Assessment of Precision Technologies for Accurate Delivery of Crop Inputs 
 
by 
 
Tyrel Lee Harbuck 
 
 
 
 
A dissertation submitted to the Graduate Faculty of 
Auburn University 
in partial fulfillment of the 
requirements for the Degree of 
Doctor of Philosophy 
 
Auburn, Alabama 
August 9, 2010 
 
 
 
 
Keywords: precision agriculture, autoguidance, GPS, subsurface-drip irrigation 
 
 
Copyright 2010 by Tyrel Lee Harbuck 
 
 
Approved by 
 
John Fulton, Co-chair, Associate Professor of Biosystems Engineering 
Donald Eakes, Co-chair, Professor of Horticulture 
Mark Dougherty, Assistant Professor of Biosystems Engineering 
Jeff Sibley, Professor of Horticulture 
Steven Taylor, Professor of Biosystems Engineering
ii 
 
Abstract 
 
 At the Tennessee Valley Research and Extension Center (TVREC), Belle Mina, 
Alabama, Global Positioning System (GPS)-assisted machine guidance and subsurface-
drip irrigation (SDI) have been implemented to study precision agriculture practices.  
Crop yield can be adversely affected if 1) not planted optimally relative to SDI product 
placement, or 2) the SDI system performance has degraded.  Recent evidence of 
guidance-system drift and soil moisture variability after irrigation at TVREC prompted 
performance evaluations of both systems.  Use of an optical total station was investigated 
to quantify positioning error in a GPS-based autoguidance system when using different 
correction-signal services.  Also developed was an assessment technique for in-situ 
pressure-compensating SDI application uniformity at variable operating pressures or 
slopes. 
  The total station provided an absolute frame of reference for comparing accuracy 
and repeatability of Wide-Area Augmentation System (WAAS), John Deere? SF1 and 
SF2, and Real-Time Kinematic (RTK) correction signals.  Recommended techniques for 
proper correction-signal testing were to perform site calibration, establish semi-
permanent reference monuments, conduct guidance tests in northward/southward or 
eastward/westward path orientations, and account for tracking-prism tilt.  Measured 
cross-track error (XTE) was influenced by drive-path orientation for RTK, SF2, and 
WAAS; SF1 performance was consistent regardless of orientation.  XTE for SF1 and 
iii 
 
WAAS were more representative of manufacturer?s claims.  Unsigned 95% short-term 
and long-term error probability was preferred to describe guidance performance over 
signed pass-to-pass (P2P) and year-to-year (Y2Y) errors, respectively. 
 A field-testing apparatus was fabricated to interface with the in-situ SDI product 
to deliver water over a range of operating pressures (48, 83, 117, and 138 kPa) while 
measuring sample discharge.  Apparatus discharge rates agreed to within a half-percent 
of per-emitter laboratory discharge rates. Lowest discharge was observed for the irrigated 
SDI product operating at 48-kPa.  Discharges from the non-irrigated and irrigated SDI 
treatments at 83 and 117 kPa were not different (?=0.05). 
 In summary, water delivery is dictated by SDI placement and functionality; as a 
consequence, other farming practices have to be conducted with a high level of spatial 
accuracy in relation to the SDI system in order to maximize crop yield, like the RTK 
correction signal. Techniques and equipment developed in this study provide 
autoguidance and SDI researchers with innovative approaches to evaluate long-term 
performance of these precision-agriculture systems. 
 
iv 
 
 
Acknowledgments 
 
 I would like to express my sincere gratitude to all of those in the Biosystems 
Engineering Department that contributed not only to my research program, but also to the 
betterment of myself from a teaching, extension, and personal point of view.  Special 
thanks are extended to Bobby Norris, David Harkins, Brad Durham, and the other station 
personnel at the Tennessee Valley Research and Extension Center for assistance with 
field work.  I would like to thank Anora Brooks, Daniel Mullenix, Corey Kichler, 
Christian Brodbeck, Seth Basinger, and Ajay Sharda for helping to facilitate this research 
effort.  And, of course, I would like to thank those who served on my committee: Dr. 
Mark Dougherty, Dr. Steven Taylor, Dr. Jeff Sibley, Dr. Joe Eakes for serving as the co-
chair of this project, and certainly Dr. John Fulton for advising me these many years.  
Lastly, I would like to thank Dr. Delos Hughes for his patronage, advice, and his 
assurances that everything ?will be fine,? my parents, and the rest of my family and 
friends for love and support that they have shown me throughout my lifetime of academic 
endeavors. 
v 
 
Table of Contents 
Abstract .......................................................................................................................... ii 
Acknowledgments .......................................................................................................... iv 
List of Tables.................................................................................................................. xi 
List of Figures .............................................................................................................. xix 
List of Abbreviations ................................................................................................... xxv 
Chapter One Introduction ................................................................................................ 1 
1.1 Preface .............................................................................................................. 1 
1.2 Justification ....................................................................................................... 3 
1.2.1 Machine Guidance ......................................................................................3 
1.2.2 SDI Application Uniformity .......................................................................5 
1.3 Objectives ......................................................................................................... 7 
1.4 Organization of Dissertation .............................................................................. 7 
Chapter Two Literature Review ....................................................................................... 9 
2.1 Overview of Static GPS ..................................................................................... 9 
2.1.1 Principles of Operation ............................................................................. 10 
2.1.2 Measures of Accuracy and Precision ........................................................ 10 
2.2 Guidance Systems ........................................................................................... 13 
2.2.1 Uses and Benefits ..................................................................................... 14 
vi 
 
2.2.2 Typical Components ................................................................................. 15 
2.3 GPS Correction ............................................................................................... 16 
2.3.1 Characteristics of GPS Correction ............................................................ 16 
2.3.2 Wide-Area Augmentation System (WAAS) .............................................. 17 
2.3.3 John Deere StarFire 1 (SF1) and StarFire 2 (SF2) ..................................... 17 
2.3.4 Real-Time Kinematic (RTK) .................................................................... 18 
2.4 Overview of Dynamic GPS ............................................................................. 19 
2.4.1 Measures of Accuracy and Precision ........................................................ 19 
2.4.2 Previous Research to Model Accuracy and Precision ................................ 22 
2.5 Overview of Total Station ................................................................................ 25 
2.5.1 Principles of Operation ............................................................................. 25 
2.5.2 Previous Research to Model Dynamic GPS Performance .......................... 26 
2.6 Subsurface-Drip Irrigation (SDI) ..................................................................... 27 
2.6.1 Uses and Benefits ..................................................................................... 27 
2.6.2 Evaluating SDI Performance .................................................................... 28 
Chapter Three Developing a Terrestrial-Based Process for Measuring Accuracy of GPS 
Correction Services in Guidance-System Applications ................................................... 31 
3.1 Abstract ........................................................................................................... 31 
3.2 Introduction ..................................................................................................... 33 
3.3 Sub-Objective .................................................................................................. 36 
vii 
 
3.4 Materials and Methods .................................................................................... 36 
3.4.1 General Test-site Preparations .................................................................. 38 
3.4.2 Test Site One Preparations ........................................................................ 39 
3.4.3 Test Site Two Preparations ....................................................................... 40 
3.4.4 Test Site One: Data Analysis .................................................................... 41 
3.4.5 Test Site Two: Data Analysis ................................................................... 43 
3.4.6 Statistical Analyses................................................................................... 46 
3.5 Results and Discussion .................................................................................... 47 
3.5.1 Site One Results ....................................................................................... 47 
3.5.2 Site Two Results ...................................................................................... 54 
3.5.3 Investigations into Normality of XTE ....................................................... 56 
3.5.4 Position Adjustments via Slope and Aspect Modeling .............................. 59 
3.5.5 Considerations for Continuing Research ................................................... 60 
3.6 Summary ......................................................................................................... 61 
Chapter Four Accuracy of GPS Correction Services Over Time for Guidance-System 
Applications .................................................................................................................. 62 
4.1 Abstract ........................................................................................................... 62 
4.2 Introduction ..................................................................................................... 63 
4.3 Sub-Objective .................................................................................................. 67 
4.4 Materials and Methods .................................................................................... 67 
viii 
 
4.4.1 Test-site Preparations ............................................................................... 69 
4.4.2 Statistical Analyses................................................................................... 72 
4.5 Results and Discussion .................................................................................... 72 
4.5.1 First-Year Evaluations of XTE ................................................................. 72 
4.5.2 Second-Year Evaluations of XTE ............................................................. 76 
4.5.3 Signed P2P Versus Unsigned Short-Term Error Evaluations..................... 82 
4.5.4 Signed Versus Unsigned Year-to-Year Error (Y2Y) Evaluations .............. 85 
4.5.5 Considerations for Continuing Research ................................................... 86 
4.6 Summary ......................................................................................................... 86 
Chapter Five Evaluating Application Uniformity of an In-Situ Pressure-Compensating 
Subsurface-Drip Irrigation Product ................................................................................ 87 
5.1 Abstract ........................................................................................................... 87 
5.2 Introduction ..................................................................................................... 89 
5.3 Sub-Objectives ................................................................................................ 92 
5.4 Materials and Methods .................................................................................... 92 
5.4.1 Laboratory Investigations ......................................................................... 92 
5.4.2 SDI Testing Apparatus ............................................................................. 93 
5.4.3 Laboratory Tests for SDI Testing Apparatus ............................................. 97 
5.4.4 Field Preparations ..................................................................................... 98 
5.4.5 Field Investigations ................................................................................ 100 
ix 
 
5.5 Results and Discussion .................................................................................. 101 
5.5.1 Laboratory Investigations ....................................................................... 101 
5.5.2 Laboratory Investigations of SDI Testing Apparatus ............................... 102 
5.5.3 Field Investigations ................................................................................ 104 
5.5.4 Considerations for Continuing Research ................................................. 108 
5.6 Summary ....................................................................................................... 109 
Chapter Six Conclusion ............................................................................................... 111 
6.1 Conclusions ................................................................................................... 111 
6.2 Opportunities for Future Research ................................................................. 113 
6.3 Final Considerations ...................................................................................... 115 
References ................................................................................................................... 117 
Appendix A    Tractor Specifications ........................................................................... 126 
Appendix B    Total-Station Specifications .................................................................. 128 
Appendix C    Targeting Prism Specifications .............................................................. 132 
Appendix D    Guidance-System Specifications ........................................................... 134 
Appendix E    Correction Signals Used in Guidance-Accuracy Tests ........................... 139 
Appendix F    Guidance Testing Site One .................................................................... 141 
Appendix G    Guidance Testing Site Two ................................................................... 143 
Appendix H    SDI Testing Apparatus Parts ................................................................. 146 
Appendix I    Test Event Shedule (2006)...................................................................... 152 
Appendix J    Test Event Schedule (2007) .................................................................... 154 
x 
 
Appendix K    Concepts and Equations Used in guidance-system performance tests .... 157 
Appendix L    Guidance assessment Office Processing Procedure (2006) ..................... 161 
Appendix M    Guidance Assessment Office Processing Procedure (2007) .................. 164 
Appendix N    Supplemental 2006 XTE Data ............................................................... 168 
Appendix O    Supplemental 2007 XTE Data ............................................................... 183 
Appendix P    SDI Product specifications and Laboratory Data .................................... 228 
Appendix Q    SDI Field Data ...................................................................................... 243 
 
  
xi 
 
List of Tables 
Table 3.1.  Typical reported accuracies for correction signals used in this study. ............ 38 
Table 3.2. Data collection events for each correction signal in 2006 (Site 1). ................. 48 
Table 3.3.  Measures of location and assumption of normality for 2006 XTE data. ........ 50 
Table 3.4.  Measures of location and assumption of normality for 2007 XTE data. ........ 56 
Table 4.1.  Typical reported accuracies for correction signals used in this study. ............ 69 
Table 4.2.  Summary of 2006 XTE data. ........................................................................ 74 
Table 4.3.  Summary of 2007 XTE data. ........................................................................ 78 
Table 4.4. Signed pass-to-pass error (P2P, cm) throughout the 2006 study. .................... 83 
Table 4.5. Signed pass-to-pass error (P2P, cm) throughout the 2007 study. .................... 84 
Table 4.6.  Year-to-year error (Y2Y, cm) from the 2006 and 2007 studies. .................... 85 
Table 5.1. Emitter discharge rates (L h-1) over a range of operating pressures on 20 
contiguous emitters subjected to a 6-percent slope rise (laboratory conditions). ........... 102 
Table 5.2. Per-emitter and apparatus discharge values over a range of operating pressures 
on 100 contiguous emitters utilizing SDI testing apparatus (laboratory conditions). ..... 103 
Table 5.3. Mean emitter-discharge-equivalent values (L h-1) on 20 contiguous emitters 
utilizing SDI testing apparatus (field conditions).  Values from sites 2 and 9 were omitted 
from the analysis. ........................................................................................................ 105 
Table E.1.  Characteristics of GPS-based correction signals used in 2006 and 2007 tests.
 .................................................................................................................................... 140 
xii 
 
Table I.1. Testing event schedule of 2006. ................................................................... 153 
Table J.1. Testing event schedule of 2007 for RTK and John Deere SF1 correction 
signals. ........................................................................................................................ 155 
Table J.2. Testing event schedule of 2007 for John Deere SF2 and WAAS correction 
signals. ........................................................................................................................ 156 
Table N.1. Summary signal-correction quality statistics for testing event 1 (February 22, 
2006); tests R11 and W11. ............................................................................................. 169 
Table N.2. Summary signal-correction quality statistics for testing event 2 (February 27, 
2006); tests R21, D21, and W21. ..................................................................................... 170 
Table N.3. Summary signal-correction quality statistics for testing event 3 (February 27, 
2006); tests R22, D22, and W22. ..................................................................................... 171 
Table N.4. Summary signal-correction quality statistics for testing event 4 (March 8, 
2006); tests R31, D31, and W31. ..................................................................................... 172 
Table N.5. Summary signal-correction quality statistics for testing event 5 (March 8, 
2006); tests R32, D32, and W32. ..................................................................................... 173 
Table N.6. Summary signal-correction quality statistics for testing event 6 (March 28, 
2006); tests R41, D41, and W41. ..................................................................................... 174 
Table N.7. Summary signal-correction quality statistics for testing event 7 (March 28, 
2006); tests R42, D42, and W42. ..................................................................................... 175 
Table N.8. Summary signal-correction quality statistics for testing event 8 (April 7, 
2006); tests R51, J51, and D51. ....................................................................................... 176 
Table N.9. Summary signal-correction quality statistics for testing event 9 (April 7, 
2006); tests R52, J52, and D52. ....................................................................................... 177 
xiii 
 
Table N.10. Summary signal-correction quality statistics for testing event 10 (June 14, 
2006); tests R61 and D61. .............................................................................................. 178 
Table N.11. Summary signal-correction quality statistics for testing event 11 (June 14, 
2006); tests R62, J62, and D62. ....................................................................................... 179 
Table N.12. Summary signal-correction quality statistics for testing event 12 (June 16, 
2006); tests R71, J71, and D71. ....................................................................................... 180 
Table N.13. Summary signal-correction quality statistics for testing event 13 (June 16, 
2006); tests R72, J72, and D72. ....................................................................................... 181 
Table N.14. Box-and-whisker plot data (cm) for the 2006 season................................. 182 
Table O.1. Summary signal-correction quality statistics for north-to-south testing event 1 
(March 7, 2007); tests RS11, JS11, DS11, and WS11. ......................................................... 184 
Table O.2. Summary signal-correction quality statistics for east-to-west testing event 1 
(March 7, 2007); tests JW11, DW11, and WW11. ............................................................... 185 
Table O.3. Summary signal-correction quality statistics for east-to-west testing event 2 
(March 7, 2007); tests JW12, DW12, and WW12. ............................................................... 186 
Table O.4. Summary signal-correction quality statistics for north-to-south testing event 2 
(March 14, 2007); tests RS21, JS21, DS21, and WS21. ....................................................... 187 
Table O.5. Summary signal-correction quality statistics for north-to-south testing event 3 
(March 14, 2007); tests RS22, JS22, DS22, and WS22. ....................................................... 188 
Table O.6. Summary signal-correction quality statistics for east-to-west testing event 3 
(March 14, 2007); tests RW21, JW21, DW21, and WW21. ................................................... 189 
Table O.7. Summary signal-correction quality statistics for east-to-west testing event 4 
(March 14, 2007); tests RW22, JW22, DW22, and WW22. ................................................... 190 
xiv 
 
Table O.8. Summary signal-correction quality statistics for north-to-south testing event 4 
(April 13, 2007); tests RS31, JS31, DS31, and WS31. ......................................................... 191 
Table O.9. Summary signal-correction quality statistics for north-to-south testing event 5 
(April 13, 2007); tests RS32, JS32, DS32, and WS32. ......................................................... 192 
Table O.10. Summary signal-correction quality statistics for east-to-west testing event 5 
(April 13, 2007); tests RW31, JW31, DW31, and WW31. ..................................................... 193 
Table O.11. Summary signal-correction quality statistics for east-to-west testing event 6 
(April 13, 2007); tests RW32, JW32, DW32, and WW32. ..................................................... 194 
Table O.12. Summary signal-correction quality statistics for north-to-south testing event 
6 (April 30, 2007); tests RS41, JS41, DS41, and WS41. ...................................................... 195 
Table O.13. Summary signal-correction quality statistics for east-to-west testing event 7 
(April 30, 2007); test DW41. .......................................................................................... 196 
Table O.14. Summary signal-correction quality statistics for north-to-south testing event 
7 (June 5, 2007); tests RS51, JS51, DS51, and WS51. ......................................................... 197 
Table O.15. Summary signal-correction quality statistics for north-to-south testing event 
8 (June 5, 2007); tests RS52, JS52, DS52, and WS52. ......................................................... 198 
Table O.16. Summary signal-correction quality statistics for east-to-west testing event 8 
(June 5, 2007); tests RW51, JW51, DW51, and WW51. ........................................................ 199 
Table O.17. Summary signal-correction quality statistics for east-to-west testing event 9 
(June 5, 2007); tests JW52, DW52, and WW52. .................................................................. 200 
Table O.18. Summary signal-correction quality statistics for north-to-south testing event 
9 (July 10, 2007); tests RS61, JS61, DS61, and WS61. ........................................................ 201 
xv 
 
Table O.19. Summary signal-correction quality statistics for north-to-south testing event 
10 (July 10, 2007); tests JS62, DS62, and WS62. ............................................................... 202 
Table O.20. Summary signal-correction quality statistics for east-to-west testing event 10 
(July 10, 2007); tests RW61, JW61, DW61, and WW61. ....................................................... 203 
Table O.21. Summary signal-correction quality statistics for east-to-west testing event 11 
(July 10, 2007); tests JW62, DW62, and WW62.................................................................. 204 
Table O.22. Summary signal-correction quality statistics for north-to-south testing event 
11 (July 18, 2007); tests RS71, JS71, DS71, and WS71. ...................................................... 205 
Table O.23. Summary signal-correction quality statistics for north-to-south testing event 
12 (July 18, 2007); tests JS72, DS72, and WS72. ............................................................... 206 
Table O.24. Summary signal-correction quality statistics for north-to-south testing event 
13 (July 18, 2007); tests JS73, DS73, and WS73. ............................................................... 207 
Table O.25. Summary signal-correction quality statistics for east-to-west testing event 12 
(July 18, 2007); tests RW71, JW71, DW71, and WW71. ....................................................... 208 
Table O.26. Summary signal-correction quality statistics for east-to-west testing event 13 
(July 18, 2007); tests JW72, DW72, and WW72.................................................................. 209 
Table O.27. Summary signal-correction quality statistics for east-to-west testing event 14 
(July 18, 2007); tests JW73, DW73, and WW73.................................................................. 210 
Table O.28. Summary signal-correction quality statistics for north-to-south testing event 
14 (July 31, 2007); tests RS81, JS81, DS81, and WS81. ...................................................... 211 
Table O.29. Summary signal-correction quality statistics for north-to-south testing event 
15 (July 31, 2007); tests RS82, JS82, DS82, and WS82. ...................................................... 212 
xvi 
 
Table O.30. Summary signal-correction quality statistics for north-to-south testing event 
16 (July 31, 2007); tests RS83, JS83, DS83, and WS83. ...................................................... 213 
Table O.31. Summary signal-correction quality statistics for east-to-west testing event 15 
(July 31, 2007); tests RW81, JW81, DW81, and WW81. ....................................................... 214 
Table O.32. Summary signal-correction quality statistics for east-to-west testing event 16 
(July 31, 2007); tests RW82, JW82, DW82, and WW82. ....................................................... 215 
Table O.33. Summary signal-correction quality statistics for north-to-south testing event 
17 (August 3, 2007); tests RS91, JS91, DS91, and WS91. ................................................... 216 
Table O.34. Summary signal-correction quality statistics for east-to-west testing event 17 
(August 3, 2007); tests RW91, JW91, DW91, and WW91. .................................................... 217 
Table O.35. Summary signal-correction quality statistics for north-to-south testing event 
18 (August 23, 2007); tests RS101, JS101, DS101, and WS101. ............................................ 218 
Table O.36. Summary signal-correction quality statistics for north-to-south testing event 
19 (August 23, 2007); tests RS102, JS102, DS102, and WS102. ............................................ 219 
Table O.37. Summary signal-correction quality statistics for north-to-south testing event 
20 (August 23, 2007); tests RS103, JS103, DS103, and WS103. ............................................ 220 
Table O.37. Summary signal-correction quality statistics for north-to-south testing event 
20 (August 23, 2007); tests RS103, JS103, DS103, and WS103. ............................................ 220 
Table O.38. Summary signal-correction quality statistics for east-to-west testing event 18 
(August 3, 2007); tests RW101, JW101, DW101, and WW101. ............................................... 221 
Table O.39. Summary signal-correction quality statistics for east-to-west testing event 19 
(August 3, 2007); tests RW102, JW102, DW102, and WW102. ............................................... 222 
xvii 
 
Table O.40. Summary signal-correction quality statistics for east-to-west testing event 20 
(August 3, 2007); tests RW103, JW103, DW103, and WW103. ............................................... 223 
Table O.41.  Box-and-whisker plot data for the 2007 season (RTK correction). ........... 224 
Table O.42.  Box-and-whisker plot data for the 2007 season (SF1 correction).............. 225 
Table O.43.  Box-and-whisker plot data for the 2007 season (SF2 correction).............. 226 
Table O.44.  Box-and-whisker plot data for the 2007 season (WAAS correction). ....... 227 
Table P.2. Five-minute emitter discharge (mL) from horizontal dripline, 83 kPa test. .. 230 
Table P.3. Five-minute emitter discharge (mL) from dripline on two slopes, 117 kPa test.
 .................................................................................................................................... 231 
Table P.4-1. Five-minute emitter discharge (mL) across 48-kPa and 83-kPa operating 
pressures, six-percent slope rise. .................................................................................. 233 
Table P.4-2. Five-minute emitter discharge (mL) across 117-kPa and 138-kPa operating 
pressures, six-percent slope rise. .................................................................................. 234 
Table P.5-1. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 48-kPa 
operating pressure, replication #1. ............................................................................... 236 
Table P.5-2. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 48-kpa  
operating pressure, replication #2. ............................................................................... 236 
Table P.5-3. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 48-kpa 
operating pressure, replication #3. ............................................................................... 237 
Table P.5-4. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 83-kPa 
operating pressure, replication #1. ............................................................................... 237 
Table P.5-5. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 83-kPa 
operating pressure, replication #2. ............................................................................... 238 
xviii 
 
Table P.5-6. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 83-kPa 
operating pressure, replication #3. ............................................................................... 238 
Table P.5-7. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 117-kPa 
operating pressure, replication #1. ............................................................................... 239 
Table P.5-8. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 117-kPa 
operating pressure, replication #2. ............................................................................... 239 
Table P.5-9. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 117-kPa 
operating pressure, replication #3. ............................................................................... 240 
Table P.5-10. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 138-
kPa operating pressure, replication #1. ......................................................................... 240 
Table P.5-11. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 138-
kPa operating pressure, replication #2. ......................................................................... 241 
Table P.5-12. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 138-
kPa operating pressure, replication #3. ......................................................................... 241 
Table P.5-13.Emitter exponent calculations as referenced in ASAE S553 (2008). ........ 242 
Table Q.3-1.  SDI-apparatus sample-discharge values for Sites 1-4. ............................. 249 
Table Q.3-2.  SDI-apparatus sample-discharge values for Sites 5-8. ............................. 250 
Table Q.3-3.  SDI-apparatus sample-discharge values for Sites 9-12. ........................... 251 
Table Q.3-4.  SDI-apparatus sample-discharge values for Sites 13-16. ......................... 252 
Table Q.3-5.  SDI-apparatus sample-discharge values for Sites 17 and 18. ................... 253 
  
xix 
 
List of Figures 
Figure 2.1. Example guidance-system setup.  Typical guidance system components 
include a GPS antenna/receiver (a), a computer module (b), and an input/output device 
(c). ................................................................................................................................. 15 
Figure 3.1.  Depiction of cross-track error (XTE) determination. ................................... 43 
Figure 3.2.  Example dataset produced from total-station observations.  The labels for the 
gridded lines represent local grid coordinates, projected into the Universal-Transverse 
Mercator (UTM) coordinate system (units are in meters).  The triangular markers 
represent stable reference monuments used in site calibration.  The green lines represent 
the vectors from the total station monument to the observed prism positions (illustrated as 
white dots). .................................................................................................................... 45 
Figure 3.3.  Example test dataset in GIS format.  Black dots within each line refer to 
position data recorded by the total station; lines drawn between dots represent travel 
?paths? made by the test vehicle.  Paths are separated by color according to correction 
signal used for each traverse in relation to the relevant reference guidance line. ............. 46 
Figure 3.4.  Portions of drive paths under WAAS correction for 2006 (site 1).  The 
negative relationship between XTE and Easting values suggests roll/pitch error and/or 
insufficient XTE calculations. ........................................................................................ 52 
Figure 3.5.  Graph of XTE observations per row for SF2 testing event on June 16, 2006.  
Observations from Row 1 were flagged as potential outliers and omitted from analyses.58 
xx 
 
Figure 4.1. Guidance testing site one, used in 2006.  Elevation contours are in 2-ft (0.6-
m) intervals; guidance lines are spaced 20 ft (6.1 m).  Direction of vehicle travel was 
westerly to easterly (approx. 80.1 degrees azimuth from true north; aerial imagery and 
contour data courtesy of City of Auburn GIS Department, Auburn, AL). ....................... 70 
Figure 4.2.  Guidance testing site two, used in 2007.  Elevation contours are in 2-ft (0.6-
m) intervals; guidance lines are spaced 40 ft (12.2 m).  Direction of vehicle travel was 
northerly to southerly (approx. 179.4 degrees azimuth from true north; aerial imagery and 
contour data courtesy of City of Auburn GIS Department, Auburn, AL). ....................... 70 
Figure 4.3. Guidance testing site two, used in 2007.  Elevation contours are in 2-ft (0.6-
m) intervals; guidance lines are spaced 40 ft (12.2 m).  Direction of vehicle travel was 
easterly to westerly (approx. 271.3 degrees azimuth from true north; aerial imagery and 
contour data courtesy of City of Auburn GIS Department, Auburn, AL). ....................... 71 
Figure 4.4.  Cumulative distributions of unsigned short-term and long-term error (cm) for 
(a) RTK, (b) SF1, (c) SF2, and (d) WAAS corrections in the 2006 study. ...................... 73 
Figure 4.5.  Box-and-whisker plots for individual testing events and Y2Y (2006) for (a) 
RTK, (b) SF1, (c) SF2, and (d) WAAS correction signals.  Testing Event Label 
corresponds to testing events listed in Appendix I. ......................................................... 76 
Figure 4.6.  Cumulative distributions of unsigned short-term and long-term error (cm) for 
(a) RTK, (b) SF1, (c) SF2, and (d) WAAS corrections in the 2007 study. ...................... 77 
Figure 4.7.  XTE box-and-whisker plots for RTK correction signal in 2007, comparing 
southward orientation (left) and westward orientation (right).  Testing Event Label 
corresponds to testing events listed in Appendix J. ......................................................... 80 
xxi 
 
Figure 4.8.  XTE box-and-whisker plots for SF1 correction signal in 2007, comparing 
southward orientation (left) and westward orientation (right).  Testing Event Label 
corresponds to testing events listed in Appendix J. ......................................................... 80 
Figure 4.9.  XTE box-and-whisker plots for SF2 correction signal in 2007, comparing 
southward orientation (left) and westward orientation (right).  Testing Event Label 
corresponds to testing events listed in Appendix J. ......................................................... 81 
Figure 4.10.  XTE box-and-whisker plots for WAAS correction signal in 2007, 
comparing southward orientation (left) and westward orientation (right).  Testing Event 
Label corresponds to testing events listed in Appendix J. ............................................... 81 
Figure 5.1. Illustration of the portable, in-field SDI testing apparatus. The white vessel 
atop the bench scale served as the water supply reservoir during testing.  The bench scale 
provided vessel mass, which equated to sample discharge.  The 12-volt battery-powered 
pump was mounted on the side and provided the operating pressure for the SDI sample 
tested. ............................................................................................................................ 94 
Figure 5.2. Pressure and flow regulation system for the SDI testing apparatus.  Most of 
the water from the pump (left side) was recirculated to the white reservoir (not shown) 
and remaining water was distributed to the in-situ SDI sample section through an 
adjustable pressure regulator (center right) and the green hose (right side).  The gate 
valve (center left) and the adjustable pressure regulator (center right) served as a means to 
target and sustain the necessary operating pressure. ....................................................... 95 
Figure 5.3. Pressure sensors located at the (a) upstream and (b) downstream ends of the 
drip-tape being sampled. ................................................................................................ 97 
xxii 
 
Figure 5.4. Depiction of test plot with respect to slope categories, represented as percent 
rise.  Sampling locations represent 12.2-m drip-line segments blocked according to 
?irrigated? versus ?non-irrigated? treatment effect and slope category. .......................... 99 
Figure 5.5.  Pressure-versus-equivalent-discharge curves for three SDI samples.  Per-
emitter discharge witnessed from product evaluation in a laboratory setting (?Baseline?) 
is represented as 1.036 L h-1 ?5% margin..................................................................... 106 
Figure 5.6. Illustration of drip-tape testing sites corresponding to cotton lint yield.  Yield 
variability is suggested in the vicinity of the SDI sample labeled as site ?9?; average 
estimated emitter discharge at site 9 was 0.54 L h-1, which is 48% less discharge than that 
stated by the manufacturer. .......................................................................................... 108 
Figure A.1. John Deere 6420. ...................................................................................... 127 
Figure B.1. Trimble 5603 Total Station. ....................................................................... 129 
Figure B.2. Trimble TSCe datalogger. ......................................................................... 130 
Figure B.3. Trimble Survey Controller. ........................................................................ 131 
Figure C.1. Trimble RMT prism. ................................................................................. 133 
Figure D.1. John Deere Greenstar display. ................................................................... 135 
Figure D.2. StarFire iTC GPS receiver (yellow dome). ................................................ 136 
Figure D.3. Trimble EZ-Guide display and accessory GPS antenna. ............................ 137 
Figure D.4. Trimble EZ-Steer assisted-steering system. ............................................... 138 
Figure F.1. Guidance testing site one, used in 2006.  Elevation contours are in 2-ft (0.6-
m) intervals; guidance lines are spaced 20 ft (6.1 m).  Direction of vehicle travel was 
westerly to easterly (approx. 80.1 degrees azimuth from true north; aerial imagery and 
contour data courtesy of City of Auburn GIS Department, Auburn, AL). ..................... 142 
xxiii 
 
Figure G.1. Guidance testing site two, used in 2007.  Elevation contours are in 2-ft (0.6-
m) intervals; guidance lines are spaced 40 ft (12.2 m).  Direction of vehicle travel was 
northerly to southerly (approx. 179.4 degrees azimuth from true north; aerial imagery and 
contour data courtesy of City of Auburn GIS Department, Auburn, AL). ..................... 144 
Figure G.2. Guidance testing site two, used in 2007.  Elevation contours are in 2-ft (0.6-
m) intervals; guidance lines are spaced 40 ft (12.2 m).  Direction of vehicle travel was 
easterly to westerly (approx. 271.3 degrees azimuth from true north; aerial imagery and 
contour data courtesy of City of Auburn GIS Department, Auburn, AL). ..................... 145 
Figure H.1. Shurflo diaphragm pump. .......................................................................... 147 
Figure H.2. Watts adjustable small-pressure regulator. ................................................. 148 
Figure H.3. Ohaus CD-11 indicator (scale platform not shown). .................................. 148 
Figure H.4. PCB pressure transducer. .......................................................................... 149 
Figure H.5. Campbell Scientific CR206 datalogger. ..................................................... 150 
Figure H.6. Campbell Scientific RF401 radio modem. ................................................. 150 
Figure H.7. X20 field computer screen shot illustrating pressure-monitoring software. 151 
Figure K.1. Depiction of cross-track error. ................................................................... 158 
Figure P.1-1. Dripline emitter data provided by manufacturer. ..................................... 229 
Figure P.1-2. Dripline product performance provided by manufacturer. ....................... 229 
Figure Q.1.  Aerial imagery of Tennessee Valley Research and Extension Center 
(TVREC), Belle Mina, Alabama (2006 data; imagery courtesy of Aerial Photography 
Field Office, United States Department of Agriculture). .............................................. 244 
Figure Q.2-1.  Irrigation treatment assignment for Field 52. ......................................... 246 
xxiv 
 
Figure Q.2-2.  Sampling locations for Field 52, based on irrigation treatment (system use 
frequency) and surface slope attributes. ....................................................................... 247 
 
xxv 
 
List of Abbreviations 
CEP Circular Error Probable 
CV Coefficient of Variation 
DEM Digital Elevation Model 
DGPS Differential GPS 
DOP Dilution of Precision 
EDM Electronic Distance Measurement 
GNSS Global Navigation Satellite System 
GPS Global Positioning System 
ID Interior Diameter 
IQR Inter-Quartile Range 
NMEA National Marine Electronic Association 
P2P Pass-to-Pass error/accuracy 
PND Personal Navigation Device 
RMS/RMSE Root Mean Square Error 
RTK Real-Time Kinematic 
SD Standard Deviation 
SDI Subsurface Drip Irrigation 
SEP Spherical Error Probable 
TCM Terrain Compensation Module 
TPS Terrestrial Positioning System 
TVREC Tennessee Valley Research and Extension Center 
U.S. DoD United States Department of Defense 
UTM Universal Transverse Mercator 
WAAS Wide-Area Augmentation System 
XTE Cross-Track Error 
Y2Y Year-to-Year error/accuracy 
 
 
1 
 
CHAPTER ONE 
INTRODUCTION 
1.1 PREFACE 
 The concept of Precision Agriculture can best be described by an almost literal 
interpretation of the term---applying practices and technologies to agriculture to better 
capture and understand the precise interactions between the crop and the land.  Precision 
Agriculture enhances current farming techniques not only by improving crop yields, but 
also to endorse a better stewardship with the environment at large by demonstrating how 
tillage, fertilizers, pesticides, and even water  usage can be minimized on a commercial 
scale. As a consequence, farmers can experiment with new and innovative ways of 
growing, maintaining, and harvesting their crops in order to maximize production and 
profits. 
 One of the obvious relationships between the crop and the land is their physical 
proximity to one another.  But, the physical position of the farm equipment in relation to 
the crop and to the land is just as important.  For instance, vehicle and implement control 
is necessary so as not to destroy the crop by inadvertent damage from moving equipment.  
Likewise, if a farmer knows that a certain area within a field tends to produce less than 
other areas, the farmer can either avoid the area altogether or develop a site-specific 
management strategy to address this inherent variability.  With these applications in 
2 
 
mind, knowing the spatial relationships between the crop, the land, and the equipment are 
crucial for implementing most precision-agriculture practices. 
 In the state of Alabama, an increasing number of farmers and producers are 
adapting their farming practices to advance the application of Precision Agriculture.  At 
the Tennessee Valley Research and Extension Center (TVREC, Belle Mina, Alabama), 
most of the agronomic crops research relates in some way to Precision Agriculture---
either directly by testing different crop management strategies, or indirectly by utilizing 
precision-agriculture technologies and strategies in daily operations.  While several plots 
exist at this facility, one particular plot of approximately 6 ha has been used in an 
experimental capacity since fall of 2003; the research objectives for this extended-year 
study involve evaluating cotton (Gossypium hirsutum L.) lint yields in response to 
accumulated biomass from cover-crops and supplemental irrigation optimized for crop 
yield. For this experiment, two particular precision-agriculture strategies are employed 
throughout the study-GPS/GNSS-based machine guidance, to aid in the precise 
placement of the cotton crop, and subsurface-drip irrigation (SDI), to provide water 
within the rooting system of the crop.  When SDI and/or controlled traffic practices are 
coupled with a high-accuracy guidance system, farming practices exhibit higher profits, 
according to Gan-Mor and Clark (2001); therefore, a reliable relationship between 
machine guidance and SDI is desired.  
 In order for the long-term experiment to remain valid from year to year, a new 
season?s cotton crop has to be planted in a precise spatial relationship with the irrigation 
product underground; also, the irrigation has to perform as anticipated to minimize yield 
variability within plots but also to maximize overall yield.  Anecdotal reports from the 
3 
 
TVREC staff have suggested that the seed placement for each season?s cotton crop in 
relation to the estimated position of the SDI tape has not been consistent from one year to 
the next.  According to the machine operators, the GPS-assisted guidance software on-
board the field equipment performed as anticipated, in that no conspicuous errors were 
evident at the time of seed placement. No discernible bias of the software or the 
equipment could be determined through routine equipment maintenance checks; also, this 
effect tended to be better or worse depending on the GPS correction signal being used to 
assist with navigation during field procedures. 
 Cotton lint yield data exhibited variability across the study site from previous 
seasons and especially within individual plots that received supplemental watering via 
SDI.  Typically, the TVREC staff monitors seasonal water usage for irrigated plots via 
flow meters installed at the head of the field.  However, yield variability did not appear to 
be evenly dispersed throughout the plots.  Flow-meter readings could offer at best an 
average measure of irrigation performance for the entire experiment plot, instead of 
focusing on the specific areas within the field that exhibited the variability.    
1.2 JUSTIFICATION 
1.2.1 MACHINE GUIDANCE 
 In order to better describe the study topic of guidance system accuracy, some 
background information regarding satellite navigation is required.  The Global 
Positioning System (GPS) is an array of about 24 navigation satellites, maintained by the 
U.S. Department of Defense (DoD), and placed in precise orbits at an altitude of 
approximately 20,000 km.  Unlike geostationary satellites, these navigation satellites are 
4 
 
revolving around the planet in six orbital planes (relative to the Equator), four satellites 
per orbital plane.  Despite the fact that these satellites are in constant motion, their 
positions relative to one another are essentially ?fixed? by careful monitoring from the 
U.S. DoD.  Essentially, each GPS satellite serves as a radio ?beacon? upon which anyone 
on Earth with a compatible GPS receiver can ?receive the GPS? information from 
satellites in view.  The software in the receiver can compute the approximate distance to 
each satellite in view, and by a geometric process of trilateration (similar to triangulation, 
except distances are used in the solution, not angles), the receiver can determine its 
position on Earth if at least four satellites are visible. 
 This raw, or autonomous, position solution could be several meters from the true 
geographical location.  Several conditions relating to the quality of the GPS receiver, the 
amount of signal interference, and even local environmental conditions may accumulate 
this error.  However, several of these conditions can be managed to a large extent 
allowing the error to be mitigated.  Several manufacturers offer ?correction signals? that 
serve to enhance the real-time positioning of GPS/GNSS receivers; autoguidance systems 
almost exclusively require some form of correction signal to operate at a performance 
level acceptable to most users. 
 Several GPS correction signals are available, and these signals typically can be 
ranked according to their accuracies.  Consequently, the higher-accuracy correction 
signals require more of a financial investment for the farmer in order to optimize farm 
equipment and/or field operations.  Manufacturers and commercial providers of 
correction-signal services advertise these accuracies in order to attract customers; 
however, as there are currently no established procedures or guidelines in the agricultural 
5 
 
or engineering community to evaluate the dynamic accuracy of GPS-assisted guidance 
systems, it is believed that the information provided by manufacturers may not be 
presented in clear context relative to their competitors. 
 As shown by the positioning-error phenomenon witnessed at TVREC, variability 
in GPS correction-signal accuracy may not be readily apparent if the vehicle operator 
relies solely upon the solution generated by the guidance-system software.  However, the 
vehicle operator may notice during a guidance maneuver that the vehicle is not tracking 
on a repeatable drive path or that proper seed placement relative to an SDI system did not 
occur. In terms of guidance-system testing, utilizing an absolute or terrestrial frame of 
reference would be equivalent to having an operator witness any positioning 
discrepancies independent of the guidance system.  Furthermore, if positioning error is 
being propagated through the GPS solutions themselves, then a terrestrial or non-GPS-
based testing procedure may provide a clear insight into investigating dynamic guidance-
system performance.  Developing a procedure that relies upon an absolute frame of 
reference would also allow for guidance-system performance testing over extended 
periods of time, in order to quantify system accuracies from a long-term point of view. 
1.2.2 SDI APPLICATION UNIFORMITY 
 While a great deal of research has been conducted regarding the benefits of SDI, 
literature is limited in evaluating the longevity or the application uniformity of a typical 
SDI system under field conditions.  Engineering standards and procedures exist that 
allow academic/industry researchers to evaluate and quantify SDI performance statistics, 
but the current standard (ASAE S553, 2008) is more suited to a laboratory situation.  It is 
6 
 
believed that a field procedure can be developed to quantify SDI product performance in 
much the same way as the current standards allow in a laboratory setting. 
 If an SDI system is malfunctioning at the field level, then a producer would likely 
be able to identify such a problem easily through observation by an overall decline in 
crop growth throughout the season or by witnessing a catastrophic failure related to water 
flow in the system.  However, not all SDI malfunctions are easy to observe; a crimped 
lateral line, clogged emitters, or miniscule holes in the lateral line caused by abrasion 
from sub-soil rocks or rodents may go unnoticed for several seasons.  It is believed that a 
field-based testing methodology should be able to diagnose SDI product samples in a 
more site-specific manner as opposed to a field level. 
 Pressure-compensating SDI provides a consistent and uniform water application 
method in spite of dynamic pressure changes created by the SDI pumping system or by 
static pressure changes created by elevation differences between emitters.  Pressure-
compensating SDI is a relatively new product in the marketplace, so little research has 
been reported on testing the reliability of this product over an extended period of time.  It 
is believed that a field-based testing methodology should provide for evaluation of an 
SDI product with respect to typical operating-system pressures and static pressure forces 
to determine if the pressure-compensating features of such an SDI product are maintained 
over time. 
 
 
7 
 
1.3 OBJECTIVES 
 The overall goal of this research is to evaluate guidance-system accuracy and in-
situ SDI application uniformity, as both of these precision-agriculture technologies are 
related to one another.  The objectives of this research are to: 
1. Evaluate techniques for using an optical total station to provide terrestrial-based, 
dynamic accuracy testing of an autoguidance system when using different GPS-
based correction services. 
2. Compare short-term and long-term accuracies of an autoguidance system when 
using different GPS-based correction services. 
3. Develop and validate in-situ testing techniques to assess application uniformity of 
SDI products. 
4. Evaluate if variable operating pressure or terrain slope impacted application 
uniformity of a pressure-compensating SDI product. 
1.4 ORGANIZATION OF DISSERTATION 
 This dissertation is presented in manuscript format.  Chapter 1 provides 
introductory statements justifying the emphasis that was put toward this research 
followed by the main objectives.  Chapter 2 is an extensive literature review supplying 
information on the characteristics of guidance-system and SDI operation.  Each of the 
Chapters 3 through 5 represents an individual manuscript that focuses on different 
portions of this research.  Chapter 3 documents the formulation of guidance-system 
testing techniques utilizing an optical total station.  Chapter 4 characterizes long-term 
absolute accuracies and short-term relative accuracies of four GPS correction signals 
8 
 
commonly utilized by guidance-system operators.  Chapter 5 documents the formulation 
of in-field SDI testing techniques as well as SDI uniformity results conducted at the 
TVREC.  Chapter 6 summarizes this research project, presents overall conclusions, and 
includes future research suggestions.  At the end of the dissertation, a Reference section 
and Appendices were developed, documenting hardware and software specifications for 
all equipment utilized throughout the project as well as supplemental guidance and SDI-
application data. 
9 
 
CHAPTER TWO 
LITERATURE REVIEW 
 
 
 With the rise in popularity of Precision Agriculture, guidance system innovations 
and irrigation innovations have developed quickly over the last decade.  Several 
publications from industry and academia were reviewed to gain knowledge on the 
operating principles behind static GPS testing, guidance systems, GPS correction signals, 
dynamic GPS testing, and total stations.  Other manuscripts pertaining to overview of 
SDI and evaluating SDI performance were reviewed. 
2.1 OVERVIEW OF STATIC GPS 
  The infrastructure behind GPS operation is based on three components: the space 
segment, the control segment, and the user segment.  The space segment consists of the 
navigational satellite array, or constellation, located in precisely-known and closely-
monitored orbits surrounding Earth.  The control segment consists of monitoring stations 
located around the world that track the ?health? of the satellites, compare their actual 
positions with their predicted positions, and are able to change the status of the 
constellation when necessary.  Both the space segment and the control segment are 
maintained by different organizations in the U.S. Government, so civilian intervention or 
manipulation is improbable.  The user segment refers to the general public that can utilize 
10 
 
the positioning or timing capabilities from the satellites, provided they have a compatible 
receiver and software program that can process the satellite signals (Hurn, 1989). 
2.1.1 PRINCIPLES OF OPERATION 
 In simple terms, the GPS satellites broadcast continuous independent radio signals 
to Earth, containing specialized data messages describing each satellite?s identity, 
operational health, predicted position, and a timestamp.  Based on the physical equation 
D=RT, where D=distance traveled, R=rate of travel, and T=time of travel, a GPS receiver 
can calculate how far away multiple satellites are based on the amount of time needed to 
receive each satellite?s data message.  This is possible because the satellites are in precise 
orbits (approximately 20,000 km from Earth) and because radio waves in the vacuum of 
space travel at the speed of light (approximately 300,000 km s-1).  Due to the assumption 
that the GPS satellites are in fixed orbits, with Earth at the center of those orbits, a 
geometric coordinate system can be derived to locate any point within the interior sphere 
of the constellation; this geometric coordinate system can then be converted into a 
geodetic coordinate system, consisting of latitude, longitude, and height.  A GPS receiver 
must process the signals from a minimum of four GPS satellites before the receiver can 
accurately derive its location within the coordinate system, but the architecture of the 
satellite constellation is designed such that at least six satellites should be available to the 
user at any time of day throughout the world (except for the polar regions) (Hurn, 1989). 
2.1.2 MEASURES OF ACCURACY AND PRECISION 
 Several factors can influence positioning accuracy: satellite clock errors, satellite 
orbital errors, errors in the GPS receiver, signal distortion caused by the atmosphere, and 
11 
 
signal multipath (Upadhyaya et al., 2003).  Errors related to the navigational satellites can 
be predicted to a high order through careful monitoring by the U.S. DoD.  Receiver errors 
can often be related to the quality of the hardware components present in the receiver, 
such as the on-board clock and the radio-frequency microchip that receives the signals 
from space.  Atmospheric errors are created by the hindrance of the satellite radio signals 
through different levels of the atmosphere; in addition, satellites that are located closer to 
the horizon tend to output radio signals that are more susceptible to atmospheric 
distortion due to the increased signal travel distance (Srivastava et al., 1993) .  
Positioning errors are often the greatest for this category, but atmospheric modeling can 
greatly reduce the error.  Multipath is a term that refers to localized radio-signal 
disturbance, often affected by nearby reflective surfaces, buildings, or vegetation; these 
errors are site-specific and can be mitigated through careful placement of the GPS 
receiver or antenna. 
 The term accuracy refers to how well the GPS receiver calculates its position 
relative to an accepted ?true? position (Buick, 2002; Upadhyaya et al., 2003).  Static 
accuracy, or the accuracy of a receiver that has not been moved during testing, is often 
quantified in statistical terms of circular error probable (CEP), spherical error probable 
(SEP), and root mean square error (RMS error, RMSE).  CEP is described as the radius 
of a circle in which 50 percent of the GPS receiver?s calculated two-dimensional 
coordinates are contained within some extended window of time.  SEP is conceptually 
the same as CEP, except that three-dimensional coordinates are evaluated.  RMS error 
can be used to evaluate accuracy with the following equation: 
 
12 
 
g3496uni2211 g2186g2191
g2779g2196
g2191g2880g2778
g2196 uni002Cuni0020uni0020 
g1875g1860g1857g1870g1857uni0020g1856g3036g2870 g3404 g4666g1850g3036 g3398 g1850g3042g4667g2870 g3397uni0020g4666g1851g3036 g3398 g1851g3042g4667g2870uni002Cuni0020 
g4666g1850g3036uni002Cg1851g3036g4667g1870g1857g1858g1857g1870g1871uni0020g1872g1867uni0020g1871g1853g1865g1868g1864g1857uni0020g1855g1867g1867g1870g1856g1861g1866g1853g1872g1857g1871uni0020g1853g1866g1856uni0020 
g4666g1850g3042uni002Cg1851g3042g4667g1870g1857g1858g1857g1870g1871uni0020g1872g1867uni0020g1853g1855g1855g1857g1868g1872g1857g1856uni0020g1872g1870g1873g1857uni0020g1855g1867g1867g1870g1856g1861g1866g1853g1872g1857g1871 
The term precision refers to how closely the receiver repeatedly calculates its current 
position relative to the previous calculated position.  RMS error can also be used to 
evaluate precision with the following equation: 
g3496uni2211 g2187g2191
g2779g2196
g2191g2880g2778
g2196 uni002Cuni0020 
g1875g1860g1857g1870g1857uni0020g1857g3036g2870 g3404 g4666g1850g3036 g3398 g1850g3364g4667g2870 g3397uni0020g4666g1851g3036 g3398 g1851g3364g4667g2870uni002C 
g1850g3364 g3404 g3533g4666g1850g3036g4667g1853g1866g1856uni0020
g3041
g3036g2880g2869
g1851g3364 g3404 g3533g4666g1851g3036g4667
g3041
g3036g2880g2869
 
(Upadhyaya et al., 2003). 
 An additional source of positioning error comes from the satellite geometry 
available at the time of data acquisition.  The term that is used to quantify the geometry 
of the overhead satellites is Dilution of Precision (DOP). While a minimum of four 
satellites are typically required for a receiver to determine its position, having those 
satellites clustered in one part of the sky is functionally not as robust as having the same 
satellites distributed evenly overhead.  A DOP value is reported as a unitless number that 
inversely relates relative to the distribution of the satellites---more spatial dispersion of 
overhead satellites means a smaller DOP, while a more clustered satellite picture means a 
larger DOP.  Because a DOP value is a function of the satellite geometry, this value can 
13 
 
also be used indirectly to determine potential accuracy and precision of positioning data 
(Stombaugh et al., 2005). 
2.2 GUIDANCE SYSTEMS 
 Machine guidance refers to the utilization of extra-sensory devices relaying 
information to a central computer controller to aid in the navigation of farming 
equipment.  Such controllers (and the peripheral devices that accompany them) are 
generally referred to as ?guidance systems?; while some guidance systems provide visual 
or audible cues to instruct the vehicle operator where to go, more advanced systems have 
the capabilities of automatically steering the vehicle as it travels over terrain.  Such 
systems are often referred to as ?autoguidance,? in order to distinguish from ?manual 
guidance? systems requiring an operator to steer the vehicle. 
 Machine guidance requires some form of navigational aid in order to direct the 
vehicle over terrain. As mentioned previously, the most commonly-available navigational 
aid for machine guidance is satellite navigation. A GPS receiver can determine a position 
relative to the orbiting satellites and derive coordinates for its physical location as they 
correspond to a standard recognized navigational format, such as latitude and longitude.  
Machine-guidance systems equipped with GPS receivers utilize the positioning 
information in order to associate the vehicle?s current position and velocity with 
navigational cues, such as a digital map or an assigned route; for farming applications, 
the assigned route may simply correspond to the orientation and layout of the crop rows 
in the field. 
 
14 
 
2.2.1 USES AND BENEFITS 
 Producers and vehicle operators spend many hours behind steering wheels 
performing routine operations, such as preparing seed beds, applying fertilizers and 
pesticides, or planting and harvesting crops.  The repetitive nature of these activities over 
the course of the work day will often contribute to operator fatigue, leading to row skips 
or overlap.  Yields can be impacted by this, but now farmers have the ability to drive crop 
rows with greater accuracy over row counting or using foam markers to indicate prior 
passes.  Also, with an increased emphasis on stewardship of the farmland through more 
efficient farming, a farmer can collect site-specific field information that allows for less 
wasteful (and more profitable) farming operations that serve to maximize a crop yield. 
 One important benefit of incorporating a guidance system into agricultural 
machinery is that new and innovative production practices can be developed.  For 
instance, researchers have utilized guidance systems and other navigational 
enhancements to develop a rice transplanter (Nagasaka et al., 2002), an agricultural ?field 
robot? (Noguchi et al., 2002) that performed tillage and planting tasks better than skilled 
labor, and a self-propelled sprayer (Giles and Downey, 2003).  Also, because many 
guidance systems are designed to ?stream? the real-time GPS positioning data to other 
software applications, everyday farming operations are being re-invented through GPS 
availability.  Supplemental GPS data contribute to mapping weeds in the field (Downey 
et al., 2003; Tian, 2002), conducting real-time soil-characteristic surveys (Freeland et al., 
2002), and even measuring field topographic changes (Westphalen et al., 2003). 
 
  
15 
 
2.2.2 TYPICAL COMPONENTS 
 Essentially, a guidance system is comprised of an antenna that receives position 
information from navigational satellites, a receiver that processes this information, a 
computer that interprets information from the vehicle and/or receiver, and an 
input/output device for the driver (Figure 2.1).  All of these components can be 
modified and enhanced for a farmer?s specific needs and farm operations.  Some optional 
features include a receiver that receives multiple position signals for increased accuracy, 
a computer with data?logging capability for recording field attributes or application rates, 
or a user-friendly interface with audible cues and graphical displays for the driver.  Other 
equipment or vehicle enhancements may be required depending on the status of the 
vehicle, such as incorporation of an electrohydraulic steering system, addition of 
mechanical ?feelers? to sense the crop?s presence, or other such sensors (Reid et al., 
2000). 
 
    
Figure 2.1. Example guidance-system setup.  Typical guidance system components 
include a GPS antenna/receiver (a), a computer module (b), and an input/output 
device (c).  
 
 
a 
b 
c 
16 
 
2.3 GPS CORRECTION 
 Positions derived from ?raw? or autonomous GPS-satellite observations are 
typically not accurate enough for most precision-agriculture-related field operations; if a 
receiver were to calculate a location using only the positioning information from the GPS 
satellites, the location would be accurate within approximately 9 m.  Enhancements have 
been developed over the last twenty years by governmental and civilian interests to 
reduce the positioning errors inherent in the GPS system.  Depending on which 
?correction signal? one utilizes will often determine the general accuracy one may expect 
from a GPS position and, consequently, from the performance of a guidance system.  
Because certain field operations require differing levels of positioning accuracy, 
guidance-system operators must understand what range of accuracies are associated with 
which correction signal. 
2.3.1 CHARACTERISTICS OF GPS CORRECTION 
 Certain errors are inherent in GPS positioning.  Most errors can be accounted for 
in a real-time sense, and when these adjustments are included in the processing, position 
accuracy increases.  A GPS receiver that is capable of making use of these enhancements 
is referred to as Differentially-corrected GPS (DGPS).  A DGPS receiver may be further 
classified as single-frequency or dual-frequency, referring to the receiver?s capabilities to 
collect additional position information from the satellites.  With a DGPS receiver, it is 
possible to refine positional accuracy to a few inches, as opposed to several feet (Hurn, 
1993).  There are three categories of DGPS receiver, each with its own inherent 
accuracies, based on the type of differential correction used by a receiver; in addition, 
17 
 
several differential-correction signals are available by commercial means, differing in 
operational costs and advertised accuracies for each, so users have options for obtaining 
differentially-corrected positioning based on application demands and budget.  
2.3.2 WIDE-AREA AUGMENTATION SYSTEM (WAAS) 
 Practically all GPS receivers in the civilian and the agricultural community are 
capable of receiving a correction signal known as Wide- Area Augmentation System 
(WAAS).  This signal was developed by the U.S. Federal Aviation Administration 
(FAA), and is available free of charge to anyone with a compatible receiver.  
Autonomous (non-corrected) positions are calculated and compared to target coordinates 
distributed throughout the United States, along with satellite and atmospheric error-
compensation models; all these factors combine to develop a new correction model 
across the continental United States, and this correction model is transmitted by 
geosynchronous satellites across the country on the same frequency as the GPS signals.  
The result is that a WAAS-capable receiver typically has a better position accuracy than 
if the position were calculated autonomously.   With a high-quality GPS receiver, static 
accuracy (or the difference between the predicted positions versus the actual location) 
ranges from ?61 to 66 cm 95% of the time over a 24-hour period (Hest, 2007); however, 
a low-cost GPS receiver, such as those found in personal navigation devices (PND) 
utilizing WAAS, may only be accurate to between 1 and 5 m (Stombaugh et al., 2005).   
2.3.3 JOHN DEERE STARFIRE 1 (SF1) AND STARFIRE 2 (SF2) 
 Some equipment manufacturers provide proprietary correction signals.  John 
Deere? offers two such DGPS-based corrections that are compatible with their StarFire? 
18 
 
DGPS receiver. These signals are referred to as SF1 and SF2, referring to single-
frequency and dual-frequency capabilities, respectively. The method of operation for SF1 
and SF2 is similar to that of WAAS, but private companies have the capabilities of 
offering more robust error modeling or worldwide signal transmission for which WAAS 
was neither configured nor intended.  The static accuracy for SF1 is approximately 76 
cm, while the static accuracy for SF2 is approximately 25 cm (Hest, 2007).  An SF1-
capable receiver is generally regarded in the same category as a WAAS-capable receiver 
(deemed ?sub-meter?), while an SF2-capable receiver is in another category (deemed 
?decimeter?).  Apart from the costs incurred by purchasing John Deere equipment, the 
SF1 correction signal is offered free of charge to customers, while the SF2 correction 
signal requires a yearly subscription fee to maintain signal access. 
2.3.4 REAL-TIME KINEMATIC (RTK) 
 The real-time kinematic (RTK) correction signal is regarded in the GPS 
community as being the most accurate and the most reliable signal available.  The 
principles behind generating the RTK correction are a bit different than other DGPS 
correction signals.  A stationary dual-frequency GPS receiver (referred to as a base 
receiver) is configured to observe the position error from a known position; this position 
error, which only consists of three numeric values relating to error in each of three spatial 
dimensions, is transmitted by radio link to a second compatible dual-frequency receiver 
that travels with the operator (referred to as the rover receiver), along with a timestamp 
for each observation.  Given the theoretical assumption that most of the errors witnessed 
at the base receiver are virtually identical to those witnessed at the rover receiver, then 
19 
 
most of the errors can be filtered out in the rover position solution.  Static accuracy can 
be ?3 cm in many situations (Hest, 2007).  In order to obtain an RTK correction, two 
high-quality receivers are required, as well as a communications link between the two 
receivers; initial equipment costs can range into the tens of thousands of dollars, but the 
correction signal itself is free to use with compatible receivers. 
2.4 OVERVIEW OF DYNAMIC GPS 
 Statistical methods exist to quantify GPS accuracy and precision in static 
situations, but what about when the GPS receiver is in motion?  Static performance of 
receivers is not indicative of dynamic performance; in addition, techniques used to 
monitor dynamic performance vary greatly from one researcher to the next (Stombaugh 
et al., 2002).  Accuracy of a receiver in motion can still be related to how closely a 
position is determined relative to an accepted ?truth?, but determining what ?truth? is, or 
should be, in dynamic situations is challenging.  The following articles describe how 
researchers have endeavored to evaluate dynamic GPS performance. 
2.4.1 MEASURES OF ACCURACY AND PRECISION 
 According to a manuscript by Buick (2002), for farming operations conducted in 
conjunction with established equipment or land features, the primary goal of a guidance 
system is to be able to operate over time such that the equipment or land features are not 
destroyed (e.g., drip tape, controlled traffic).  When a terrestrial-based frame of reference 
is considered, Buick calls this absolute accuracy.   The author then presents the term 
relative accuracy, which refers to ?accuracy relative to some location other than a 
surveyed ?truth?.?  The author uses this term synonymously with pass-to-pass accuracy 
20 
 
(P2P); because relative accuracy degrades in a fairly short amount of time, a testing 
interval of about 15 minutes is recommended.  The term precision is used only to refer to 
the exactness of the solution, i.e. the number of decimal places the solution offers, while 
variability is the term used to conceptually define the spread of position data (inversely 
known as ?repeatability?).  Statistically, sigma values refer to variability, and RMS error 
measures spread and offset from the truth. 
 Because accuracy depends on a comparison to a true value, many researchers 
have developed testing protocols that involve ?truth in motion? procedures.  Various 
researches have been conducted utilizing either mechanical or conceptual travel paths.  A 
mechanical travel path refers to evaluating the receiver through a predictable, repeatable 
motion; examples included mounting receivers onto a rotating boom apparatus with 
constant motion and acceleration (Stombaugh et al., 2002) and mounting receivers onto a 
mobile cart that traveled back and forth via rail (Taylor et al., 2004).  A conceptual travel 
path refers to comparing receiver performance through mathematical modeling; examples 
included evaluating several receivers by studying their performance according to post-
process filtering (Han et al., 2001) and regression modeling of each ?attempted? straight-
line pass (Ehsani et al., 2003; Wu et al., 2005). Positioning performance relative to site-
specific monumentation would be considered conceptual travel as well. 
 In order to measure absolute accuracy, most researchers collect cross-track error 
(XTE) information, which relates to the ?distance between the currently measured GPS 
position and the desired track? (Han et al., 2001).  This value is typically represented in a 
direction perpendicular to the orientation of the travel path; as a result, the change in XTE 
from one end of a travel path to another would be expected to reflect how well the 
21 
 
receiver (guidance system, vehicle) followed the travel path.  In order to measure relative 
accuracy, some researchers evaluate pass-to-pass error (P2P), which relates to how well 
the guidance system can calculate the position of the next anticipated travel path, usually 
in relation to the previous one.  Utilizing mean XTE values on a per-path basis, pass-to-
pass average error, which is referred to as the difference in mean XTE of the relevant 
path and the preceding path (XTEi ? XTEi-1), can indicate whether a guidance system 
exhibits skips or overlaps by interpreting the positive or negative values (Han et al., 
2004; Ehsani et al., 2002). 
 When measuring dynamic accuracy, precautions and caveats have been offered by 
researchers.  For example, some authors describe an increase in XTE when the mobile 
test vehicle traveled latitudinally (parallel with the Equator) versus longitudinally 
(oriented with the North and South Poles) (Wu et al., 2005; Ehsani et al., 2003).  Because 
GPS satellites are not placed in an orbit relative to the north and south poles, a ?hole? 
exists in satellite coverage for mid-latitude users in the U.S.; the result is smaller cross-
track errors when driving north/south versus driving east/west (Buick, 2002).  Other 
factors to consider when evaluating guidance performance in agricultural settings involve 
vehicle speed, swath width/length, field conditions, different test days and different times 
during the day, presence of potential multipath objects, and operator bias (Ehsani et al., 
2002; Han et al., 2004); those factors are in addition to the potential error sources that 
affect overall GPS accuracy (timing errors, constellation shift, etc.). 
 
 
22 
 
2.4.2 PREVIOUS RESEARCH TO MODEL ACCURACY AND PRECISION 
 Most researchers are convinced of the reliability and repeatability of RTK 
correction such that solutions derived through RTK correction are considered as the 
?reference? positions by which to compare other systems or GPS receivers (Buick, 2002; 
Ehrl et al., 2003; Ehrl et al., 2004; Ehsani et al., 2002; Ehsani et al., 2003: Taylor et al., 
2004).  Also, according to Ehrl et al. (2004), an independent reference system should 
have at least three times the accuracy as the systems being evaluated; however, standard 
testing practice suggests accuracy of a measurement system should be at least ten times 
greater than what is being evaluated (ION, 1997).  For sub-meter and decimeter-level 
GPS receivers, an RTK-based measuring system may be adequate, but validating the 
performance of an RTK-based guidance system with an independent RTK-based 
measuring system may not provide adequate basis for analysis, especially if local site 
conditions serve to adversely affect both the reference and the measured solutions 
(Adamchuk et al., 2007).  A positioning system that offers greater accuracy than RTK 
should be considered. 
 Han et al. (2001) utilized a mathematical ?filter? to improve position and velocity 
data on a DGPS navigation system.  The filtering technique is useful for reducing effects 
from random GPS error but not GPS ?bias? error---a term that is defined but not 
addressed by this study.  The authors believe that anticipated accuracy of GPS positions 
should be 0.04 to 0.12 meter for chemical application and 0.02 to 0.04 meter for seeding, 
row-crop cultivation, and harvesting applications.  Along with summarizing results using 
mean XTE and standard deviation, maximum XTE per dataset is also presented, 
presumably in an effort to illustrate how the observed position was so contrasting from 
23 
 
the mean and was witnessed by the operators.  These values could reflect the amount of 
non-statistical variation that a farmer may witness during a single random guidance 
event. 
 Buick (2002) suggested that high relative accuracy can be expected shortly after 
establishing a reference, or AB, line for the first time, but a different position solution 
should be expected when returning to the starting point after a large passage of time.  If 
an operator desires no greater than 1- to 2-inch P2P error for farming operations, then 
RTK-based guidance is recommended, along with an autosteer system.  Buick also states 
that a GPS system that has low absolute accuracy will be undesirable for controlled-
trafficking situations, because the system cannot be relied upon to re-determine the 
positions for the established vehicle paths consistently. 
 Among other concepts, Wu et al. (2005) discussed the relationship between the 
GPS constellation and position accuracy.  Several Dilution of Precision (DOP) terms 
were discussed and modified to explain effects from N/S and E/W travel directions.  The 
authors were able to verify that there was a difference between North DOP and East DOP 
in ?mid-latitude? locations.  Standard deviations of error were always less in the northerly 
orientation than in the easterly orientation. 
 A study by Ehsani et al. (2002) outlined the issues that need to be considered in 
comparing the accuracy of different lightbar guidance systems; their contention is that, 
while the ideal situation is to have mean P2P error of zero, generally some overlap is 
more desirable in farming operations than skip.  Also, when trying to evaluate the 
performance of guidance systems, both the RMS error and standard deviation ?range of 
variation should be very close? for these values. 
24 
 
 A later study by Ehsani et al. (2003) introduced a method of dynamic accuracy 
testing for DGPS receivers and compared the dynamic accuracy of five commonly used 
low-cost DGPS receivers.  RTK GPS positions were utilized as the reference positions; to 
measure cross-track error, standard deviation of error, and root mean square error of 
positions from the tested receivers, a regression line was modeled from a subset of 
positions mapped per driving path.  (Pass-to-pass accuracy was not studied in this test.)  
The testing vehicle (a Chevy Suburban) was equipped with a lightbar computer to assist 
the driver with the location of the desired paths.  Accuracy and precision of the systems 
are defined similarly to their paper in 2002; RMSE was considered as ?absolute? cross-
track error, but the ?relative? cross-track error was not clearly stated---presumably, this 
value was witnessed by mean XTE.  Cross-track error (relative or absolute) was also 
observed to be worse (higher) when driving east-west direction instead of north-south 
direction, so one of their conclusions was to measure error in both directions for future 
evaluations. 
 In Taylor?s study (Taylor et al., 2004), testing methods for the compared GPS 
receivers involved mounting the receivers on a rail cart that traveled in an east/west 
direction.  Cross-track error was deemed the ?most important measure? for guidance 
evaluation and, in this testing scenario, northing error was considered synonymous with 
cross-track error.  Error was determined by finding the regression line that passed through 
a sample of RTK-based positions collected while the rail cart was in motion. 
 Ehrl et al. (2004) investigated the performance of an RTK-based guidance system 
that utilized two independent RTK GPS receivers (three in total on the test vehicle).  The 
purpose of this was to capture roll-and-pitch information while the vehicle was in motion 
25 
 
through a winter-wheat seeding; XTE results were ?25 mm, with a max. deviation of 100 
mm, on some straight paths; planting on the side of a slope resulted in vehicle yaw and 
implement drift up to 240 mm between passes, illustrating that vehicle terrain can impact 
path accuracy significantly in an agricultural environment.   
2.5 OVERVIEW OF TOTAL STATION 
 In the surveying industry, a total-station unit offers target positioning that can be 
correlated with GPS-derived positions and, more commonly, physical on-site monuments 
providing a local frame of reference.  While surveyors generally collect positions with a 
stationary target (like a prism), in recent years, total-station technology has progressed 
such that the target can be in motion as well.  Through redundant position calculation 
across fixed on-site monumentation, the potential exists to utilize total-station technology 
to document the travel path of a vehicle equipped with a guidance system.   
2.5.1 PRINCIPLES OF OPERATION 
 A total station, also known as a tachymeter, offers very precise angular and 
distance measurements.  Through electronic distance measuring, a laser pulse is 
transmitted by the total station and reflected back to the total station by the target, which 
can be a mirrored prism or some form of highly-reflective material.  A distance 
measurement can be derived by recording the amount of time required for the laser pulse 
to return to the device from its origin.  Also, as the sighting mechanism on the total 
station pivots horizontally and vertically, vector measurements can be recorded and 
associated with each recording (Ehrl et al., 2004).  In the surveying profession, the total 
station is often oriented over a stable monument upon which the positioning coordinates 
26 
 
are either known or can be easily assumed; in addition, the total station can be oriented 
relative to north.  The result is that any observations made of the target by the total station 
can be related to the total station?s position in three dimensions and over time.  Recent 
advances in total-station technology allow for passive autonomous tracking of the 
assigned target by the total station, resulting in a hands-off approach by the operator in 
following the target.  Some potential sources of error include radial inaccuracy, or how 
accurately each degree of horizontal or vertical rotation can be measured, and 
atmospheric distortion over long distances (Ehrl et al., 2004). 
2.5.2 PREVIOUS RESEARCH TO MODEL DYNAMIC GPS PERFORMANCE 
 Freimann (2000) and Bittner (2000) utilized a total station with tracking 
capabilities to corroborate the driving performance of an experimental ?autonavigating? 
tractor, reporting that the tractor performed better than ? 8 cm when required to follow a 
preprogrammed path.  In order to study characteristics between and within agricultural-
vehicle components, Ehrl et al. (2002) developed a system to synchronize output between 
CAN-based communication, GPS positioning information, and controller software 
performance.  The positioning system utilized for this test incorporates both RTK-based 
positioning and a ?target tracking tachymeter? (total station).  Results were that, either 
stationary or in motion, the combined reference system provided an ?absolute? accuracy 
of less than ? 3 cm.  In a later study, Ehrl et al. (2004) posed potential deficiencies with 
using a total station as a referencing system for autoguidance testing: acute angle 
measurements reduce accuracy, downrange measurements lose accuracy after 200 to 300 
meters, and tracking-vehicle speeds can exceed servo-motor performance of the 
27 
 
positioning device.  Potential dusty conditions in most agricultural settings could also 
hinder observations with a total station (Bell, 2000). 
2.6 SUBSURFACE-DRIP IRRIGATION (SDI) 
 Subsurface drip irrigation is a method of applying precise amounts of water to 
agronomic crops by delivering the water directly to the plants? root zone via underground 
tape or tubing, often referred to as ?drip tape?, ?drip line?, or ?drip tube?.  In its various 
forms, drip tape offers water-use savings due not only to its reduced application volume 
but also to its reduced evaporative losses.  The research topics discussed below involve 
the uses and benefits of SDI, and previous attempts to study SDI performance in 
laboratory and field situations. 
2.6.1 USES AND BENEFITS 
 SDI can be adapted to many typical crop applications, both agronomic and 
horticultural in nature.  In a review of all published research pertaining to SDI, crop 
yields were found to be either comparable or higher than other conventional forms of 
supplemental irrigation (center-pivot, overhead sprinkler, etc.), and usually less water 
was required (Camp, 1998).  Other studies have focused on utilizing SDI for fertigation 
or effluent disposal (Lamm et al., 2001; Lamm et al., 2004; Song et al., 2003; Trooien et 
al., 2000), determining optimum lateral spacing and lateral depth for various crop 
systems (Enisco-Medina et al., 2002; Bryla et al., 2003; Charlesworth and Muirhead, 
2003; Camp et al., 1997), and studying SDI effects on crop water-use efficiency 
(Bordovsky and Porter, 2003; Lamm and Trooien, 2003).  Increased yields and greater 
application efficiency add up to better profits for producers.  For example, in western 
28 
 
Kansas, corn irrigated via SDI provided substantial net returns on a per-hectare basis 
compared with center-pivot sprinkler irrigation (O?Brien et al., 1998).  However, the 
authors of that study determined that SDI was unprofitable relative to center-pivot 
sprinklers for SDI product life of less than ten years.   
 In economic terms, the initial costs associated with designing and installing an 
extensive SDI system can be daunting.  While installation costs may be considered a 
drawback in some situations, the cost of drip irrigation can be reduced by using both 
wider lateral spacings and the same laterals for multiple years, as with subsurface 
placement.  From this point of view, there could be a financial incentive to install and 
maintain an SDI system that could last a great deal longer than ten years (Dougherty et 
al., 2009). 
2.6.2 EVALUATING SDI PERFORMANCE 
 While much research has focused on improved agronomic growth, relatively few 
studies discuss the longevity of the drip tape systems themselves.  Effects on crop yield 
and crop quality of SDI tape design (from five manufacturers) and position along 152-m 
lateral runs were investigated by Steele et al. (1996) for three consecutive years.  Yields 
were not statistically different between tape designs for any upstream distance transect in 
any year.  However, measured emitter discharge rates decreased dramatically with 
distance downstream from the start of the tape. 
 Camp et al. (1997) used three drip irrigation systems; two with lateral lines on the 
soil surface and one with laterals about 30 cm below the soil surface for several 
experiments from 1985 to 1992.  Emitter plugging, system uniformity, and overall 
29 
 
performance were evaluated via several methods.  All uniformity values were lower for 
the subsurface system, primarily because of plugged emitters.  Entry of soil particles into 
an eight-year-old subsurface system during construction and/or repair operations 
probably caused the emitter plugging, according to the authors. 
 Ayars et al. (1999) operated several SDI experimentation plots throughout 
California over a fifteen-year period, testing fertigation regimes against crops such as 
cantaloupe, corn, cotton, and tomato.  The authors reported that some of their plots 
performed comparably to initial installation after as much as nine years.  The 
presumption was that when the system was flushed and maintained in a proper manner 
root intrusion was minimized to the emitters. 
 A commonly-referenced standard pertaining to SDI performance is ASAE S553: 
Collapsible Emitting Hose (Drip Tape) --- Specifications and Performance Testing 
(ASAE, 2008).  This testing standard outlines methods for certifying performance and 
quality characteristics of drip tape products.  The methods were developed such that 
industry testing (or consumer testing) could take place under laboratory conditions---
specifically, the tape product should be above ground in a controlled environment as 
opposed to field conditions.  Lesikar et al. (2004) adopted this standard to a field setting 
to evaluate drip-tape flow rates from SDI products that had been used for wastewater 
effluent for up to five years; this required the authors to excavate the soil to such an 
extent that small catch pans could be placed underneath the emitters in order to perform 
in-field application uniformity tests.  The authors did not refer to any plants of agronomic 
value that were removed or relocated during excavation; however, the presumption is that 
soil was disturbed within several inches of the drip tape.   In an effort to protect 
30 
 
neighboring plants, Steele et al. (1996) employed a less-intrusive method of exposing a 
few emitters in order to make a rough determination about in-field application 
uniformity.  Catch pans were also placed under the exposed emitters in this situation, and 
collected volumes were averaged and extrapolated to an equivalent flow rate based on 
manufacturer?s stated flow rates and emitter spacing. 
31 
 
CHAPTER THREE 
DEVELOPING A TERRESTRIAL-BASED PROCESS FOR MEASURING 
ACCURACY OF GPS CORRECTION SERVICES IN GUIDANCE-SYSTEM 
APPLICATIONS 
 
3.1 ABSTRACT 
 Navigational satellite services (GPS, GNSS) provide real-time position 
information for autoguidance applications used today in agriculture production.  Under 
dynamic situations (i.e. when the GPS/GNSS rover receiver is in motion), performance 
assessments are often based on measured positions relative to a target ?path?.  Creating a 
testing method that is not reliant on satellite positioning may serve to reveal the potential 
drift component of any GPS/GNSS-based correction signal when an autoguidance system 
is in operation.  The objective of this study was to evaluate techniques for using an 
optical total station to provide terrestrial-based, dynamic accuracy testing of an 
autoguidance system when using different GPS-based correction services. An optical 
total-station with target-tracking capability was utilized in conjunction with a tracking 
prism mounted to an autoguidance-equipped tractor to provide an absolute frame of 
reference for comparing accuracy and repeatability of Wide-Area Augmentation System 
(WAAS), John Deere? SF1 and SF2, and Real-Time Kinematic (RTK).  Several 
techniques relating to test-site preparation, data collection, and office processing are 
32 
 
presented.  Recommended techniques for proper correction-signal testing by the total-
station instrument included conducting site calibration, establishing semi-permanent 
monumentation for reference purposes, and conducting guidance tests in neutral-easting 
or neutral-northing path orientations.  Short-term and long-term testing can be 
accomplished with this technology for SF1, SF2, and WAAS; tracking capability of the 
total station was only four to five times more accurate than typical RTK performance, so 
this process may not be sufficient for RTK evaluation.   Total-station positioning 
accuracy can potentially be degraded unless prism tilt is considered.  Therefore, a digital 
elevation model (DEM) served to capture slope and aspect characteristics for the test site 
such that derived positions could be adjusted based on the roll and pitch modeled by the 
DEM; however, for our test site, the prism positions were only improved by about 2 cm. 
33 
 
3.2 INTRODUCTION 
 The benefits of autoguidance systems include reducing driver fatigue, extending 
operational hours, increasing field efficiency and potentially lowering fuel usage.  A 
common industry standard involves utilizing a navigational satellite service to derive 
vehicle location in real time, such as the United States? Global Positioning System (GPS), 
or the more inclusive Global Navigation Satellite System (GNSS), which includes 
satellites from the United States (GPS) and Russia (GLONASS) currently but could 
include future operational systems developed by other governments. 
 Accuracy and precision measurements of GPS- or GNSS-capable receivers have 
traditionally been derived from static-operation tests, and internationally-accepted 
engineering testing standards exist for making such evaluations (Institute of Navigation, 
1997).  Terms such as Circular Error Probable (CEP), Spherical Error Probable (SEP), 
and Root Mean Square (RMS) error pertain to differences in measured positions from a 
known target position (Upadhyaya et al., 2003).  Today, no engineering test standard has 
yet been developed to quantify GPS/GNSS positioning accuracy under dynamic 
conditions, yet most manufacturers report accuracy and performance values such as pass-
to-pass error (P2P) (also known as relative accuracy, or short-term error), which 
typically refers to accuracy of a new position relative to a position previously determined 
within a short period of time, such as fifteen minutes.  The assumption is that some form 
of dynamic testing must have taken place in order to derive such values, but as there is no 
accepted test standard in place, comparing guidance systems among manufacturers in 
equal terms becomes a challenge. 
34 
 
 Under dynamic situations (i.e. when the GPS receiver is in motion), performance 
assessments are often based on measured positions relative to a target ?path?.  Stombaugh 
et al. (2002) proposed various testing methods for GPS receivers under dynamic 
conditions, either by repeatedly driving a vehicle over known predictable paths or by 
mounting a receiver to an apparatus with a fixed travel path.  Taylor et al. (2004) also 
mounted a receiver to a fixed-path apparatus in order to investigate accuracy through 
straight-line tests.  These methods have their advantages in that most of the 
measurements generated through these processes can be easily evaluated and compared 
with an intuitive path template; also, evenly-spaced, straight-line drive patterns are often 
desired with everyday farming operations. Conversely, several equipment- and terrain-
related variables can influence the performance of a system; Ehsani et al. (2002) advised 
that dynamic evaluations should be conducted under field conditions representative of 
standard operational performances and at typical operational speeds. 
 In recent years, farmers across the U.S. have invested in autoguidance systems to 
use in their cropping systems.  Some farming practices require more accurate positioning 
from the autoguidance system than others; for example, applying manure with a 
broadcast spreader would not require autoguidance accuracy on the order of a few cm; 
accuracy to that order would be essential for banded-fertilizer applications, however.  The 
tasks desired by the operator often determine what level of accuracy is required, and an 
appropriate autoguidance system or correction signal is selected.   
 By utilizing high-quality GPS- or GNSS-capable receivers with conventional 
surveying techniques, positions can be derived to a high order of accuracy under static 
conditions. When these techniques are not available to autoguidance operators, an 
35 
 
element of drift may be introduced into the positioning error.  GPS drift can have an 
effect on field operations that rely heavily on autoguidance technologies.  Han et al. 
(2004) observed that a receiver?s dynamic performance fluctuates over time, whether by 
different hours of the test day or by different test days.  This trait could be problematic 
over the course of several seasons in situations where controlled traffic or conservation 
tillage is practiced requiring repeatable, accurate equipment paths from year-to-year.   
 Han et al. (2004) and Ehsani et al. (2002) utilized multiple GPS sensors receiving 
multiple decimeter- and sub-meter wide-area differential GPS (DGPS) correction signals 
mounted onto an agricultural vehicle in an effort to measure their performance; for these 
studies, the paths mapped via Real-Time Kinematic (RTK) correction were considered 
the ?absolute? or ?reference? paths from which to compare the other receivers? paths.  
Autoguidance operations conducted with RTK correction are often regarded as 
?repeatable? from one season to the next due to the operating parameters and 
assumptions maintained in the RTK correction.  However, when one GPS receiver is 
established as an independent ?base? from which to generate corrected positions, and few 
or no redundancies exist to support the position derivations made in the autoguidance 
software, the potential exists for GPS drift to occur under RTK operation, albeit a small 
effect.  Additionally, if local site conditions contribute to a less-than-optimum correction 
signal, then this phenomenon will be present in both the ?reference? positioning as well 
as the autoguidance-established positioning (Adamchuk et al., 2007).  Therefore, creating 
a testing method that is not reliant on satellite positioning may serve to reveal more 
accurate and absolute measures of performance for any GPS/GNSS-based correction 
signal when an autoguidance system is in operation. 
36 
 
 In the surveying industry, a total-station unit (also referred to as a tachymeter) 
offers target positioning that can be correlated with GPS-derived positions and, more 
commonly, physical on-site monuments providing a local frame of reference.  While 
surveyors generally collect positions with a stationary target (like a prism), in recent 
years, total-station technology has progressed such that the target can be in motion as 
well.  This characteristic provides the potential to plot the driving path of a vehicle.   Ehrl 
et al. (2004) discussed concerns regarding the use of total-station technology (referred to 
as Terrestrial Positioning Systems or TPS) as a form of absolute-position referencing.  
Their concerns included the drop in accuracy from a tilted prism, down-range limitations 
beyond 200 to 300 m, and issues related to the travel speed of the target vehicle.   
3.3 SUB-OBJECTIVE 
 The objective of this study was to evaluate techniques for using an optical total 
station to provide terrestrial-based, dynamic accuracy testing of an autoguidance system 
when using different GPS-based correction services. 
3.4 MATERIALS AND METHODS 
 The tractor used for this study was a John Deere? 6420 (Deere and Co., Moline, 
IL, USA; see Appendix A) factory-equipped with the StarFire? ITC? GPS receiver (part 
no. PF80732; software ver. 2.52A; Deere and Co., Moline, IL, USA) and the GreenStar? 
autoguidance system (hardware ver. PF80444 2.5; software ver. PF303192F; Deere and 
Co., Moline, IL, USA: see Appendix D).  The tractor was also outfitted with a Trimble? 
EZ-Guide? lightbar guidance computer (firmware ver. 4.10.002 Feb 22 2007 11:53:46 
for 2007 tests; Trimble Agriculture, Sunnyvale, CA, USA) and an EZ-Steer? assisted-
37 
 
steering system (software ver. 3.00.01 for 2007 tests; Trimble Agriculture, Sunnyvale, 
CA, USA: see Appendix D). 
 In order to capture moving-vehicle position data with a terrestrial frame of 
reference, we used a Trimble? 5603 DR+ total station (firmware ver. 696-03.08; Trimble 
Engineering, Sunnyvale, CA, USA) in combination with a Trimble? TSCe field data 
collector (part no. 45185-20; Trimble Engineering, Sunnyvale, CA, USA) utilizing 
Trimble? Survey Controller software (versions 11.20 ? 11.40; Trimble Engineering, 
Sunnyvale, CA, USA: see Appendix B).  Point of measurement on the tractor was 
maintained using a Trimble? RMT ATS Multi-Channel tracking prism (part no. 
571233035; Trimble Engineering, Sunnyvale, CA: see appendix C), mounted to the rear 
frame hitch approximately 66 cm above the ground and aligned with the vehicle?s 
centerline.  Distance measurements were calculated via Electronic Distance Measurement 
(EDM) technique; the stated accuracy of such measurements was ? (5 mm + 2 ppm; 
Trimble Engineering, 2001).  In addition, our total station was equipped with Trimble?s 
AutoLock? function such that the total station, via servo motors integrated within its 
base, could automatically track the moving target prism and simultaneously collect 
distance data. 
 Four different GPS-based correction signals were evaluated: Wide-Area 
Augmentation System (WAAS), John Deere SF1 (SF1), John Deere SF2 (SF2), and Real-
Time Kinematic (RTK).  Position accuracies typically reported by manufacturers and/or 
system users are referenced in Table 3.1.  Measures of short-term error (P2P) and long-
term error are often reported; however, long-term error is often reported from static tests, 
not dynamic tests.   No implements were attached during the tests, so draft or drag was 
38 
 
not considered; also, the tractor had performed only light-duty tasks during its life, so 
steering and/or transmission wear was considered negligible. 
Table 3.1.  Typical reported accuracies for correction signals used in this study. 
Correction Signal Type of Correction Accuracy 
WAAS wide-area ?15 to 30 cm short-term (P2P); ?61 to 66 cm long-term (static) 
SF1 wide-area ?33 cm P2P; ?76 cm static 
SF2 wide-area ?10 cm P2P; ?25 cm static 
RTK local-area ?3 cm P2P; ?3 cm static 
Source: Hest, 2007.  For additional details, see Appendix E. 
3.4.1 GENERAL TEST-SITE PREPARATIONS 
 This study was performed in Auburn, Alabama (32.6? N, 85.5? W) beginning in 
February 2006 and ending in August 2007.  A different testing site was selected each year 
(see Appendices F and G).  Except for differences in local terrain, both sites shared the 
characteristics of minimal overhead obstruction (desired for maximum reception of 
available satellites), minimal sources of radio interference (desired for maximum 
reliability of positioning data), and allowances for line-of-sight testing (desired for 
uninterrupted prism tracking by the total station).  
 Semi-permanent position monuments, or ?benchmarks,? were installed at each 
testing site to serve as points of reference for total-station data collection.  In order to best 
relate total-station derived positions to the local terrain, a four-point site-calibration 
technique was employed via Site Calibration procedure in Trimble Survey Controller 
program (Trimble Surveying, 2007). Additional monumentation was installed and 
39 
 
documented as necessary for consistent total-station viewing location and station 
backsight monument(s). 
 For both sites, guidance ?paths? were input to the guidance computers at the first 
data-collection event (referred to as ?AB? line placement) by parking the tractor at the 
start of the path, recording a coordinate at this location (point ?A?), driving to the end of 
the path, parking, and recording a subsequent coordinate (point ?B?).  Path placement and 
direction of travel were optimized for total-station line-of-sight availability.  Positions 
were collected via the total station during tests when all the following conditions were 
met: 
1. The tractor operator selected the relevant correction service. 
2. The operator aligned the tractor on the relevant pass (1, 2, etc.). 
3. The autoguidance device was engaged by the operator. 
4. The guidance system steered the tractor on path.  Total-station positioning was 
terminated before the tractor reached the end of the pass. 
 Five straight-line parallel paths were navigated by the guidance system(s) for each 
correction signal being evaluated.  Vehicle speed was maintained at 2.2 m s-1 for all tests.  
A testing ?event? was represented by traversing all five reference lines within a 15-
minute time period.    A different correction signal was enabled prior to each testing 
event.   
3.4.2 TEST SITE ONE PREPARATIONS 
 The RTK base receiver/radio was installed atop a standard instrument tripod over 
a dedicated benchmark at the start of each testing day, located approximately 36 m south 
of the AB line with clear line-of-sight to the vehicle at all times.  A GPS-derived 
40 
 
coordinate for the base receiver was determined instantaneously on the first day of testing 
and stored in the guidance-system program as a designated base location. 
 AB-line establishment was made for both the John Deere GreenStar system and 
the Trimble EZ-Guide system concurrently.  However, these systems operated 
independently of one another; the GreenStar system computed an AB line using RTK-
quality positioning, while the Trimble system computed an equivalent AB line using 
WAAS-quality positioning.  Guidance lines were established at 6.1-m spacing; direction 
of travel was approximately 80.1 degrees azimuth relative to true north (Appendix F).  
3.4.3 TEST SITE TWO PREPARATIONS 
 For the second site, a semi-permanent base-receiver mount was installed with a 
clear line-of-sight of the testing site, located approximately 30 m from the AB line(s).  In 
the effort to further maximize RTK repeatability, instead of collecting an instantaneous 
coordinate, the base receiver was programmed to store a coordinate representative of 
mean position after a 24-hour data compilation.   
 While we utilized the same John Deere GreenStar system for SF1, SF2, and RTK 
corrections, the Trimble EZ-Guide computer used in 2006 for WAAS correction was 
different from the one used in 2007 (but make and model were identical for both).  Two 
additional configuration changes were made to the newer Trimble EZ-Guide.  First, we 
devised a method of outputting the NMEA-0183 data stream (an ASCII standardized data 
format developed by the U.S. National Marine Electronic Association) from the StarFire 
iTC receiver and into the Trimble system.  This modification served two purposes: (1) 
while maintaining an RTK-quality position fix, AB-line establishment was conducted 
41 
 
concurrently with both systems utilizing the same vehicle receiver, thereby reducing 
potential offset error between the systems, and (2) while the StarFire receiver and the 
GreenStar system could still process WAAS, the signal was sent via NMEA output to the 
Trimble system so that the EZ-Steer module could function. The second change to the 
EZ-Guide configuration was the addition of a Terrain-Compensation Module (TCM: see 
Appendix D); the belief was that roll/pitch conditions exhibited by the vehicle throughout 
the testing area would tend to not adversely affect position accuracy as may have been 
the case in the previous year?s tests. 
 More intensive topographic surveys were conducted at this site than the previous 
one, as the intent for the second year of tests was to attempt to compensate for prism-tilt 
error.  Utilizing a survey-grade RTK receiver, elevation data were recorded throughout 
the plot.   
 Finally, this test site allowed both north-to-south and east-to-west orientations for 
vehicle paths to be studied.  This setup provided a means to evaluate whether drive-path 
orientation impacted dynamic positioning accuracy.  A swath width of 12.2 m was used 
during this testing; directions of vehicle travel were approximately 179.4 degrees and 
271.3 degrees relative to true north for the north-to-south and east-to-west tests, 
respectively (Appendix G). 
3.4.4 TEST SITE ONE: DATA ANALYSIS 
 During both years of testing, positions derived using the total station were 
converted into Universal Transverse Mercator (UTM) coordinates (Northing and Easting, 
42 
 
in meters). Ground distances could be easily derived with UTM coordinates, and the 
UTM projection allowed for easy transfer to existing data imagery of the sites. 
 While the guidance computers used stored AB line information on each 
successive test, no clear method of documenting the AB line with the total station was 
determined in 2006.  Referencing research conducted by Taylor et al. (2004), a linear-
regression model was used to evaluate path placement on successive testing events 
relative to the first path created by the guidance system on the first day of testing at Site 
1.  For example, the first pass along the AB line when using the RTK signal generated 
about 30 total-station derived point coordinates.  The linear regression equation obtained 
from those 30 points became the ?reference? by which to compare cross-track error 
(XTE) and pass-to-pass error (P2P) for RTK performance throughout the season.  An 
illustration of XTE determination is presented in Figure 3.1.  Furthermore, the target 
positions were ?rotated? about a predicted point on the regression line in order to relate 
all the vehicle paths relative to a positive easting and zero northing direction by using the 
following equations: 
g1850g3033 g3404 g1850g3036 g4674uni0063uni006Funi0073g4672 g3343g3117g3124g3116g4666uni0033uni0036uni0030g3398 g2016g4667g4673g4675g3398 g1851g3036 g4674uni0073uni0069uni006Eg4672 g3343g3117g3124g3116g4666uni0033uni0036uni0030g3398 g2016g4667g4673g4675  (1) 
g1851g3033 g3404 g1851g3036 g4674uni0063uni006Funi0073g4672 g3343g3117g3124g3116g4666uni0033uni0036uni0030g3398 g2016g4667g4673g4675 g3397 g1850g3036 g4674uni0073uni0069uni006Eg4672 g3343g3117g3124g3116g4666uni0033uni0036uni0030g3398 g2016g4667g4673g4675  (2) 
where (Xi , Yi) are initial prism coordinates (Easting, Northing), (Xf , Yf) are modeled 
prism coordinates, and ? represents a degree conversion of the arc-tangent of the slope 
parameter from the regression-line equation.  The results were that the northing error was 
equivalent to XTE for a given path, making performance calculations much easier to 
conduct.  (Additional details are in Appendix L.) 
43 
 
 
 
Figure 3.1.  Depiction of cross-track error (XTE) determination. 
3.4.5 TEST SITE TWO: DATA ANALYSIS 
 To account for prism movement during the second-year testing, a digital elevation 
model (DEM), produced using ArcMap GIS Ver. 9 (ESRI, Redlands, CA, USA) was 
created to represent the study site.  The DEM was produced with a 1.0-m2 resolution.  
Further, slope and aspect models were also generated.  The benefit of these models was 
that, when tracking positions were included, a derived slope (in degrees) and aspect 
(related to true north) could be associated with the tracking positions to account for prism 
tilt.  Further details regarding production of slope/aspect models and adjusting target 
coordinates for prism tilt are available in Appendix M. 
 While no equivalent AB line was recorded with the total station in the first year, a 
representative AB line was recorded for the tests in the second year.  On the initiation 
date, target-prism positions were recorded concurrently with the establishment of the 
guidance-system field records (under RTK correction) via an ?A + B? method.  Relating 
this information to the slope and aspect models used to adjust position data, any total-
44 
 
station-derived position could be modeled from an [Aunadj. + Bunadj.] or [Aadj. + Badj.] target 
path. 
 Example target position data are illustrated in Figure 3.2.  One additional change 
to XTE derivations was made in the second year of testing.  Target positions (represented 
by point coordinates) were exported as GIS-compatible files for further analyses in 
ArcMap.  Through GIS procedures, polyline files were modeled according to anticipated 
swath width and corresponding with total-station-derived AB line model (expected 
guidance path).  To obtain XTE, we used a GIS procedure that provided perpendicular 
distance offset of each point coordinate relative to the expected guidance path.  (Details 
of ArcMap procedures are provided in Appendix M.)  For further visual representation, 
polyline files were created by connecting consecutive points per swath, reinforcing the 
concept of observed vehicle travel compared to expected guidance path (Figure 3.3).  
45 
 
 
Figure 3.2.  Example dataset produced from total-station observations.  The labels 
for the gridded lines represent local grid coordinates, projected into the Universal-
Transverse Mercator (UTM) coordinate system (units are in meters).  The 
triangular markers represent stable reference monuments used in site calibration.  
The green lines represent the vectors from the total station monument to the 
observed prism positions (illustrated as white dots). 
46 
 
 
Figure 3.3.  Example test dataset in GIS format.  Black dots within each line refer to 
position data recorded by the total station; lines drawn between dots represent 
travel ?paths? made by the test vehicle.  Paths are separated by color according to 
correction signal used for each traverse in relation to the relevant reference 
guidance line.   
3.4.6 STATISTICAL ANALYSES 
 For both years of the study, mean XTE, standard deviation of XTE (SD), and root 
mean square error of XTE (RMSE), were calculated as outlined by Ehsani et al. (2002)  
(also outlined in Appendix K).  Pass-to-pass errors (P2P) were calculated by subtracting 
the signed value of the mean XTE of a row from the preceding row.  Per-event 
observations were screened for potential outliers, defined by ?any observation farther 
than 1.5 [Inter-Quartile Range] from the closest quartile? (Devore and Farnum, 2005). 
Basic statistical measures of XTE and tests for normality were derived through the 
47 
 
univariate-statistics procedure (PROC UNIVARIATE; SAS Institute, Cary, NC, USA).  
Graphical depictions were generated in MS Excel 2007. 
3.5 RESULTS AND DISCUSSION 
3.5.1 SITE ONE RESULTS 
 Data collection dates and times are referenced in Table 3.2.  Implementation of 
the desired correction signals was a large challenge throughout the test period.  While the 
intention was to navigate the five drive paths within a 15-minute time period, select a 
different correction signal, and repeat the navigation, certain correction signals would not 
initiate in a timely manner between RTK, SF1, and SF2.  One potential reason may have 
been that the processor in the GreenStar computer could not re-compile the satellite data, 
filter the data using the newly-chosen correction-signal algorithm, and re-compute a 
vehicle position as quickly as was demanded.  This conjecture does not relate to WAAS 
data collection being unavailable after March 28, however; the Trimble EZ-Guide 
computer and EZ-Steer accessory had been loaned to the Biosystems Department for the 
first half of the study and had to be returned.  Because of time constraints, problematic 
signals were abandoned during a testing day. 
48 
 
  
Table 3.2. Data collection events for each correction signal in 2006 (Site 1). 
Test 
Namea 
 Dateb Time of 
Dayc 
 Test 
Namea 
 Dateb Time of 
Dayc 
         
R11  February 22 2:30 pm  R51  April 7 1:33 pm 
J11  --- ---  J51  April 7 1:59 pm 
D11  --- ---  D51  April 7 1:45 pm 
W11  February 22 2:45 pm  W51  --- --- 
         
R21  February 27 3:03 pm  R52  April 7 2:15 pm 
J21  --- ---  J52  April 7 2:36 pm 
D21  February 27 3:28 pm  D52  April 7 2:25 pm 
W21  February 27 3:16 pm  W52  --- --- 
         
R22  February 27 3:40 pm  R61  June 14 12:28 pm 
J22  --- ---  J61  --- --- 
D22  February 27 4:00 pm  D61  June 14 12:42 pm 
W22  February 27 3:50 pm  W61  --- --- 
         
R31  March 8 1:47 pm  R62  June 14 3:22 pm 
J31  --- ---  J62  June 14 3:49 pm 
D31  March 8 2:15 pm  D62  June 14 3:35 pm 
W31  March 8 2:04 pm  W62  --- --- 
         
R32  March 8 2:30 pm  R71  June 16 9:43 am 
J32  --- ---  J71  June 16 10:07 am 
D32  March 8 2:52 pm  D71  June 16 9:55 am 
W32  March 8 2:42 pm  W71  --- --- 
         
R41  March 28 T1d  R72  June 16 12:21 pm 
J41  --- ---  J72  June 16 12:43 pm 
D41  March 28 T1  D72  June 16 12:32 pm 
W41  March 28 T1  W72  --- --- 
         
R42  March 28 T2      
J42  --- ---      
D42  March 28 T2      
W42  March 28 T2      
         
a  Test name is in the format A
xy, where A represents correction signal used (R=RTK, J=StarFire SF1, 
D=StarFire SF2, W=WAAS), x represents testing day, and y represents event number for that day. 
b  Dashed lines represent absence of data for that iteration. 
c  Time of Day refers to local time at which data collection began, recorded by program clock embedded in 
datalogger. 
c  Events T
1 and T2 have no time stamps. 
 
49 
 
 Correction-signal results for all signals will be discussed in more detail in Chapter 
4; however, a summary of XTE measures for the 2006 season is presented in Table 3.3.  
Statistical evidence implied that XTE data did not always follow a normal distribution.  
XTE generated through RTK correction had the greatest number of testing events (85%) 
that were normally distributed (if [Pr<W] > 0.05 for Shapiro-Wilk test); 40 to 50 percent 
of the test events evaluating other correction signals suggested normal XTE distributions.  
Negative or positive XTE median and XTE mean values represent XTE trends that were 
to the northern or southern side of the reference line, respectively.  RTK correction 
offered very low position error in terms of variability (5-cm SD, 6-cm IQR), location (5-
cm XTE median/XTE mean), and overall accuracy (7-cm RMSE), but the values are 
slightly higher than values typically reported for RTK performance.  SF1 correction 
varied over a meter throughout the season, yet tended to be centered close (5 cm) to the 
desired path; however, high RMSE suggested that this signal was the most inaccurate for 
the time period used in this study.  SF2 correction offered XTE farthest from the 
reference lines (37-cm XTE median; 35-cm XTE mean).  Finally, WAAS correction 
offered good general location throughout the season (9-cm XTE median/mean) despite 
the overall accuracy (31-cm RMSE), which was consistent with typical reported accuracy 
for this correction signal. 
50 
 
 Table 3.3.  Measures of location and assumption of normality for 2006 XTE data. 
Correction 
Signal 
? Test 
Events 
# Events 
Rejecting 
Normality 
Median 
XTE 
(cm) 
Inter-
Quartile 
Range 
(IQR) of 
XTE (cm) 
Mean 
XTE 
(cm) 
SD 
(cm) 
RMSE 
(cm) 
RTK 13 2 -5 6 -5 5 7 
SF1 5 3 5 116 38 68 78 
SF2 12 6 37 30 35 22 41 
WAAS 7 4 -9 38 -9 30 31 
 
 Considerations should be made concerning the measured accuracy of the RTK 
correction signal as related to the measurement accuracy of the total station.  The 
?tracking? capabilities of the total station were accurate to ?0.5 cm under ideal 
circumstances (Trimble Engineering, 2001).  While position bias may have been 
introduced from a tilted targeting prism, we nonetheless believed we were achieving 
optimum performance of the total station.  However, typical static-testing standards for 
GPS receivers require that measurement devices should be at least one order of 
magnitude more accurate than the product being evaluated (Institute of Navigation, 
1997); ergo, a correction signal that was expected to achieve an accuracy of 3-cm should 
have been evaluated with a unit or process that could achieve 0.3-cm accuracy or better.  
Standard (static) total station measurements would have met this criterion, but a 
stationary prism measurement could only be made if the tractor were stationary as well.  
More modern total stations may be able to provide necessary accuracy under target-
tracking conditions, but the unit used during this study could not meet this threshold.  
Therefore, for ?typical? RTK-derived XTE observations, this measuring process was 
51 
 
considered questionable.  RTK statistics are nevertheless presented in this and the next 
chapter to offer comparisons with the wide-area correction signals. 
 Several points can be made regarding the lessons learned from first-year data 
collection.  Unfortunately for our testing regime, the autoguidance controller system 
integrated into the tractor was unable to operate using the WAAS correction with the 
StarFire iTC receiver.  It was this limitation that required the use of a secondary ?hands 
free? guidance system solution to evaluate WAAS, namely, the Trimble EZ-Guide and 
EZ-Steer hardware bundle.  Of course, this platform was entirely independent of the 
GreenStar computer and StarFire iTC receiver used for the other three correction signals.  
A second antenna was used for the Trimble guidance system, and while the ideal antenna 
placement for optimum system performance was the point corresponding to the vehicle 
centerline and the forward-most point on the cab roof, this location was used by the 
StarFire receiver.  Additionally, the Trimble autoguidance package used for this study 
was not equipped with terrain-compensation capability like the StarFire iTC receiver.  
The consequences of this scenario were not apparent in the overall performance of 
WAAS (as represented by mean XTE per event), but instead were apparent when a close 
evaluation of the drive paths over time was conducted (Figure 3.4). In Figure 3.4, there is 
a negative relationship between XTE and distance from the origin. This information led 
us to consider the possibility that the trend was indicative of roll/pitch error from the site, 
and therefore needed to be accounted in future tests.  However, another possibility that 
would have explained this trend was that XTE observations were not perfectly adjusted 
after subjecting the positions to rotation equations (1) and (2).  This tendency was 
52 
 
witnessed in other datasets as well; therefore, an alternative method of calculating XTE 
had to be considered.  
 
 
Figure 3.4.  Portions of drive paths under WAAS correction for 2006 (site 1).  The 
negative relationship between XTE and Easting values suggests roll/pitch error 
and/or insufficient XTE calculations. 
 
 Regarding data processing, using the regression-line technique to evaluate 
repeatability of correction signals was not intuitive from an operator?s point of view.  In a 
production setting, an operator would not have had the capabilities of documenting an 
AB line by this procedure, but rather by the AB-line establishment method described 
above.  Therefore, we considered the feasibility of capturing the prism positions 
concurrently with future AB-line establishment, allowing for a total-station-derived AB-
line equivalent that could serve as an absolute path reference during office processing 
calculations.  
 Site conditions in 2006 led to several set-up changes that were implemented in 
2007.  For example, conditions were not conducive to leaving the RTK base receiver in 
53 
 
its original location throughout the duration of this study, or for establishing a permanent 
monument to accommodate the receiver.  While the coordinate stored in the guidance 
computer referred to a three-dimensional location (represented as latitude, longitude, and 
altitude), no precise measures were undertaken to reposition the receiver at the identical 
vertical height or in the identical orientation relative to true north; either condition may 
have contributed to reduced repeatability of RTK positioning.  Only replication of reset 
relative to latitude and longitude was conducted by centering the base receiver over an 
established benchmark using a plumb-bob.  Therefore, in an effort to reduce potential 
measurement error, one of the improvements sought for the next season was to provide a 
semi-permanent receiver mount such that latitude, longitude, altitude, and receiver 
orientation could be repeated throughout all tests. 
 Due to size limitations, an average of 187 observations per test event were 
documented with the total station; while this number was considered sufficient to provide 
evidence of a normal distribution, more observations were desired for each pass during a 
test.  Also, previous research (Wu et al., 2005) had suggested that path orientation 
affected XTE trends, so we wanted to find a larger test site that would allow for neutral 
easting (north-to-south or south-to-north) and neutral northing (east-to-west or west-to-
east) comparison studies. 
 Finally, measurement error contributing from a tilted prism was not considered 
for 2006 tests.  From a conventional-surveying perspective, in order to accurately relate 
surface positions relative to existing monumentation (or simply to the total station 
position), the leading assumption is that the total station and the targeting prism are 
plumb.  Given that assumption, ground positions would be consistent horizontally and 
54 
 
adjusted vertically based on instrument and/or target height above the ground.  Having 
either the total station or the target prism out of plumb violates this assumption, and 
horizontal and vertical accuracy is degraded.  Internal sensors in the total station provided 
immediate indication of instrument error if the device was not plumbed within acceptable 
tolerances; however, no such mechanism existed for the target prism.  Therefore, we 
wanted to find a method of predicting prism tilt as the vehicle traveled throughout the test 
plot. 
3.5.2 SITE TWO RESULTS 
 Data collection dates and times are referenced in Appendix J.  While initializing 
the SF1 correction signal was problematic in 2006, initializing the RTK signal was more 
problematic in 2007.  Because of the overall lack of testing events recorded on a 
correction-signal basis in the previous year, the research objective for the 2007 study was 
to observe twenty testing events per signal and per orientation; only 13 events were 
recorded for east-to-west RTK, but all other signal-orientation combinations had 17 to 20 
test events. 
 The size of the second test site allowed for much longer traverses in the north-to-
south direction than was witnessed at the earlier site, but the traverses in the east-to-west 
direction were about the same length as those at the earlier site.  An average of 289 
observations was recorded per north-to-south testing event, while east-to-west testing 
events averaged 157 observations. 
 Summary positioning statistics for all correction signals for north-to-south and 
east-to-west path orientations are presented in Table 3.3.  Similar to 2006, XTE generated 
55 
 
through RTK correction had the greatest number of testing events (70%) that were 
normally distributed (if [Pr<W] > 0.05 for Shapiro-Wilk test); only 20 to 25 percent of 
the test events evaluating SF1 and SF2 correction signals suggested normal XTE 
distributions, and only about 30 percent of test events evaluating WAAS exhibited 
normality.  As evidenced from IQR values, the drive-path orientation tended to influence 
XTE; as this tendency had already been suggested by previous research efforts, the 
validation of the concept through the total-station-derived data was acceptable.  However, 
this tendency was not observed for the SF2 signal.  Given that variability of SF2 signal 
was about equal (43-cm southward, 45-cm westward) and that XTE median values of the 
SF2 signal were closer to zero (expected value) than with either SF1 or WAAS, the 
suggestion is that the SF2 correction was the most reliable sub-meter correction signal 
evaluated.  Measure of variability for SF2 in 2006 supports this suggestion as well (Table 
3.2).  SF1 offered consistent position error regardless of path orientation (21-cm XTE 
median), but the variability in the southward orientation was more than twice that 
observed in the westward orientation.  XTE from WAAS correction tended to be more 
?on target? throughout the season in the southward, as opposed to the westward, 
orientation, but XTE variability was about equal to that observed under SF1 correction. 
56 
 
Table 3.4.  Measures of location and assumption of normality for 2007 XTE data. 
Correction 
Signal a 
? Test 
Events 
# Events 
Rejecting 
Normality 
Median 
XTE 
(cm) 
Inter-
Quartile 
Range 
(IQR) of 
XTE (cm) 
Mean 
XTE 
(cm) 
SD 
(cm) 
RMSE 
(cm) 
RTKS 17 4 -1 10 -1 8 8 
RTKW 13 7 4 19 5 15 15 
SF1S 20 15 21 83 12 47 48 
SF1W 19 15 21 38 17 33 37 
SF2S 20 16 3 43 5 29 29 
SF2W 20 16 6 45 3 52 52 
WAASS 20 13 -10 56 -13 65 67 
WAASW 19 14 40 75 27 73 78 
a Subscripts ?S? and ?W? refer to southward-oriented and westward-oriented reference lines, respectively. 
3.5.3 INVESTIGATIONS INTO NORMALITY OF XTE 
 Because several testing events for the 2006 and 2007 seasons produced non-
normal XTE, further analyses were performed to understand this result.  Outlier filtering 
was performed as described by Devore and Farnum (2007) for XTE observations per 
testing event; an alternative approach was to perform a similar outlier filtration across the 
entirety of each season.   This approach was investigated; depending on the correction 
signal being evaluated, the results were that entire testing events should have been 
omitted.  This scenario was unacceptable for our project goal of comparing XTE trends 
on a per-event basis; further, we believed that, as a producer does not necessarily have 
the capability or the time to simply ?omit? and ?redo? a guidance-based operation that 
may not be indicative of typical correction-signal performance, we wanted to document 
and evaluate XTE observations in a similar fashion. 
 Outlier filtration on a per-testing-event basis did prove useful in spotting XTE 
observations that should not have been included in the datasets.  Usually these 
57 
 
observations were located at the beginning or the ending of individual guidance rows, 
implying that prism targeting occurred before the tractor ?locked? on the guidance (AB) 
line, or that prism targeting continued briefly (and erroneously) after the autoguidance 
feature was disabled at the end of the guidance line, respectively. In a few situations, 
outlier flagging suggested evidence that XTE observations were different on a per-row 
basis.  From the 2006 study, out of a total 178 rows across all correction signals, 
observations from Row 1 during a June 16 SF2 test event were flagged and removed 
from analysis (see Appendix N); a graph of XTE observations for this event (with XTE 
from Row 1 included) is presented in Figure 3.5.  In 2007 from a total of 385 rows 
involved with north-to-south path orientation across all correction signals, only 6 rows 
were flagged and omitted from analyses (Rows 1 and 2 from two SF1 testing events; 
Row 2 from an SF2 event; three instances of Row 1 from separate WAAS events; see 
Appendix O).  Out of 355 rows involved with east-to-west path orientation, only 2 rows 
were omitted from analyses (two instances of Row 1 from SF1 and SF2 testing events; 
see Appendix O). 
58 
 
 
Figure 3.5.  Graph of XTE observations per row for SF2 testing event on June 16, 
2006.  Observations from Row 1 were flagged as potential outliers and omitted from 
analyses. 
 Despite these evaluations, XTE observations still suggested evidence of non-
normality on a test-event basis.  A contributing factor to this tendency was likely from 
poor satellite geometry or low satellite count at the time of data collection; situations that 
are generally regarded as adverse to GPS positioning.  Neither satellite number nor 
measures of geometry were recorded during testing events. 
 Another factor contributing to non-normality may have been the strong 
dependence of XTE performance to the error-modeling accuracy (or lack thereof) of the 
wide-area correction signals.  In differential GPS techniques, a wide-area correction 
refers to modeling timing, satellite, and atmospheric error over a ?wide area? of the earth.  
For WAAS, base stations throughout the United States contribute to an error-mitigation 
model representative of the United States (Hurn, 1993), and for SF1 and/or SF2, base 
stations are distributed around the world to create three wide-area correction models; one 
of which models North and South America from 76 degrees north latitude to 76 degrees 
59 
 
south latitude (Navcom Technology, 2003). Error modeling for a very small, specific 
portion of the planet?s surface is best served through ?local area? corrections like RTK.  
The algorithms that comprise SF1 and/or SF2 error modeling are proprietary to Navcom 
Technology (a subsidiary of John Deere), so an evaluator would be unlikely to simulate 
or replicate the error modeling of these services.  In fact, due to the dynamic nature of the 
GPS satellite constellation, true testing replications are impossible to achieve, but 
approximate replications may be possible. 
3.5.4 POSITION ADJUSTMENTS VIA SLOPE AND ASPECT MODELING 
 Digital elevation modeling for the test site revealed that the site, although level in 
appearance, had a uniform negative slope almost directly towards the west (see 
Appendices F and G).  The average slope value was 1.5 degrees.  Given that the tracking 
prism, mounted on the rear hitch of the tractor, was 90 cm above ground at this site (in 
order to overcome line-of-sight issues from ground vegetation such as weeds), this 
translated to a theoretical eastward position adjustment of approximately 2.4 cm.  The 
positions were not adjusted in such a broad manner, however; each unadjusted total-
station position was adjusted based on the grid value from its underlying DEM.  By 
adjusting the positions in this manner, overall XTE error was more representative of 
autoguidance performance.  Average position adjustment with the southward drive paths 
was 2 cm, while average adjustment with the westward drive paths was <1 cm. 
 One issue that was introduced during the conceptualization and implementation of 
the slope/aspect modeling involved choosing the DEM resolution.  Because the vehicle 
speed during these tests was 2.2 m s-1, and, because the total station could only process 
60 
 
positions no faster than 1 Hz, slope and aspect values were calculated on a 1.0-m2 grid.  
A greater DEM resolution would serve to capture a more complete picture of the ground 
surface in the test plot but require far greater labor resources to accurately produce and 
ground-truth the model.  On the other hand, as the tractor footprint is far larger than 1.0 
m2, a less-resolute DEM may better explain the characteristics of a suspension-enabled 
vehicle in motion.  We believe a far simpler solution to capturing prism tilt information 
would have been to install accelerometers or gyroscopic devices at the prism.  Acquiring 
additional equipment would have incurred more project costs, so we wanted to utilize 
tools that were ?in-house? (like GIS software and survey-grade GPS units) to account for 
the phenomenon. 
3.5.5 CONSIDERATIONS FOR CONTINUING RESEARCH 
 Using a total station with target-tracking capability served us well for 
documenting the drive path of our autoguidance-equipped tractor.  This technique served 
particularly well in evaluating SF1, SF2, and WAAS.  Because RTK correction is 
typically regarded as accurate to less than 3-cm, the tracking capabilities of our total 
station were, at best, five times more accurate than RTK observations.  However, a 
measurement accuracy of 10 times better than typical RTK could have provided more 
robust statistical details about RTK.  Modern total stations may satisfy this requirement. 
 Documenting vehicle prism positions during AB-line establishment served as a 
more intuitive reference method for XTE distribution throughout subsequent test events.  
Also, XTE observations supported the notion that position determination has different 
error levels between the northing and easting components, so evaluating correction-signal 
61 
 
performance with easting-neutral and northing-neutral path orientations should be 
strongly considered for future testing.   While accounting for potential prism tilt may not 
have contributed greatly to position accuracy in our study, reducing error is nonetheless 
good policy, so some form of error mitigation should always be considered when travel 
conditions become more representative to typical farming applications.  
3.6 SUMMARY 
 The techniques evaluated for this study involved following a GPS-based 
autoguidance-assisted vehicle with an optical total station with target-tracking capability 
in order to create a terrestrial absolute frame of reference when comparing correction 
signals.  Recommended techniques for proper correction-signal testing by the total-
station instrument include conducting site calibration, establishing semi-permanent 
monumentation for reference purposes, and conducting guidance tests in neutral-easting 
or neutral-northing path orientations. Proper application of a total station can be used to 
investigate autoguidance behavior using correction signals with typical decimeter 
accuracy or greater, like SF1, SF2, and WAAS; XTE error from RTK correction signal 
could not be sufficiently documented in comparison to inherent tracking-mode error of 
the total station.  Due to the potential measurement errors derived from a tilted tracking 
prism, a digital elevation model served to capture slope and aspect characteristics for the 
test site such that derived positions could be adjusted.  This adjustment method relies 
strongly on the susceptibility of the tracking prism to excessive pitch/roll of the vehicle, 
and this susceptibility can be minimized by installing the prism as close to the ground 
surface as possible,or by adding accelerometer/gyroscopic hardware to the prism. 
62 
 
CHAPTER FOUR 
ACCURACY OF GPS CORRECTION SERVICES OVER TIME FOR 
GUIDANCE-SYSTEM APPLICATIONS 
 
4.1 ABSTRACT 
 Navigational satellite services (GPS, GNSS) provide real-time position 
information for autoguidance applications.  Under dynamic situations (i.e. when the 
GPS/GNSS receiver is in motion), performance assessments are often based on measured 
positions relative to a target ?path?.    The objective of this study was to compare short-
term and long-term accuracies of an autoguidance system when using different GPS-
based correction services.  An optical total-station with target-tracking capability was 
utilized to provide an absolute frame of reference when comparing accuracy and 
repeatability of Wide-Area Augmentation System (WAAS), John Deere? StarFire 1 
(SF1) and SF2, and Real-Time Kinematic (RTK) during 2006 and 2007 seasons.  Cross-
track error (XTE) was influenced by drive-path orientation for RTK, SF2, and WAAS 
correction signals; SF1 performance was consistent regardless of orientation.  Southward 
XTE was typically more accurate than westward XTE.  SF2 short-term and long-term 
errors were much greater than typically reported values.  XTE derived under SF1 and 
WAAS correction were more representative of typical manufacturer?s claims.  Unsigned 
63 
 
95% short-term and unsigned 95% long-term error was preferred to describe guidance 
performance over signed P2P and Y2Y, respectively.  
4.2 INTRODUCTION 
 One of the most significant advancements introduced into the precision-
agriculture market has been that of assisted-steering, or autoguidance, for farm 
machinery.  Primarily this technology allows the vehicle operator to focus his or her 
efforts on monitoring real-time operations other than navigating the vehicle, such as 
implement activity.  A few of the many benefits of autoguidance systems include 
reducing driver fatigue, extending operational hours, and potentially lowering fuel usage.  
A common industry practice involves utilizing a navigational satellite service to derive 
vehicle location in real time, such as the United States? Global Positioning System (GPS), 
or the more inclusive Global Navigation Satellite System (GNSS), which includes 
satellites from the United States (GPS), Russia (GLONASS), and other governments as 
more compatible satellite vehicles are launched into space and positioned into orbits.  
Unlike many geo-stationary satellites used for communications purposes, GNSS satellites 
are in constant motion relative to Earth?s surface; therefore, many satellite vehicles can 
come in to and drop out of view of a user?s GPS/GNSS receiver over the course of 
several minutes.  A GPS/GNSS receiver, or sensor, operates by ?receiving? radio signals 
from GNSS satellites overhead, and sensor software calculates the distance from each 
satellite to the user; of course, because of the movement of the GNSS constellation, 
position solutions fluctuate. 
64 
 
 Although an inexpensive personal navigation device (PND), equipped with a 
GPS/GNSS sensor, can provide the user with his or her location over the majority of the 
planet, the accuracy of such a location could be several meters from the true geographical 
location.  Several conditions relating to the quality of the PND, the amount of signal 
interference, and even local environmental conditions may accumulate this error.  
However, several of these conditions can be managed to a large extent allowing the error 
to be mitigated.  Several manufacturers offer ?correction signals? that serve to enhance 
the real-time positioning of GPS/GNSS sensors; autoguidance systems almost 
exclusively require some form of correction signal to operate at a performance level 
acceptable to most users.  Correction signals do not necessarily provide consistent, 
repeatable results for mitigating error, however; Han et al. (2004) observed that, when 
comparing multiple GPS sensors operating on different wide-area (non-RTK) correction 
signals, a sensor?s dynamic performance was extremely variable throughout time, 
whether by different hours of the test day or by different test days.    
 Accuracy and precision measurements of GPS- or GNSS-capable sensors are 
derived from static-operation tests using internationally-accepted engineering testing 
standards to estimate such determinations (Institute of Navigation, 1997).  Static-
accuracy terms such as Circular Error Probable (CEP), Spherical Error Probable (SEP), 
and Root Mean Square (RMS) Error pertain to differences in measured positions from a 
known target position (Upadhyaya et al., 2003).  Today, no engineering test standard has 
been developed to quantify GPS/GNSS positioning accuracy under dynamic conditions, 
yet most manufacturers report performance of their system in terms of pass-to-pass error 
(P2P).  Also known as relative accuracy, or short-term error, P2P refers to a guidance 
65 
 
system?s accuracy in determining a new position (point, drive path, etc.) relative to a 
recently-determined position within a short period of time (about 15 minutes).  This 
concept expresses to the operator how closely the current drive path should relate to the 
previous drive path, which is a direct relation to skip and overlap in most tillage and 
sprayer operations. The assumption is that some form of dynamic testing must have taken 
place in order to derive such values, but as there is no accepted test standard in place, 
comparing guidance systems among manufacturers in equal terms becomes a challenge.  
Adamchuk et al. (2007) discussed how several parametric and non-parametric statistical 
values are utilized to describe autoguidance error.  The authors preferred to describe both 
types of values when they reported findings, so that readers may choose for themselves 
which values are more representative of autoguidance system performance.   
 Under dynamic situations (i.e. when the GPS/GNSS receiver is in motion), 
performance assessments are often based on measured positions relative to a target 
?path?.  Several researchers have followed this template for guidance testing (Stombaugh 
et al., 2002; Taylor et al., 2004).  These methods have their advantages in that most of the 
measurements generated through these processes can be easily evaluated and compared 
with an intuitive path template.  Also, some researchers prefer to evaluate dynamic 
autoguidance tests in field conditions at suitable operating speeds, as several equipment- 
and terrain-related variables can influence the performance of a system (Ehsani et al., 
2002).  A commonly-used response variable for assessing guidance error is Cross-Track 
Error (XTE), which refers to the shortest possible distance between an observed receiver 
position and its theoretical, or anticipated, position in relation to a mathematical line 
model.  With this variable, other statistical measures of location and variability can be 
66 
 
computed, such as mean XTE, standard deviation of XTE (SD), root mean square error of 
XTE (RMSE), P2P, and year-to-year accuracy (Y2Y).  (Definitions for these terms are in 
Appendix K.) 
 By utilizing high-quality GPS- or GNSS-capable sensors with conventional 
surveying techniques, positions can be derived to a high order of accuracy under static 
conditions; when these techniques are not available to autoguidance operators, an element 
of drift may be introduced into the positioning error.  This position drift is largely a direct 
result of fluctuating position solutions caused by the motions of the GNSS constellation.  
Position drift can have an effect on field operations that rely heavily on autoguidance 
technologies, such as seasonal crop-seed placement over underground irrigation tubing.  
No accepted method of quantifying position drift in dynamic situations exists; however, it 
is generally accepted by researchers that this phenomenon can be measured in terms of 
Y2Y and/or long-term error by comparing the amount of XTE variability over a long 
period of time, providing some measure of how ?repeatable? a guidance system would be 
in traversing a known path from season to season.  Han et al. (2004) and Ehsani et al. 
(2002) utilized an array of vehicle-mounted GPS sensors receiving multiple decimeter- 
and sub-meter wide-area differential GPS (DGPS) correction signals in an effort to 
measure their performance; for these studies, the paths mapped via Real-Time Kinematic 
(RTK) correction were considered the ?absolute? or ?reference? paths from which to 
compare the other receivers? paths.  Autoguidance operations conducted with RTK 
correction are often regarded as sub-inch accurate from one season to the next due to the 
operating parameters and assumptions maintained in the RTK correction.  However, 
when one GPS receiver is established as an independent ?base? from which to generate 
67 
 
correction data or position errors, and few or no redundancies exist to support the 
position derivations made in the autoguidance software, the potential exists for GPS drift 
to occur under RTK operation, albeit a small effect or negligible for some users in 
agricultural field operations.  Because of this error potential, another form of absolute 
positioning was investigated by utilizing a total station, or tachymeter.  In the surveying 
industry, a total-station unit offers target positioning that can be correlated with GPS-
derived positions and, more commonly, physical on-site monuments providing a local 
frame of reference.  While surveyors generally collect positions with a stationary target 
(like a prism), in recent years, total-station technology has progressed such that the target 
can be in motion as well.  This characteristic provides the potential to plot the driving 
path of a vehicle or any moving target in which the prism is mounted. 
4.3 SUB-OBJECTIVE 
 The objective of this study was to compare short-term and long-term accuracies of 
an autoguidance system when using different GPS-based correction services. 
4.4 MATERIALS AND METHODS 
 The tractor used for this study was a John Deere? 6420 (Deere and Co., Moline, 
IL, USA; see Appendix A) factory-equipped with the Starfire? ITC? GPS receiver (part 
no. PF80732; software ver. 2.52A; Deere and Co., Moline, IL, USA) and the GreenStar? 
autoguidance system (hardware ver. PF80444 2.5; software ver. PF303192F; Deere and 
Co., Moline, IL, USA: see Appendix D).  The tractor was also outfitted with a Trimble? 
EZ-Guide? lightbar guidance computer (firmware ver. 4.10.002 Feb 22 2007 11:53:46 
for 2007 tests; Trimble Agriculture, Sunnyvale, CA, USA) and an EZ-Steer? assisted-
68 
 
steering system (software ver. 3.00.01 for 2007 tests; Trimble Agriculture, Sunnyvale, 
CA, USA: see Appendix D). 
 In order to capture moving-vehicle position data with a terrestrial frame of 
reference, we used a Trimble? 5603 DR+ optical total station (firmware ver. 696-03.08; 
Trimble Engineering, Sunnyvale, CA, USA) in combination with a Trimble? TSCe field 
data collector (part no. 45185-20; Trimble Engineering, Sunnyvale, CA, USA) utilizing 
Trimble? Survey Controller software (versions 11.20 ? 11.40; Trimble Engineering, 
Sunnyvale, CA, USA: see Appendix B).  Point of measurement on the tractor was 
maintained using a Trimble? RMT ATS Multi-Channel tracking prism (part no. 
571233035; Trimble Engineering, Sunnyvale, CA: see appendix C), mounted to the rear 
frame hitch approx. 66 cm above the ground and aligned with the vehicle?s centerline.  
Distance measurements were calculated via Electronic Distance Measurement (EDM) 
technique; the stated accuracy of such measurements was ? (5 mm + 2 ppm; Trimble 
Engineering, 2001).  In addition, the total station was equipped with Trimble?s 
AutoLock? function such that the total station, via servo motors integrated within its 
base, could automatically track the moving target prism under dynamic conditions and 
simultaneously collect vector (angle and distance) data. 
 Four different GPS-based correction signals were evaluated: Wide-Area 
Augmentation System (WAAS), John Deere StarFire 1 (SF1), John Deere StarFire 2 
(SF2), and Real-Time Kinematic (RTK).  Position accuracies typically reported by 
manufacturers and/or system users are referenced in Table 4.1.   No implements were 
attached during the tests, so draft or drag was not considered; also, the tractor had 
69 
 
performed only light-duty tasks during its life, so steering and/or transmission wear was 
considered negligible. 
Table 4.1.  Typical reported accuracies for correction signals used in this study. 
Correction Signal Type of Correction Accuracy 
WAAS wide-area 
?15 to 30 cm short-term 
(pass-to-pass); 
?61 to 66 cm long-term 
(static) 
SF1 wide-area ?33 cm short-term; ?76 cm long-term 
SF2 wide-area ?10 cm short-term; ?25 cm long-term 
RTK local-area ?3 cm short-term; ?3 cm long-term 
Source: Hest, 2007.  For additional details, see Appendix E. 
4.4.1 TEST-SITE PREPARATIONS 
 This study was performed in Auburn, Alabama (32.6? N, 85.5? W) beginning in 
February 2006 and ending in August 2007.  A different testing site was selected each year 
(see Figures 4.1, 4.2, and 4.3).  Except for differences in local terrain, both sites shared 
the characteristics of minimal overhead obstruction (desired for maximum reception of 
available satellites), minimal sources of radio interference (desired for maximum 
reliability of positioning data and correction-service reception), and allowances for line-
of-sight testing (desired for uninterrupted prism tracking by the total station).  For the 
2006 study, guidance lines were established at 6.1-m spacing; direction of travel was 
approximately 80.1 degrees azimuth relative to true north (Figure 4.1).  For the 2007 
study, a swath width of 12.2-m was used during this testing; directions of vehicle travel 
were approximately 179.4 degrees (Figure 4.2) and 271.3 degrees (Figure 4.3) relative to 
true north. 
70 
 
 
Figure 4.1. Guidance testing site one, used in 2006.  Elevation contours are in 2-ft 
(0.6-m) intervals; guidance lines are spaced 20 ft (6.1 m).  Direction of vehicle travel 
was westerly to easterly (approx. 80.1 degrees azimuth from true north; aerial 
imagery and contour data courtesy of City of Auburn GIS Department, Auburn, 
AL). 
 
Figure 4.2.  Guidance testing site two, used in 2007.  Elevation contours are in 2-ft 
(0.6-m) intervals; guidance lines are spaced 40 ft (12.2 m).  Direction of vehicle 
travel was northerly to southerly (approx. 179.4 degrees azimuth from true north; 
aerial imagery and contour data courtesy of City of Auburn GIS Department, 
Auburn, AL). 
71 
 
 
 
Figure 4.3. Guidance testing site two, used in 2007.  Elevation contours are in 2-ft 
(0.6-m) intervals; guidance lines are spaced 40 ft (12.2 m).  Direction of vehicle 
travel was easterly to westerly (approx. 271.3 degrees azimuth from true north; 
aerial imagery and contour data courtesy of City of Auburn GIS Department, 
Auburn, AL). 
 Five straight-line parallel paths were navigated by the guidance system(s) for each 
correction signal being evaluated.  Vehicle speed was maintained at 2.2 m s-1 for all tests.  
A test ?event? was represented by traversing all five reference lines within a 15-minute 
time period.    A different correction signal was enabled prior to each test event.  Field 
set-up and data analysis procedures for each year of testing are discussed in Chapter 3 
and Appendices L and M.  XTE values were derived from positioning data and rounded 
to the nearest centimeter.  Testing event schedules are presented in Appendix I (2006) 
and Appendix J (2007). 
 
 
72 
 
4.4.2 STATISTICAL ANALYSES 
 For both years of the study, mean cross track error (XTE), standard deviation of 
XTE (SD), and root mean square error of XTE (RMSE), were calculated as outlined by 
Ehsani et al. (2002; also outlined in Appendix K).  Pass-to-pass errors (P2P) were 
calculated by subtracting the signed value of the mean XTE of a row from the preceding 
row; year-to-year errors (Y2Y) were calculated as the range between the signed testing-
event XTE across a season.  Unsigned short-term error was calculated as the absolute 
difference from average XTE per testing event (Adamchuk et al., 2007); unsigned long-
term error was considered the same as unsigned XTE.  Individual testing events and Y2Y 
observations were screened for potential XTE outliers, defined by ?any observation 
farther than 1.5 [Inter-Quartile Range, or IQR] from the closest quartile? (Devore and 
Farnum, 2005). Basic statistical measures of XTE were derived through the univariate-
statistics procedure (PROC UNIVARIATE; SAS Institute, Cary, NC, USA).  Box-and-
whisker plots were created to illustrate signed XTE comparisons of individual testing 
events across a given season as well as year-to-year XTE assessment.  Cumulative 
frequency distributions of short-term and long-term error were created to illustrate 
unsigned XTE trends for each correction signal.  Graphical depictions were generated in 
MS Excel 2007. 
4.5 RESULTS AND DISCUSSION 
4.5.1 FIRST-YEAR EVALUATIONS OF XTE 
 As mentioned in Chapter 3, XTE evaluations under the RTK correction are 
offered solely as comparisons to the other correction signals, since the total station used 
73 
 
in this study was not considered accurate enough to evaluate dynamic RTK situations.
 Cumulative short-term and long-term errors for the 2006 season are presented in 
Figure 4.4.  The values at which the 95%-probability lines intersect each error curve are 
presented as 95% short-term and long-term error in Table 4.2.  This method of presenting 
  
Figure 4.4.  Cumulative distributions of unsigned short-term and long-term error 
(cm) for (a) RTK, (b) SF1, (c) SF2, and (d) WAAS corrections in the 2006 study. 
 
unsigned short-term error is similar to the method discussed by Adamchuk et al. (2007), 
in that short-term error was considered the amount of path deviation witnessed from the 
average XTE value of the testing event on which the deviation was recorded.  For 
74 
 
example, if an individual XTE value of 12 cm from the target path was observed during a 
testing event in which all XTE averaged 10 cm from the target path, then the short-term 
error for the individual XTE value was 2 cm.  However, Adamchuk et al. (2007) used a 
different calculation for deriving long-term error that depended on finding the amount of 
Table 4.2.  Summary of 2006 XTE data. 
Correction 
Signal 
Median 
XTE 
(cm) 
XTE 
IQR 
(cm) 
Mean 
XTE 
(cm) 
SD 
(cm) 
RMSE 
(cm) 
95% 
Short-
term error 
(unsigned), 
cm 
95% 
Long-term 
error 
(unsigned), 
cm 
RTK -5 6 -5 5 7 7 13 
SF1 5 116 38 68 78 32 159 
SF2 37 29 36 21 41 22 68 
WAAS -9 37 -10 27 29 36 58 
 
path deviation witnessed from the average XTE value of the entire testing season.  To 
extend the previous example, if all XTE values for the entire season averaged 5 cm, then 
the individual XTE value would have a long-term error of 7 cm.  This technique of 
evaluating long-term error was changed to reflect simple unsigned XTE across the testing 
season.  Firstly, it was apparent that the locations of the GPS-derived drive paths 
changed, or ?drifted?, across the season.  It was surmised that, by adjusting individual 
XTE relative to a seasonal average, this ?drift? would not be as readily recognized in the 
datasets.  Secondly, we wanted to use values more reflective of what a vehicle operator 
may witness over time in a typical autoguidance operation, and individual XTE was 
suitable.  However, the 2006 study was considered as more of a procedural evaluation for 
data collection, and the lessons learned during these tests helped us to refine our 
75 
 
procedures and expand upon other concepts and variables that may have helped to better 
quantify short- and long-term error. 
 RTK correction offered very low position error in terms of variability (5-cm SD, 
6-cm IQR), location (5-cm XTE median/XTE mean), and overall accuracy (7-cm 
RMSE), but the values were slightly higher than values typically reported for RTK 
performance (Table 4.2).  Short-term and long-term errors were also greater than 
anticipated.  SF1 correction varied over a meter throughout the season, yet tended to be 
centered close (5 cm) to the desired path; however, high RMSE suggested that this signal 
was the most inaccurate signal of the season, and this inaccuracy contributed to a high 
long-term error.  SF2 correction offered XTE farthest from the reference lines (37-cm 
XTE median; 35-cm XTE mean).  Finally, WAAS correction offered good general 
location throughout the season (9-cm XTE median/10-cm XTE mean) despite the overall 
accuracy (29-cm RMSE), which was consistent with typical reported accuracy for this 
signal (Table 4.1). 
 Box-and-whisker plots for all testing events in 2006 are presented in Figure 4.5; 
values pertaining to creation of box-and-whisker plots are in Appendix N.  The 
?summary? box-and-whisker plot for each correction signal corresponds with median and 
IQR values presented in Table 4.2.  As mentioned above, positioning for SF1 correction 
signal had the greatest error of all other signals tested; however, XTE variability between 
individual testing events was somewhat consistent (Fig. 4.5b). XTE median varied 
considerably on events J71 and J72 from the previous three events.  On these occasions, 
XTE suggested that SF1 is consistent in a short-term sense but not very repeatable in the 
long-term sense.  Only one individual testing event under SF2 correction was considered 
76 
 
more ?on target? than other events (D62; Fig. 4.5c); apart from that event, XTE median 
was somewhat consistent at 37 cm throughout the period tested.  WAAS-derived XTE 
had consistent variability throughout the period tested, as represented by individual IQR 
per testing event. 
 
Figure 4.5.  Box-and-whisker plots for individual testing events and Y2Y (2006) for 
(a) RTK, (b) SF1, (c) SF2, and (d) WAAS correction signals.  Testing Event Label 
corresponds to testing events listed in Appendix I. 
4.5.2 SECOND-YEAR EVALUATIONS OF XTE 
 Cumulative short-term and long-term errors for the 2007 season are presented in 
Figure 4.6.  The values at which the 95%-probability lines intersect each error curve are 
presented as 95% short-term and long-term error in Table 4.3.  RTK correction provided 
77 
 
close short-term and long-term error in the southward direction, while the westward 
short-term and long-term errors were not similar (Figure 4.6a). Short-term and long-term 
 
Figure 4.6.  Cumulative distributions of unsigned short-term and long-term error 
(cm) for (a) RTK, (b) SF1, (c) SF2, and (d) WAAS corrections in the 2007 study. 
 
errors were considerably different for the SF1 (Figure 4.6b) and WAAS (Figure 4.6d) 
corrections and for each drive-path orientation.  Westward SF2 provided closer short-
term and long-term error than southward SF2 (Figure 4.6c).  Both SF1 and SF2 had 
similar maximum 95% long-term error of around 70 cm, suggesting that both John Deere 
correction signals performed equally.  
78 
 
 As evidenced from RMSE values, the drive-path orientation tended to influence 
XTE (Table 4.3).  Southward drive paths yielded more accurate XTE than westward 
drive paths for 3 of the 4 correction signals tested.  RTK short-term and long-term errors 
were much larger than values typically reported.  Median XTE and mean XTE values 
suggested, however, that the target path was closely obtained throughout the season.  
Given that variability of SF2 signal was about equal (43-cm southward, 39-cm westward) 
and that XTE median values of the SF2 signal were closer to zero (expected value) than 
with either SF1 or WAAS, the suggestion was that the SF2 correction was the most 
reliable sub-meter correction signal evaluated.  (Measure of variability for SF2 in 2006 
supported this suggestion as well---Table 4.2.)  Short-term and long-term errors observed 
for SF2 did not support this suggestion, however;  
Table 4.3.  Summary of 2007 XTE data. 
Correction 
Signal a 
XTE 
Median 
(cm) 
XTE 
IQR  
(cm) 
XTE 
Mean 
(cm) 
SD 
(cm) 
RMSE 
(cm) 
95% 
Short-
term error 
(unsigned), 
cm 
95% 
Long-term 
error 
(unsigned), 
cm 
RTKS -1 10 -2 7 7 9 15 
RTKW 4 18 4 13 14 17 27 
SF1S 21 83 12 47 48 24 81 
SF1W 18 36 18 31 36 26 70 
SF2S 3 43 5 29 29 24 56 
SF2W 7 39 6 35 35 57 72 
WAASS -10 49 -8 43 43 35 92 
WAASW 44 66 38 59 70 55 130 
a Subscripts ?S? and ?W? refer to southward-oriented and westward-oriented reference lines, respectively. 
 SF1 offered consistent position error regardless of path orientation (21-cm XTE 
median southward/18-cm XTE median westward; Table 4.3), but the variability in the 
southward orientation was more than twice that observed in the westward orientation; a 
79 
 
histogram of XTE for the southward SF1 dataset revealed that, instead of XTE being 
normally distributed about a single mean XTE, there were two distinct ?peaks? in the 
distribution at approximately 60 cm and -50 cm locations.  Signed statistics are likely 
confounded as a result, but unsigned short-term and long-term errors for SF1 are 
consistent regardless of path orientation.  SF1 short-term and long-term errors agree 
closely with typical reported accuracy of this signal as well (Table 4.1). 
 Long-term XTE error from WAAS correction was approximately one meter 
regardless of direction (Table 4.3).  To determine whether WAAS correction performed 
according to typical reports for this signal, some interpretation of Table 4.1 must be 
considered.  Typical WAAS errors are ?15 to 30 cm short-term and ?61 to 66 cm long-
term (Table 4.1); if ?15 to 30 cm for short-term error means that the position could be 
either 15 to 30 cm on one side of the target line or the other, then the error ?spread? could 
be considered 30 to 60 cm.  Likewise, long-term WAAS error could be considered as a 
spread of 122 to 132 cm.  Unsigned error represents this error spread, therefore, WAAS 
short-term and long-term errors agree with those typically reported by manufacturers and 
researchers regardless of travel direction. 
 Box-and-whisker plots for all testing events in 2007 are presented in Figures 4.7 ? 
4.10; values pertaining to creation of box-and-whisker plots are in Appendix O.  The 
?Summary? box-and-whisker plot for each category corresponds with median and IQR 
values presented in Table 4.3.  For each drive-path orientation, XTE spread under RTK 
correction appeared similar across the season (Figure 4.7); however, westward RTK 
events exhibited more variability and less accuracy than southward events.  Presentation 
of southward RTK performance over the season provides strong support that RTK 
80 
 
correction had the greatest accuracy and reliability from one event to the next, yet 
westward RTK performance tends to counter-act this notion. 
 
 
Figure 4.7.  XTE box-and-whisker plots for RTK correction signal in 2007, 
comparing southward orientation (left) and westward orientation (right).  Testing 
Event Label corresponds to testing events listed in Appendix J. 
 
 
 
Figure 4.8.  XTE box-and-whisker plots for SF1 correction signal in 2007, 
comparing southward orientation (left) and westward orientation (right).  Testing 
Event Label corresponds to testing events listed in Appendix J. 
81 
 
 
 
Figure 4.9.  XTE box-and-whisker plots for SF2 correction signal in 2007, 
comparing southward orientation (left) and westward orientation (right).  Testing 
Event Label corresponds to testing events listed in Appendix J. 
 
 
 
 
Figure 4.10.  XTE box-and-whisker plots for WAAS correction signal in 2007, 
comparing southward orientation (left) and westward orientation (right).  Testing 
Event Label corresponds to testing events listed in Appendix J. 
82 
 
 By studying Figures 4.8 ? 4.10, there was the tendency for testing-event XTE 
median to fluctuate throughout time in an almost-regular sine-wave fashion.  SF1 and 
SF2 illustrated this phenomenon more strongly than WAAS, but this trend was witnessed 
for WAAS as well.  No explanation can be offered to explain this behavior; however, 
more investigation should be considered to better document this trend and to determine 
whether a more precise method of evaluating short-term or long-term error could be 
derived. 
 Concerning the WAAS correction (Figure 4.10), XTE for testing events that 
occurred on or after July 31, 2007 (events W*8*, W*9*, and W*10*) may have been 
influenced by the scheduled removal of one of the two available communications 
satellites that broadcasted the WAAS correction signal.  The GreenStar guidance system, 
that relayed WAAS correction from the StarFire iTC receiver to the EZ-Guide system, 
could not be configured or reviewed to determine whether a particular communications 
satellite was being monitored, or whether the system received the correction without 
regard to its source.  If this incident provided an additional source of error, then per-event 
XTE for the southward drive path reflected this concept, starting at event WS81.  While 
this incident may have affected per-event XTE for the westward drive paths, starting at 
event WW81, assessment was difficult considering the large amount of per-event XTE 
variability witnessed previously in the season.        
4.5.3 SIGNED P2P VERSUS UNSIGNED SHORT-TERM ERROR EVALUATIONS 
 In 2006, RTK correction resulted in no skip and only 2-cm overlap (Table 4.4).   
Under SF2, signed P2P averaged 2 cm for the study, but separate events witnessed either 
a skip as great as 12 cm or an overlap as great as 8 cm.  Unsigned short-term error for 
83 
 
SF2 was 22 cm (Table 4.2), and both values were consistent with typical reports under 
this correction (Table 4.1).  SF1 correction exhibited greatest amount of overlap 
throughout the study, and signed P2P under WAAS correction varied greatly. 
Table 4.4. Signed pass-to-pass error (P2P, cm) throughout the 2006 study. 
Event No.[a] RTK SF1 SF2 WAAS 
1 -2   7 
2 -2  5 -1 
3 0  1 -5 
4 -1  4 -35 
5 -1  0 -2 
6 -2  -2 3 
7 -1  1 4 
8 -2 -2 5  
9 -1 -8 2  
10 -1  -1  
11 0 0 -8  
12 0 -5 12  
13 -1 -15 11  
     
Avg. P2P -1 -6 2 -4 
[a]   Events are numbered in sequential order by time and date of collection (see Appendix I). 
 
 In 2007, southward P2P under RTK correction ranged from neutral to 8-cm 
overlap, and westward P2P ranged from 2-cm skip to 2-cm overlap throughout the study 
(Table 4.5).  SF2 correction yielded skip and overlap as high as 8 cm traveling 
southward, while skip was as high as 39 cm and overlap was as high as 26 cm traveling 
westward.  Under SF1, both path orientations averaged neutral P2P.  WAAS correction 
yielded overlap in the southward direction and skip in the westward direction; southward 
P2P illustrated an overlap as high as 19 cm, and westward P2P illustrated overlap and 
skip nearing a half meter.  To compare the signed short-term error to the unsigned short- 
 
84 
 
Table 4.5. Signed pass-to-pass error (P2P, cm) throughout the 2007 study. 
Test Date[a] RTKS[b] SF1S SF2S WAASS RTKW[c] SF1W SF2W WAASW 
March 7 -2 6 -2 4  0 -8 14 
      -10 -25 -9 
         
March 14 -2 9 4 -4 -2 4 14 -1 
 -1 -17 2 -13 -1 -4 39 8 
         
April 13 -1 7 8 -13 0 -3 -5 25 
 -2 7 1 -4 -1 8 0 16 
         
April 30 -2 -4 -6 -2   -4  
         
June 5 -2 -4 3 -4 2 4 -2 -8 
 -3 -1 -5 -6  0 -12 58 
         
July 10 0 -6 2 -5 0 -11 -13 -7 
  4 6 -6  -3 -6 3 
         
July 18 -2 4 3 0 -2 -1 -1 -10 
  -6 -3 -9  2 6 5 
  -2 2 6  -3 -1 -4 
         
July 31 -2 -3 1 2 -2 -2 3 9 
 -1 1 4 -8 -1 6 -2 22 
 0 0 -2 2     
         
August 3 -2 9 -8 1 -1 5 -26 3 
         
August 23 -2 -1 -8 4 0 3 -24 12 
 -8 -5 -6 -1 -2 -6 -1 4 
 0 -2 3 -19 -1 1 2 -42 
         
Avg P2P -2 0 0 -4 -1 0 -3 5 
[a]   Events are numbered in sequential order by time and date of collection (see Appendix J). 
[b]   Correction signals labeled with ?S? refer to tests conducted while driving southward. 
[c]   Correction signals labeled with ?W? refer to tests conducted while driving westward. 
 
85 
 
term error documented in Table 4.3, southward RTK, westward SF2, and either SF1 
orientation had similar values. 
 Neither signed nor unsigned short-term error as calculated in this analysis can be 
sufficiently compared to one another.  Average P2P, as presented in Tables 4.4 and 4.5, 
do not adequately represent the situations witnessed during autoguidance trials, although 
the range of signed P2P is somewhat representative.  Most researchers consider P2P and 
short-term error as similar in the sense that both testing methods typically require a 
limited timeframe for data collection, such as 15 minutes, however. 
4.5.4 SIGNED VERSUS UNSIGNED YEAR-TO-YEAR ERROR (Y2Y) EVALUATIONS 
 Year-to-year errors are presented in Table 4.6.  When comparing these values to 
95% long-term error in the same categories for the 2006 study, some similarities were 
observed for the SF1, SF2, and WAAS corrections.  For the 2007 study, Y2Y was not 
similar to any of the long-term error values derived.  The unsigned long-term error value 
may be a more representative method to quantify guidance drift.   
 
Table 4.6.  Year-to-year error (Y2Y, cm) from the 2006 and 2007 studies. 
Correction Signal 
2006 Y2Y 
(cm) Correction Signal a 
2007 Y2Y 
(cm) 
RTK 5 RTKS 12 
  RTKW 34 
SF1 163 SF1S 151 
  SF1W 113 
SF2 78 SF2S 91 
  SF2W 204 
WAAS 49 WAASS 316 
  WAASW 261 
a Subscripts ?S? and ?W? refer to southward-oriented and westward-oriented reference lines, respectively 
(2007 data). 
86 
 
 
4.5.5 CONSIDERATIONS FOR CONTINUING RESEARCH 
 Further research would involve evaluating other factors that may influence short-
term and long-term error, such as vehicle ground speed.  Factors that have been known to 
affect general position accuracy, such as GPS/GNSS satellite geometry or satellite count, 
should be collected for future evaluations in order to more consistently compare 
experimental conditions from one test to another.  An ambitious research prospect would 
be to conduct time-series analyses for dynamic XTE in an effort to better understand if 
there is a cyclical nature to correction-signal performance as was suggested in the 2007 
study.   
4.6 SUMMARY 
 An autoguidance-assisted vehicle was tracked with an optical total station in order 
to measure short-term and long-term error under four correction signals.  Cross-track 
error was influenced by drive-path orientation (North-South vs. East-West) for RTK, 
SF2, and WAAS correction signals; SF1 performance was consistent regardless of 
orientation.  Southward XTE was typically more accurate than westward XTE.  RTK and 
SF2 short-term and long-term errors were much greater than typically reported values by 
manufacturers.  XTE derived under SF1 and WAAS correction were more representative 
of typical manufacturer?s claims.  Unsigned 95% short-term and unsigned 95% long-term 
error was preferred to describe guidance performance over signed P2P and Y2Y, 
respectively.
87 
 
CHAPTER FIVE 
EVALUATING APPLICATION UNIFORMITY OF AN IN-SITU PRESSURE-
COMPENSATING SUBSURFACE-DRIP IRRIGATION PRODUCT 
 
5.1 ABSTRACT 
 In the southeast U.S., an available agricultural irrigation practice is Subsurface 
Drip Irrigation (SDI).  Research performed at the Tennessee Valley Research and 
Extension Center (TVREC), Belle Mina, AL indicated moisture variability after irrigation 
in an active experimental plot equipped with pressure-compensated SDI.  The need 
therefore exists to quantify the uniformity of the SDI product without disrupting the 
ongoing crop research.  A study was conducted to develop and validate an in-situ testing 
technique to assess application uniformity of SDI products, and to evaluate if variable 
operating pressure or terrain slope impacted application uniformity.  Laboratory 
investigations indicated that new pressure compensated SDI tape of the type used at 
TVREC had an emitter discharge of 1.04 L h-1 at 5% coefficient of variability (CV), 
above the manufacturer?s specification of 0.98 L h-1 at 2.5% CV.  A field testing 
apparatus was fabricated to interface with the in-situ SDI product in order to deliver 
water over a range of operating pressures while measuring sample discharge. This 
apparatus was tested under laboratory conditions to compare emitter-discharge estimates 
88 
 
with per-emitter discharge from drip-tape samples; apparatus accumulated measurements 
agreed to within a half-percent of per-emitter discharge rates. Global Positioning System 
(GPS) and Geographic Information System (GIS) technologies were utilized to quantify 
surface slope conditions and to target SDI samples throughout the field test plot based on 
irrigation system use frequency and terrain slope.   Drip-tape discharge was affected by 
an interaction between operating pressure and system use frequency (p=<0.0001).   
Lowest discharge was observed for the irrigated SDI product operating at 48 kPa 
pressure.  Discharges from non-irrigated SDI product and from irrigated product, both 
operated at 83 and 117 kPa, were not statistically different (?=0.05) from one another.  
Emitter discharge estimates were within ?5% of the baseline discharge rate determined 
from laboratory evaluations (1.04 L h-1) regardless of operating pressure or system use 
frequency, except for irrigated samples operating at 48 kPa (-7%).  Certain tested samples 
exhibited flows exceeding ?5% of the baseline discharge that may prompt a system 
manager to investigate situations more closely.  Spatial yield estimates suggested that 
cotton lint yield was adversely affected by poor SDI uniformity.  Techniques and 
equipment developed in this study provide SDI researchers and users a prototype field 
method to evaluate in-situ SDI performance. 
89 
 
5.2 INTRODUCTION 
 Subsurface drip irrigation (SDI) is a method of applying precise amounts of water 
to agronomic crops by delivering the water directly to the plants? root zone via 
underground tape or tubing, often referred to as ?drip line?, ?drip tape?, or ?drip tube?.  
Previous studies have focused on utilizing SDI for fertigation or effluent release (Lamm 
et al., 2001; Lamm et al., 2004; Song et al., 2003; Trooien et al., 2000), determining 
optimum lateral spacing and lateral depth for various crop systems (Enisco-Medina et al., 
2002; Bryla et al., 2003; Charlesworth and Muirhead, 2003; Camp et al., 1997), and 
studying SDI effects on crop water-use efficiency (Bordovsky and Porter, 2003; Lamm 
and Trooien, 2003).  While much research has focused on improved agronomic growth, 
relatively few studies discuss the longevity of drip tape systems.  Ayars et al. (1999) 
operated several SDI experimentation plots throughout California over a fifteen-year 
period, testing fertigation regimes against crops such as cantaloupe, corn, cotton, and 
tomato.  The authors reported that some of their plots performed similarly to initial 
installation after nine years, and when the system was flushed and maintained in a proper 
manner, root intrusion and overall emitter clogging was minimized.  The case was 
presented by O?Brien et al. (1998) that longevity of an SDI system became an economic 
factor; when compared to center-pivot irrigation the implementation of the SDI system 
was not deemed profitable if the system did not function for at least ten years. 
 Much research has focused on evaluating emitter discharge uniformity and 
efficiency.  One often-used testing standard pertaining to SDI performance is ASAE 
S553 (2008).  This standard outlines methods for quantifying performance of drip tape 
90 
 
products, and includes testing criteria for selecting sample populations and evaluating 
emitter discharge rates at constant pressure or by pressure response.  Testing criteria 
documented in this standard are most favorable to laboratory conditions; specifically, the 
tape product should be above ground in a controlled environment as opposed to in-situ 
field conditions.  Several researchers have adapted the criteria from this testing standard 
to evaluate drip tape products under field conditions.  When Lesikar et al. (2004) 
followed the ASAE standard to evaluate drip-tape flow rates from tape products that had 
been used for wastewater effluent for up to five years, the authors were required to 
excavate the soil to such an extent that small catch pans could be placed underneath the 
emitters in order to perform in-field application uniformity tests.  The authors did not 
refer to any plants of agronomic value that were removed or relocated during excavation.   
However, in an effort to protect neighboring plants, Steele et al. (1996) exposed a 
reduced number of emitters in order to make an estimate of in-field application 
uniformity.  Catch pan volumes were averaged and extrapolated to an equivalent flow 
rate based on manufacturer?s flow rates.  A similar practice of placing catch pans under 
exposed emitters was employed by Safi et al. (2007) to evaluate emitter-discharge 
uniformity between new and 3-year used drip-tape product across a range of operating 
pressures; the authors also relied upon evaluation procedures and terminology 
documented in ASAE EP458 (1996) and ASAE EP405 (2005). 
 In recent years, manufacturers have developed SDI products that are ?pressure 
compensated,? meaning that regardless of pressure changes caused by system surges or 
rolling terrain, emitter discharge rates are consistent.  There is a lack of research literature 
available regarding pressure-compensated SDI performance in agronomic settings, while 
91 
 
some research has been conducted in wastewater or sewage-effluent settings (Li et al., 
2009; Liu and Huang, 2009).  The primary reason for the lack of research may be that 
pressure-compensated SDI products are still relatively new, and manufacturers continue 
to refine designs as products are released into the marketplace.  Additionally, acceptance 
of SDI technology by producers may be relatively slow. 
 At the Tennessee Valley Research and Extension Center (TVREC, Belle Mina, 
Alabama), station-management practices involved the utilization of pressure-
compensated SDI in conjunction with precision agriculture technology to grow cotton 
and other crops.  Research by Sullivan et al. (2007) indicated that moisture variability 
existed within a study site containing plots using pressure compensated SDI tape in ways 
that suggested either clogged emitters or flow rate differences between emitters.  In 
theory, pressure compensated SDI emitters should maintain equivalent flow rate 
regardless of pressure differences over a specified range.  The surface of the TVREC 
study site exhibits rolling terrain (0% to 5% slope range) that may influence water 
distribution from one end of the lateral to the other if the pressure-compensation feature 
was inadequate or had deteriorated over time.  The tape product selected in this study had 
been operational for six years and had not been evaluated for specific in-field 
performance; however, as a long-term experiment was being conducted on this test plot, 
major disruption through excavation was not a preferred option.  From a research 
perspective, the need existed to quantify the in-situ performance of selected portions of 
drip tape coinciding with locations identified as having undesirable moisture distribution.  
A scenario such as this could provide either preliminary diagnostic information for SDI 
92 
 
system-management or independent data for moisture variability and precision-
agriculture related crop yield studies. 
5.3 SUB-OBJECTIVES 
 The objectives of this investigation were to (1) develop and validate in-situ testing 
techniques to assess application uniformity of SDI products, and (2) evaluate whether 
variable operating pressure or terrain slope impacted application uniformity. 
5.4 MATERIALS AND METHODS 
5.4.1 LABORATORY INVESTIGATIONS 
 To determine a feasible method to assess application uniformity of the SDI 
product at the TVREC, we conducted laboratory evaluations of new drip tape of identical 
design.  The SDI product evaluated was Netafim? DripNET PC?, 0.381 mm tape 
thickness, 2.2 cm I.D.  For this product, each emitter was rated by the manufacturer to 
distribute 0.984 liter hr-1 at 0.6-m spacing (Netafim USA, 2010; see Appendix P).  
Adapting from ASAE S553 (2008), a portion of drip tape containing 20 contiguous 
emitters was selected and subjected to specified performance tests.  A tape sample 
containing 20 emitters was the desired number for in-situ field tests. 
 The municipal water supply provided in the laboratory offered a higher operating 
pressure than was desired; therefore, pressure was reduced and regulated via a system 
bypass (ball valve) and a brass-body pressure regulator (Model 26A; Watts Regulator 
Co., No. Andover, MA, USA), capable of adjusting from 21 to 344 kPa pressure range.  
Operating pressure was monitored throughout the tests via analog pressure gauges 
plumbed on the upstream and downstream ends of the tape sample. Emitter discharge 
93 
 
values were recorded over a period of five minutes by suspending the product over an 
array of catch cups in order to validate manufacturer specifications.  In order to test 
potential slope effects on discharge uniformity, a sample of new drip tape was subjected 
to a constant operating pressure of 117 kPa at  a horizontal (0% slope) grade and a 6% 
slope grade uphill of the water supply.    Then, another sample of drip tape was subjected 
to a range of operating pressures (48, 83, 117, and 138 kPa) while also subjected to a 6% 
slope grade to evaluate potential discharge differences according to pressure. Three 
replications were conducted during this testing.  Per-emitter discharge and coefficients of 
variation (CV) were collected and tabulated using MS Excel 2007.  A studentized t-test 
was conducted  in MS Excel 2007 to determine if a discharge difference existed (?=0.05) 
between slope classes, and a General Linear Model (GLM) procedure was conducted 
using SAS ver. 9.1.3 to determine if discharge differences existed (?=0.05) between 
pressure classes.  Mean discharge values were separated according to Tukey?s 
Studentized Range (SAS Institute, Cary, NC, USA). 
5.4.2 SDI TESTING APPARATUS 
 Because the premise behind this study was to evaluate application uniformity with 
minimal soil-profile disturbance, we needed to determine whether a discharge value 
could be derived for previously installed SDI tape without having access to per-emitter 
flow rates, thereby avoiding complete excavation of SDI tape to access each emitter 
within the test specimen.  Based on the manufacturer?s specifications, twenty emitters per 
drip-tape sample distributes 19.7 L h-1 under the recommended operating conditions 
(Netafim USA, 2010).  It was therefore necessary to design and construct a portable field 
94 
 
apparatus that would supply water to the drip-tape samples over a suitable range of 
operating pressures; in addition, this apparatus needed to be equipped such that the 
quantity of water distributed through the drip-tape samples could be recorded to within 
?2% of actual discharge (ASAE, 2003).  The result was a trailer-mounted apparatus with 
a 12V, low-flow (10.6 L min-1) centrifugal pump (Model 2088; SHURflo, Cypress, CA, 
USA) to deliver water from a 19-L reservoir to the drip-tape sample (Figure 5.1) and a 
low-flow pressure regulator in-line with the pump discharge (Figure 5.2).  This 
equipment configuration allowed for sustainable operating pressures from 20 - 310 kPa. 
 
 
Figure 5.1. Illustration of the portable, in-field SDI testing apparatus. The white 
vessel atop the bench scale served as the water supply reservoir during testing.  The 
bench scale provided vessel mass, which equated to sample discharge.  The 12-volt 
battery-powered pump was mounted on the side and provided the operating 
pressure for the SDI sample tested. 
 
 
95 
 
 
Figure 5.2. Pressure and flow regulation system for the SDI testing apparatus.  Most 
of the water from the pump (left side) was recirculated to the white reservoir (not 
shown) and remaining water was distributed to the in-situ SDI sample section 
through an adjustable pressure regulator (center right) and the green hose (right 
side).  The gate valve (center left) and the adjustable pressure regulator (center 
right) served as a means to target and sustain the necessary operating pressure.  
 Due to power consumption by the water pump and on-board measurement 
devices, field testing was performed over a relatively short time interval.  Apart from 
initially conditioning the drip line samples at 5- to 15-minute intervals, data collection 
events were limited to 5 minutes per pressure level tested.  Because of this limitation, the 
design criteria for the apparatus required that a method of measuring a small flow rate 
(approx. 1.64 L 5 min-1) be incorporated.  In order to obtain the temporal resolution of 
data collection, we devised a method of measuring the water mass as it was distributed to 
the drip line.  A water vessel was placed atop a 100-kg capacity, 0.02-kg resolution 
digital bench scale (model B100S platform with model CD-11 indicator, software ver. 
1.22; Ohaus Corp., Pine Brook, NJ, USA) during tests (Figure 5.1).  Water mass (kg) was 
converted into a volumetric value (mL) and then into an equivalent sample discharge rate 
per emitter given the number of emitters per unit length of tape tested. 
 Operating pressure had to remain constant during each test, yet pressure settings 
needed to be adjustable from test to test.  Pressure regulation was achieved using a 2-way 
96 
 
brass-body adjustable pressure regulator (Model 26A; Watts Regulator Co., No. Andover, 
MA, USA), capable of adjusting from 21 to 344 kPa pressure range (Figure 5.2).  An 
analog precision pressure gauge (Type 332.54; F.N. Cuthbert, Inc., Toledo, OH, USA) 
was installed immediately after the pressure regulator, the gauge served primarily as a 
quick approximation of operating pressure.  More accurate pressure monitoring was 
achieved through two digital pressure sensors (Model 1501B02EZ50PSIG; PCB 
Piezotronics, Depew, NY, USA) capable of detecting a pressure range from 0 to 344 kPa 
with an accuracy of ?0.15%.  Each pressure sensor was installed at either end of the drip-
line sample via plastic fittings that were compatible for connection to the drip line 
(Figures 5.3a and 5.3b).  Pressure-sensor readings were relayed wirelessly from data 
loggers (Model CR206, PakBus software; Campbell Sci., Logan, UT, USA) to a Topcon 
X20 field computer (Windows XP platform; Topcon Positioning Systems, Inc., 
Livermore, CA, USA) mounted on the apparatus. A real-time sensor-monitoring software 
program (LoggerNet 3.4.1; Campbell Sci., Logan, UT, USA) was installed on the X20 
computer to allow an operator to view instantaneous operating pressure exhibited on the 
drip-tape sample.  (Equipment specifications are listed in Appendix H.) 
97 
 
 
Figure 5.3. Pressure sensors located at the (a) upstream and (b) downstream ends of 
the drip-tape being sampled. 
5.4.3   LABORATORY TESTS FOR SDI TESTING APPARATUS 
 To validate discharge data obtained from the apparatus, laboratory tests were 
performed with new SDI product identical to that in the field plot. Emitter uniformity was 
evaluated for a random sample of 100 contiguous emitters in order to calculate the CV 
under 48, 83, 117, and 138 kPa operating pressures, adapted from clause 8.2 of ASAE 
S553 (2008). Tape-section preparation involved >15 min of conditioning at four 
operating pressures.  Operating pressures were monitored via the aforementioned 
pressure transducers installed upstream and downstream of the drip-tape sample.  The 
drip-tape sample was suspended above the laboratory floor, 152 cm over an array of 100, 
240-mL plastic cups.   Tests were replicated three times at each pressure level.  Discharge 
values were tabulated and compared in MS Excel 2007, and CVs were computed 
according to ASAE S553 (2003).  Differences in discharge values according to pressure 
were analyzed using GLM procedure and separated according to Tukey?s Studentized 
Range (SAS Institute, Cary, NC, USA). 
98 
 
5.4.4 FIELD PREPARATIONS 
 For the field tests, a 3x2 randomized complete block design was developed to 
select potential sampling locations ranked according to field slope classes (0-2%, 2-4%, 
and >4% rise) and irrigation-system use frequency (?irrigated? versus ?non-irrigated? 
treatment) with three replications. Topographic data were collected utilizing a Trimble 
survey-grade, real-time kinematic (RTK) Global Positioning System (GPS) receiver, 
capable of elevation information within 5 cm relative accuracy (Trimble Navigation, 
Sunnyvale, CA, USA).  The GPS receiver was mounted to an ATV, and position data 
were recorded on a 2- to 3-m spacing; position data were then imported into ArcGIS ver. 
9.3 (ESRI, Redlands, CA, USA) for mapping and further analyses.  To classify the test 
plot according to slope, geospatial data were interpolated first into a digital elevation 
model (DEM) via 3D Analyst Tools / Raster Interpolation / Topo to Raster function, and 
then into a slope model via 3D Analyst Tools / Raster Surface / Slope function (Figure 
5.4). 
 
 
99 
 
 
Figure 5.4. Depiction of test plot with respect to slope categories, represented as 
percent rise.  Sampling locations represent 12.2-m drip-line segments blocked 
according to ?irrigated? versus ?non-irrigated? treatment effect and slope category. 
 To clarify the meaning of irrigation-system use frequency, since the experimental 
plot?s inception (fall 2003), a concurrent field experiment involved evaluation of 
treatment effects (?irrigated? versus ?non-irrigated?) on an independently-conducted 
long-term seasonal cotton crop (see Appendix Q).  While the entirety of the plot was 
equipped with SDI tape, only those portions of the system pertaining to the ?irrigated? 
treatment were operated during the growing season according to site-specific pan-
evaporation estimates.  The portions of the system pertaining to the ?non-irrigated? 
treatment variable were only operated during spring maintenance cycles which involved a 
sustained clean-and-flush.  From a concept of SDI longevity, we wanted to test whether 
irrigated portions of the drip-line system would be more fatigued than the non-irrigated 
portions of the system with respect to drip-tape sample discharge. 
100 
 
 The resulting GIS depiction also served as a navigational aid during field testing.  
The map of sampling locations was uploaded into a handheld differential GPS (DGPS) 
mapping device with a nominal sub-meter horizontal accuracy (Trimble Navigation, 
Sunnyvale, CA, USA).  This device allowed for easy navigation to all sampling locations. 
5.4.5 FIELD INVESTIGATIONS 
 Field investigations were conducted in mid-April, 2009, prior to crop planting.  
Approximately two weeks earlier, an annual spring maintenance cycle was performed for 
the SDI system, which consisted of flushing the laterals and running the system 
uninterrupted for several days in order to locate and repair obvious leaks before the 
growing season.  As a result, the SDI system performance was deemed optimum in mid-
April.   At each sample location, a small volume of soil was excavated so as to reveal 
approximately 30 cm of drip line at points along the drip line that were 12.2-m apart; this 
method would result in 20 contiguous emitters being tested. While the emitters were 
spaced 0.6-m, an emitter was not always visible after excavation; where an emitter was 
uncovered, the soil was further excavated so as to not include any exposed emitter in the 
test.  The justification for this action was that application uniformity may have been 
impacted by the immediate soil microenvironment around the buried emitters, and we 
wanted to maintain that potential integrity during testing.  The drip line was severed at 
both ends of the 12.2-m test section in order to install the necessary plumbing and 
pressure transducers for interaction with the testing apparatus (Figures 3a and 3b). 
 The isolated test section was charged with water from the apparatus for 
approximately 15 minutes at the first pressure reading of 48 kPa. Obvious leaks were 
101 
 
repaired during this time in order to reliably determine that the only water being lost from 
the apparatus was through the buried emitters.  Data collection then occurred for a 5-
minute period; water reservoir mass was recorded at the beginning and ending of this 
period, and target operating pressure was maintained via pressure regulation and 
transducer feedback. Vessel-mass differential was equated to cumulative discharge for 
that sample.  After the 5-minute period expired, water pressure was readjusted to the next 
desired threshold, and the sample line was conditioned to the new pressure setting for > 5 
minutes. This process continued for all four desired pressure levels. When the test was 
completed for the drip-line sample, the interface was removed, and drip-line couplers 
were used to reattach the sample to the existing drip line.  In addition, position readings 
were recorded for the excavated locations using the RTK GPS system.  Positions were 
recorded for two reasons: 1) to provide documentation of accurate locations of the line 
couplers for maintenance records, and 2) to provide a ?ground-truth? solution to the 
accuracy of the slope model derived through GIS. 
 Emitter discharge equivalents for drip-line samples were subjected to GLM 
procedure and separated according to Tukey?s Studentized Range (SAS Institute, Cary, 
NC, USA) to evaluate differences among operating pressure, irrigation use and field 
slope effects.  Flow-rate versus pressure curves were produced using MS Excel 2007. 
5.5 RESULTS AND DISCUSSION 
5.5.1 LABORATORY INVESTIGATIONS 
 At a constant operating pressure of 117 kpa, mean emitter discharge rate was 
measured for the horizontal drip-tape sample at 1.04 L h-1 and 1.06 L h-1 for the sloped 
102 
 
sample (CV=3.7% for both conditions); the emitter discharges in both cases were not 
statistically different (p=0.8081) from one another.  Results from variable-pressure tests 
are presented in Table 5.1.  Results indicated that operating pressure significantly 
affected emitter discharge rate (p=0.0024).  Only the discharge at the 48 kPa pressure 
differed from the other pressures tested by approximately 0.010 L h-1.  Because the 
resolution of per-emitter discharge testing was 1 mL emitter-1 (5 min)-1, statistical 
differences between pressure levels should be regarded as tenuous.  For all four operating 
pressures, the observed mean emitter discharge rate did not exceed ?10% of the nominal 
discharge rate of 0.984 L h-1 (Netafim USA, 2010). 
Table 5.1. Emitter discharge rates (L h-1) over a range of operating pressures on 20 
contiguous emitters subjected to a 6-percent slope rise (laboratory conditions). 
Operating 
Pressure 
(kPa) 
Mean Emitter Discharge 
Rate 
(L h-1)a 
Change from Nominal 
Rate 
(%)b 
Avg. 
CV 
(%) 
48 1.052 a +7.3 4.3 
83 1.041 b +6.2 4.3 
117 1.043 b +6.4 4.3 
138 1.043 b +6.4 4.0 
a Mean separation according to Tukey?s Studentized Range.  Discharges with the same letter are not 
significantly different (?<0.05). 
b Nominal rate = 0.984 L h-1 from manufacturer specifications (Netafim USA, 2010). 
 
5.5.2 LABORATORY INVESTIGATIONS OF SDI TESTING APPARATUS 
 Results from laboratory tests are summarized in Table 5.2.  Pressure response was 
a significant factor (p=<0.0001) in per-emitter sampling.  Per-emitter sampling revealed a 
0.016 L h-1 difference between 48 and 138 kPa, a similarity between 48 and 83 kPa, and a 
similarity between 83 and 117 kPa (?=0.05).  Laboratory emitter discharge rate across all 
pressure levels was 1.036 L h-1.  That value, with a ?5% margin of error, was used as a 
103 
 
baseline performance value in making field management decisions regarding in-situ 
product performance with respect to drip-tape sample discharge. 
Table 5.2. Per-emitter and apparatus discharge values over a range of operating 
pressures on 100 contiguous emitters utilizing SDI testing apparatus (laboratory 
conditions). 
Operating 
Pressure 
(kPa) 
Mean Emitter 
Discharge 
(L h-1)a 
CV 
(%) 
Mean 
Apparatus 
Discharge 
(L h-1)a,b 
SD 
Apparatus 
Discharge 
(L h-1) 
Change from 
Emitter 
Discharge 
(%) 
48 1.028   c 4.8 1.032a 0.012 +0.4 
83 1.035 bc 5.0 1.032a 0.012 -0.3 
117 1.036   b 5.5 1.032a 0.000 -0.4 
138 1.044   a 5.0 1.040a 0.007 -0.4 
a Mean separation according to Tukey?s Studentized Range.  Discharges (per column) with the same letter 
are not significantly different (?<0.05). 
b Derived from kilograms water emitted through drip-tape sample. 
 
 Pressure response was not a significant factor (p=0.5376) in influencing sample 
discharge rate utilizing laboratory data derived from the apparatus (Table 5.2).  More 
importantly, apparatus equivalents derived through this process varied less than one-half 
percent from per-emitter discharge averages. The small variation between per-emitter 
values and apparatus equivalents provided strong evidence for validating that the testing 
apparatus could be utilized for estimating in-situ drip-line discharge throughout the field. 
 For the laboratory study, discharge values were recorded for equal to or greater 
than 95% of the emitter population.  On occasion, cups were either misplaced so as to not 
catch the discharge, or cups were knocked over before a volume could be recorded.  
Therefore, each replication had between 95 and 100 emitters that contributed to the mean 
emitter discharge rate and CV.  Consistently throughout all laboratory tests, CV was 
higher than that stated by the manufacturer of 2.5% (Netafim USA, 2010).  One 
explanation for this discrepancy is that CV, measured as standard deviation of a single 
104 
 
emitter?s discharge over the same emitter?s mean discharge rate, did not exceed 2.5% for 
above tests.  The calculation method of CV we employed was measured as standard 
deviation of all contiguous emitters per drip-tape sample over the mean discharge rate of 
same.  Clause 8.2 of ASAE S553 (2008) may be interpreted to justify either scenario; 
however, we believe that our method is more useful from a product-evaluation 
standpoint, as an end-user would be more concerned with application uniformity for a 
population of emitters rather than individual emitters. 
5.5.3 FIELD INVESTIGATIONS 
 The interaction between operating pressure and system use frequency affected 
equivalent emitter discharges (p=<0.0001).  Results are presented in Table 5.3.  Lowest 
discharge was observed for the irrigated SDI product operating at 48 kPa.  Six of the 
eight classifications were not statistically different (?=0.05) from one another.  Emitter 
discharge estimates were within ?5% of baseline discharge rate determined from earlier 
laboratory evaluations, except for irrigated samples operating at 48 kpa (-7%). 
 
 
 
 
 
 
 
105 
 
Table 5.3. Mean emitter-discharge-equivalent values (L h-1) on 20 contiguous 
emitters utilizing SDI testing apparatus (field conditions).  Values from sites 2 and 9 
were omitted from the analysis. 
System Use 
Frequency 
Operating 
Pressure 
(kPa) 
Mean 
Equivalent 
Discharge 
(L h-1) 
Significancea 
SD 
Discharge 
(L h-1) 
     
Irrigated 48 0.967 c 0.082 
 83 1.040 ab 0.054 
 117 1.064 ab 0.042 
 138 1.073 a 0.039 
     
Non-Irrigated 48 1.025 b 0.042 
 83 1.043 ab 0.051 
 117 1.032 ab 0.019 
 138 1.043 ab 0.026 
     
a Mean separation according to Tukey?s Studentized Range.  Discharges with the same letter are not 
significantly different (?<0.05). 
  
 Surface-slope modeling was not an accurate depiction of anticipated drip-tape 
slope.  Essentially, the calculated surface-slope model served as a reasonable three-
dimensional representation of the field, but the drip-line orientation could be better 
described in a two-dimensional sense.  As the GIS model interpolated slopes between 
neighboring locations, slope trends were spatially distributed in a 360-degree ?view? 
across the ground surface; however, the drip-line orientation was spatially fixed in two 
directions opposite to one another.  The result was that the slope model, while depicting a 
given slope in a given area, did not accurately indicate the estimated slope relative to the 
orientation of the drip tape.  Drip-tape samples exhibited slope between 0.3 to 1.9 percent 
rise; unfortunately, the drip-tape samples collected in the field were unbalanced in 
regards to the original experimental design, so no proper evaluation of the effect of slope 
on product performance could be conducted. 
106 
 
 Equivalent-discharge data from three testing sections were beyond ?5% of 
baseline discharge estimate (Figure 5.5) which would warrant further investigation in a 
diagnostics setting.  For example, at site 2, a pin-hole leak appeared during the 83-kPa 
test, and the leak was not repaired until the final pressure test was concluded; the trend 
line illustrates the fact that a non-pressure-compensated emission was present.  Also, sites 
9 and 11 depict potential emitter under- and overwatering, respectively.  Having the 
ability to document flow-rate estimates offers the potential benefit of utilizing this 
procedure and equipment as a site-specific diagnostic tool for in-situ SDI performance. 
 
 
Figure 5.5.  Pressure-versus-equivalent-discharge curves for three SDI samples.  
Per-emitter discharge witnessed from product evaluation in a laboratory setting 
(?Baseline?) is represented as 1.036 L h-1 ?5% margin.   
 
 One example of how this ?site-specific? SDI diagnostics tool can be integrated 
with supplemental GPS and/or GIS information is represented in Figure 5.6.  Cotton is 
107 
 
the marketable crop for this experimental plot, and spatial yield data as bales/ha lint from 
the 2008 growing season were collected using a yield monitoring system.  Referring to 
the Site 9 example from above, in which the estimated discharge was 48 percent lower 
than the expected, baseline value, lint yield for the treatment block that contained the Site 
9 sample averaged 1.12 bales/ha, while the mean yield for all eight irrigated plots was 
1.26 bales/ha.  However, summary yield data in the immediate vicinity of Site 9 indicated 
that 1.26 bales/ha was produced, and the mean yield witnessed in the vicinity of irrigated 
drip-line samples was 1.29 bales/ha.  In addition, the lowest yield observed at an 
irrigation-active sampling location was 1.09 bales/ha at Site 6, a sample with estimated 
discharge that corresponded with other drip-line samples.  While the implications with 
the available data suggested that other factors may be contributing to yield variability, 
site-specific irrigation variability cannot be ruled out for certain locations within the plot.  
Regardless, the estimated discharge observed with the SDI testing apparatus, along with 
spatially-referenced yield data, suggests that the drip tape had some degree of clogging 
for that drip-tape lateral. 
108 
 
 
 
Figure 5.6. Illustration of drip-tape testing sites corresponding to cotton lint yield.  
Yield variability is suggested in the vicinity of the SDI sample labeled as site ?9?; 
average estimated emitter discharge at site 9 was 0.54 L h-1, which is 48% less 
discharge than that stated by the manufacturer. 
5.5.4 CONSIDERATIONS FOR CONTINUING RESEARCH 
 For future SDI research, the potential for field product evaluation has been 
demonstrated using an SDI testing apparatus that can interface with a drip-tape test 
section over time.  As the drip product is used over time, performance could be 
documented to derive effective or useful design life of the product.  This testing 
technique could be a highly-desirable planning tool for SDI users and researchers.  
Recommended follow-up studies include larger or smaller drip-tape samples to evaluate 
precision of the testing apparatus and procedures.  Moisture-variation studies can be used 
to further characterize the site by combining soil-moisture sensors or other moisture-
monitoring equipment. 
 
109 
 
 
5.6 SUMMARY 
 In-situ subsurface-drip irrigation performance was evaluated using a mobile, 
trailer-mounted apparatus consisting of a water reservoir, a digital bench scale, a low-
flow centrifugal pump, and a pressure regulator.  Laboratory tests were conducted to 
evaluate behavior of an agricultural SDI product using the portable testing apparatus.  
Surface terrain and lateral locations were compiled in a GIS program to select 
representative portions of SDI laterals for field analyses.  While laboratory per-emitter 
discharge values were not statistically different as a function of slope from 0% to 6% 
(p=0.8081), values were typically greater than the discharge of 0.984 L h-1 stated by the 
manufacturer and within ?10% of the manufacturer?s value.  Operating pressure was 
found to be a significant factor in per-emitter discharge for SDI tape samples containing 
twenty emitters (p=0.0024) and 100 emitters (p=<0.0001), but mean comparisons were 
not large enough to be considered practical.  Mean equivalent discharge determined using 
the testing apparatus agreed to within 0.5 percent of laboratory results, validating the 
testing apparatus.  Slope effect could not be verified in the field, as observed drip-tape 
slope did not correspond to surface-slope estimates.  Drip-tape discharges were affected 
by an interaction between operating pressure and irrigation-system use frequency 
(p=<0.0001).   The lowest discharge was observed for the irrigated SDI product operating 
at 48 kPa pressure.  Mean discharges from non-irrigated SDI product were not 
statistically different (?=0.05) from irrigated product operating at 83 and 117 kPa.  
Emitter discharge estimates were within ?5% of baseline discharge rate determined from 
110 
 
laboratory evaluations using 100 emitters (1.036 L h-1), except for irrigated samples 
operating at 48 kpa.  Pressure-versus-discharge curves for three SDI samples indicated 
the possibility of substandard discharge.  By combining sample discharges with 
supplemental spatially-related information about the laterals (like cotton lint yield), 
variability in the cropping system can be more fully investigated with regard to precise 
application of water and nutrients in a field situation. 
111 
 
CHAPTER SIX 
CONCLUSION 
6.1 CONCLUSIONS 
 Utilizing total-station technology to document guidance system signal 
performance was deemed practicable.   Site preparation has to be made favorable for the 
line-of-sight limitations of the total station, and site-specific permanent monumentation 
must be constructed in order to relate each cross-track error (XTE) dataset to the same 
terrestrial frame of reference over time.  Provided that line-of-sight is not violated, the 
potential exists for researchers to accurately track the movements of guidance-system-
enabled agricultural vehicles over a variety of field conditions, driving speeds, and 
driving patterns.  From an office-processing perspective, two different methods of 
recording XTE were developed: mathematical rotation of positioning coordinates relative 
to a regression-line model of the drive path, and offset distance calculation from a 
terrestrial-based line model in GIS software.  Either method may be considered worthy 
from a research point of view; however, the intuitive nature of using a terrestrial-based 
line model makes the GIS approach the more desirable option.  Utilizing a tracking prism 
on the target vehicle means accounting for potential positioning error resulting from pitch 
and roll of the moving vehicle; developing a slope and aspect model of the terrain is one 
method of quantifying perceived pitch and roll of a vehicle as it traverses the field.  As 
112 
 
the test site used in this study had no major slope undulations in the direction of travel, 
tilt corrections did not significantly improve cross-track error values. 
 The next objective regarding guidance-system performance was to quantify short-
term and long-term accuracies of the four correction signals tested, with the hypothesis 
that long-term error would quantify the drift phenomenon.  RTK correction offered very 
accurate XTE throughout the durations of each study.  Drive path orientation (north-to-
south versus east-to-west) had an effect on XTE for RTK, SF2, and WAAS corrections; 
SF1 was consistent regardless of orientation.  XTE derived under SF1 and WAAS 
correction were more representative of typical manufacturer?s claims.  Per-event XTE 
values served as a more instantaneous estimation of signal repeatability, and I believe the 
variability that is associated from one testing event to the next is what operators are 
referring to when they talk about ?drift? of the guidance system.  However, a simpler 
statistical quantifier of absolute accuracy was unsigned short-term and long-term 95% 
probability. 
 Regarding the subsurface-drip irrigation (SDI) performance evaluation, the 
methodology and principles discussed provide an innovative way of determining SDI 
product performance in as unobtrusive a manner as possible for field conditions.  Because 
the principle behind the constructed testing apparatus involved using differences in water 
mass over time to equate per-emitter flow rate, the testing apparatus provided a precise, 
site-specific SDI flow-rate analysis that agreed to within a half percent of per-emitter 
discharge tests conducted in the laboratory.  For the Netafim? pressure-compensating 
product being evaluated, lowest discharge was observed for the irrigated SDI product 
operating at 48-kPa pressure.  All discharges from non-irrigated SDI product and from 
113 
 
the irrigated product operating at 83 and 117 kPa (12 and 17 psi) were not statistically 
different (?=0.05) from one another.  Emitter discharge estimates were within ?5% of 
baseline discharge rate determined from laboratory evaluations (1.036 L h-1), except for 
irrigated samples operating at 48 kpa (-7% from baseline).  Because one drip-line sample 
had substantially lower flow rate than other samples tested, geospatially-linked cotton 
yield data was compared relative to the location of the drip-line sample; evidence 
indicated that yield was affected by reduced SDI output in that location. 
6.2 OPPORTUNITIES FOR FUTURE RESEARCH 
 Regarding guidance-system testing, several potential research opportunities are 
available. While the slope and aspect models of the test site were used to estimate 
potential roll and pitch of the testing vehicle, other means of capturing vehicle roll and 
pitch should be considered, such as gyroscopic devices or accelerometers.  While the 
guidance systems used for this study provided some form of internal self-correction 
through pitch and roll, this information was not available for supplemental collection to 
compare ?observed? prism tilt from ?expected? prism tilt.  Therefore, other guidance 
systems must either have the capability of providing roll/pitch information to an outside 
source (like a data logger), or supplemental accelerometers would have to be installed on 
the target prism.  With vehicle pitch/roll information, it may also be possible to mount the 
target prism in a more accessible location on the vehicle to reduce line-of-sight total-
station limitations, such as on the vehicle roof.  Such a procedure would also involve 
some form of synchronizing the prism tilt with the instant at which the target prism is 
observed by the total station. 
114 
 
 For the 2006 and 2007 studies, testing intervals were not as uniform as initially 
desired. As has been noted from previous research and confirmed through this project, 
guidance-system accuracy can be affected by time of day.  A more precise way of 
describing this phenomenon may be to conjecture that guidance-system accuracy can be 
affected by satellite geometry and availability.  While Dilution of Precision (DOP) values 
(particularly Horizontal DOP or Position DOP) are often regarded in the GPS community 
as a measure of satellite geometry, synchronizing testing events with predicted DOP 
values may provide a more apples-to-apples comparison of system performance over 
time.  Relating to this topic, the concept of sidereal time refers to the amount of time that 
a celestial object returns to its original position relative to reference points on Earth; 
instinctively, one may think that 24 hours constitutes sidereal time, but the GPS 
constellation has a slightly smaller interval of time than 24 hours.  For sake of simplicity, 
this would suggest that, if a GPS sidereal day is 23 hours instead of 24 hours, a truer 
replication of a testing event from a previous test may require testing one hour earlier on 
the second day than the first day, as the GPS constellation would be more consistent 
throughout the two tests. 
 For the SDI testing, the methodology presented provides for potential continued 
product evaluation in terms of sample discharge, as the locations for each interface were 
documented with accurate positioning equipment, and the SDI testing apparatus can 
easily interface with the same drip-tape sample over time.  As the product ages, 
reductions in discharge uniformity could be documented accordingly in order to derive an 
estimated ?design life? of the product, which would be a highly-desirable planning tool 
for SDI users.  Further moisture-variation studies can be supplemented by the site-
115 
 
specific nature of this application, by combining soil-moisture meter data or other 
moisture-monitoring equipment, in an effort to capture the nature of water movement in 
the localized soil zone.  Also, while one of the motivating factors behind this SDI 
research was to explain whether certain locations exhibiting within-plot variability 
existed for the test plot from SDI-emission variability, this evaluation was not 
specifically conducted. 
6.3 FINAL CONSIDERATIONS 
 Using a total station to document guidance-system performance is, in my opinion, 
a worthwhile technique to continue.  For researchers that wish to evaluate dynamic 
accuracy, yet may have limited acreage in which to perform large drive tests, a total 
station is ideal.  While the accuracy of our total station was not considered accurate 
enough to evaluate the RTK correction, the total station would be sufficient to evaluate 
other available correction signals.  Also, while the total station provided 5-mm accuracy 
in the tracking mode, some of the claims about RTK performance can still be validated 
regarding excellent accuracy (centimeter level) and repeatability over time, and I would 
ultimately recommend that farmers invest in this correction signal if their intent is to 
maintain any row-cropping system.  Utilizing RTK correction is an expensive investment 
initially, but ways exist to reduce costs by utilizing other civilian-owned or government-
owned reference stations to obtain an RTK solution.  Also, from a producer?s perspective, 
practical positioning accuracy is not only related to how well the vehicle follows a line, 
but also how well the implement follows the same line.  If the farming ?system,? meaning 
116 
 
the vehicle plus the implement, has a limiting performance error of >5 cm, then our total 
station may be considered as an appropriate measurement tool. 
 The need exists for researchers and industry to agree on a dynamic testing 
standard.  When quantifying guidance accuracy, there is not one single value that can be 
offered at this time to describe, in plain language, just how well a guidance system 
performs.  GPS-based dynamic accuracy is a complex topic; many variables contribute to 
positioning error from a static-observation perspective alone, and when the GPS/GNSS 
receiver is in motion, additional error sources exist.  While my desire was to derive a 
statistical term for ?drift,? with one value to capture the fact, the topic is too complex to 
currently reach that goal. 
 The techniques presented for in-situ SDI evaluations are simple, repeatable, and 
the values are easily understood.  I believe the limiting factor regarding the testing 
apparatus was its power-consumption and the resolution of the bench scale; however, the 
premise of in-situ uniformity testing is sound, and I would recommend to other 
researchers that this concept be expanded upon over a range of drip-tape products, 
operating pressures, fluid temperatures, and fertigation/effluent mixes. 
 
117 
 
REFERENCES 
 
Adamchuk, V. I., R. M. Hoy, G. E. Meyer, and M. F. Kocher.  2007.  GPS-based auto-
guidance test program development.  Precision agriculture ?07: presented at the 
6th European Conference on Precision Agriculture, Skiathos, Greece, June 3-6, 
2007. 
Ayars, J., C. Phene, R. Hutmacher, K. Davis, R. Shoneman, S. Vail, and R. Mead.  1999.  
Subsurface drip irrigation of row crops: a review of 17 years of research at the 
Water Management Research Laboratory.  Agricultural Water Management 
42(1): 1-27. 
ASAE Standards, 43rd ed.  1996.  EP458.  Field evaluation of microirrigation systems.  
St. Joseph, Mich. ASAE. 
ASAE Standards.  2008.  S553 DEC2000 (R2008): Collapsible emitting hose (drip tape) - 
specifications and performance testing. St. Joseph, Mich.: ASAE. 
ASAE Standards.  2005.  EP405.1 FEB03.  Design and installation of microirrigation 
systems.  St. Joseph, Mich.: ASAE. 
Bell, T.  2000.  Automatic tractor guidance using carrier-phase differential GPS.  
Computers and Electronics in Agriculture 25: 53-66. 
 
 
118 
 
Bittner, G.  2000.  AGRO NAV? autonomous, off-road vehicle navigation and 
implement control system, using CDGPS and inertial backup.  AgEng 2000, 
Proceedings of the International Conference on Agricultural Engineering, 
Warwick, England.  00-IE-007. 
Bryla, D. R., G. S. Banuelos, J. P. Mitchell.  2003.  Water requirements of subsurface 
drip-irrigated faba bean in California.  Irrigation Science 22(1): 31-37. 
Bordovsky, J. P. and D. Porter.  2003.  Cotton response to pre-plant irrigation level and 
irrigation capacity using spray, LEPA, and subsurface drip irrigation.  ASAE 
Meeting Paper No. 032008.  St. Joseph, Mich.: ASAE.   
Buick, R.  2002.  GPS guidance ? making an informed decision.  Proceedings of 6th 
International Conference on Precision Agriculture and Other Precision 
Resources Management.  Minneapolis, MN, 14-17 July 2002.  American Society 
of Agronomy, Madison, WI. 
Camp, C. R., P. J. Bauer, P. G. Hunt.  1997.  Subsurface drip irrigation lateral spacing 
and management for cotton in the southeastern Coastal Plain.  Transactions of the 
ASAE 40(4): 993-999.   
Camp, C. R., E. J. Sadler, W. J. Busscher.  1997.  A comparison of uniformity measures 
for drip irrigation systems.  Transactions of the ASAE 40(4): 1013-1020.   
Camp, C. R.  1998.  Subsurface drip irrigation: a review.  Transactions of the ASAE 
41(5): 1353-1367.   
Charlesworth, P. B. and W. A. Muirhead.  2003.  Crop establishment using subsurface 
drip irrigation: a comparison of point and area sources.  Irrigation Science 22: 
171-176.   
119 
 
Deere and Company. 2004. GreenStar? guidance systems---parallel tracking and 
AutoTrac assisted steering systems. Moline, Ill.: Deere and Company. Manual 
OMPC20353: p.146.  
Devore, J. and N. Farnum. 2005.  Applied statistics for engineers and scientists: second 
edition.  Thomson Brooks/Cole, Belmont, California, USA. 
Dougherty, M., A. H. AbdelGadir, J. P. Fulton, E. van Santen, L. M. Curtis, C. H. 
Burmester, H. D. Harkins, and B. E. Norris.  2009.  Subsurface drip irrigation and 
fertigation for north Alabama cotton production.  Journal of Cotton Science 13: 
227-237. 
Downey, D., D. K. Giles, D. C. Slaughter.  2003.  Ground based vision identification for 
weed mapping using DGPS.  ASAE Meeting Paper No. 031005.  St. Joseph, 
Mich.: ASAE.   
Ehrl, M., M. R. Demmel, H. Auernhammer, W. V. Stempfhuber, W. Maurer, and T. 
Wunderlich.  2002.  Spatio-temporal quality of precision farming applications.  
ASAE Meeting Paper No. 023084.  St. Joseph, Mich.: ASAE.   
Ehrl, M., W. V. Stempfhuber, M. R. Demmel, M. Kainz, and H. Auernhammer.  2004.  
AutoTrac ? accuracy of a RTK DGPS based autonomous vehicle guidance system 
under field conditions.  ASAE Publication Number 701P1004.   
Ehsani, M. R., M. Sullivan, J. T. Walker, T. L. Zimmerman.  2002.  A method of 
evaluating different guidance systems.  ASAE Meeting Paper No. 021155.  St. 
Joseph, Mich.: ASAE.   
120 
 
Ehsani, M. R., M. D. Sullivan, T. L. Zimmerman, T. Stombaugh.  2003.  Evaluating the 
dynamic accuracy of low-cost GPS receivers.  ASAE Meeting Paper No. 031014.  
St. Joseph, Mich.: ASAE.   
Enisco-Medina, J., B. L. Unruh, J. C. Henggeler, W. L. Multer.  2002.  Effect of row 
pattern and spacing on water use efficiency for subsurface drip irrigated cotton.  
Transactions of the ASAE 45(5): 1397-1403.   
Freeland, R. S., R. E. Yoder, J. T. Ammons, L. L. Leonard.  2002.  Mobilized surveying 
of soil conductivity using electromagnetic induction.  Applied Engineering in 
Agriculture 18(1): 121-126. 
Freimann, R.  2000.  Investigation of position accuracy of an autonomous, off-road 
vehicle navigation.  AgEng 2000: Proceedings of the International Conference on 
Agricultural Engineering. Warwick, England.  00-IE-007ii. 
Gan-Mor, S. and R. L. Clark.  2001.  DGPS-based automatic guidance ? implementation 
and economical analysis.  ASAE Meeting Paper No. 011192.  St. Joseph, Mich.: 
ASAE.   
Giles, D. K. and D. Downey.  2003.  Quality control verification and mapping for 
chemical application.  Precision Agriculture 4(1): 103-124.   
Guo, L. S., Q. Zhang, L. Feng.  2003.  A low-cost integrated positioning system of GPS 
and inertial sensors for autonomous agricultural vehicles.  ASAE Meeting Paper 
No. 033112.  St. Joseph, Mich.: ASAE.   
Han, S., Q. Zhang, H. K. Noh.  2001.  Applying filtering techniques to improve GPS 
positioning accuracy.  ASAE Meeting Paper No. 011158.  St. Joseph, Mich.: 
ASAE.   
121 
 
Han, S., Q. Zhang, H. Noh, and B. Shin. 2004. A dynamic performance evaluation 
method for DGPS receivers under linear parallel-tracking applications. 
Transactions of the ASAE 47(1): 321-329. 
Hest, D.  2007.  On target?  Penton Media, Inc.  Available at http:// 
farmindustrynews.com/ farm-equipment/ precision-farming/ farming_on_target/ .  
Last accessed February, 2008. 
Hurn, J.  1989.  GPS: A Guide to the Next Utility.  Trimble Navigation, Sunnyvale,CA.   
Hurn, J.  1993.  Differential GPS Explained.  Trimble Navigation, Sunnyvale, CA.   
Institute of Navigation (ION) Standards.  1997.  Recommended test procedures for GPS 
receivers, ION STD 101.  ION, Alexandria, Virginia, USA. 
Lamm, F. R., T. P. Trooien, H. L. Manges, H. D. Sunderman.  2001.  Nitrogen 
fertilization for subsurface drip-irrigated corn.  Transactions of the ASAE 44(3): 
533-542.   
Lamm, F. R. and T. P. Trooien.  2003.  Subsurface drip irrigation for corn production: a 
review of 10 years of research in Kansas.  Irrigation Science 22: 195-200.   
Lamm, F. R., A. J. Schlegel, G. A. Clark.  2004.  Development of a best-management 
practice for nitrogen fertigation of corn using SDI.  Applied Engineering in 
Agriculture 20(2): 211-220.   
Lesikar, B., V. Weynand, and R. Persyn.  2004.  Evaluation of the application uniformity 
of subsurface drip distribution systems.  ASAE Publication Number 701P0104. 
Li, J., L. Chen, and Y. Li.  2009.  Comparison of clogging in drip emitters during 
application of sewage effluent and groundwater.  Transactions of the ASABE 
52(4): 1203-1211. 
122 
 
Liu, H. and G. Huang.  2009.  Laboratory experiment on drip emitter clogging with fresh 
water and treated sewage effluent.  Agricultural Water Management.  96(5): 745-
756. 
Nagasaka, Y., N. Umeda, Y. Kanetai.  2002.  Automated rice transplanter with GPS and 
FOG.  Automation Technology for Off-Road Equipment: Proceedings of the 
ASAE.  Publication No. 701P0502, pp. 190-195.  St. Joseph, Mich.: ASAE. 
Navcom Technology, Inc.  March 2003.  StarFire datasheet.  Available at:   
http://www.navcomtech.com/support/Download/StarFire%20Network_050302-
4%20low%20res.pdf. Last accessed April 2010. 
Netafim USA.  2010.  DripNet PC 875 product specifications.  Available at: 
http://www.netafimusa.com/ files/ literature/ agriculture/ thinwall-dripperline/ 
DNPC-875.pdf.  Last accessed April, 2010. 
Noguchi, N., M. Kise, K. Ishii, H. Terao.  2002.  Field automation using robot tractor.  
Automation Technology for Off-Road Equipment.  Publication No. 701P0502:  
239-245.  St. Joseph, Mich.: ASAE. 
O?Brien, D. M., D. H. Rogers, F. R. Lamm, G. A. Clark.  1998.  An economic 
comparison of subsurface drip and center pivot sprinkler irrigation systems.  
Applied Engineering in Agriculture 14(4): 391-398.   
Reid, J. F., Q. Zhang, N. Noguchi, M. Dickson.  2000.  Agricultural automatic guidance 
research in North America.  Computers and Electronics in Agriculture 25(1): 155-
167. 
   
123 
 
Safi, B., M. R. Neyshabouri, A. H. Nazemi, S. Massiha, and S. M. Mirlatifi,  2007.  
Water application uniformity of a subsurface drip irrigation system at various 
operating pressures and tape lengths.  Turkish Journal for Agriculture and 
Forestry 31(5): 275-285. 
Sharpe, T., R. Hatch, and F. Nelson.  2002.  StarFire and real-time GIPSY: a global high-
accuracy differential GPS system.  Torrance, Cal.: NavCom Technology, Inc.  
Available at http://www.navcomtech.com/ Support/ Download/ 
StarFire%20and%20Real-Time%20GIPSY.pdf .  Last accessed March 2008. 
Song, I., S. Stine, J. Pimentel, C. Y. Choi, C. Gerba.  2003.  The role of irrigation and 
wastewater: comparison of subsurface drip irrigation and furrow irrigation.  
ASAE Meeting Paper No. 032373.  St. Joseph, Mich.: ASAE. 
Srivastava, A. K., C. E. Goering, R. P. Rohrbach, and D. R. Buckmaster.  1993.  Chapter 
6: Precision Agriculture. In Engineering Principles of Agricultural Machines, 2nd 
ed., p. 123 ? 138.  St. Joseph, Mich.: ASABE.   
Steele, D. D., R. G. Greenland, B. L. Gregor.  1996.  Subsurface drip irrigation systems 
for specialty crop production in North Dakota.  Applied Engineering in 
Agriculture 12(6): 671-679. 
Stombaugh, T., S. A. Shearer, J. Fulton, R. Ehsani.  2002.  Elements of a dynamic GPS 
test standard.  ASAE Meeting Paper No. 021150.  St. Joseph, Mich.: ASAE. 
Stombaugh, T., D. McLaren, and B. Koostra.  2005.  The global positioning system.  
University of Kentucky Cooperative Extension circular AEN-88.  Issued August, 
2005. 
124 
 
Sullivan, D. G., J. P. Fulton, J. N. Shaw, and G. Bland.  2007.  Evaluating the sensitivity 
of an unmanned thermal infrared aerial system to detect water stress in a cotton 
canopy.  Transactions of the ASABE 50(6): 1963-1969.   
Taylor, R. K., M. D. Shrock, J. Bloomfield, G. Bora, G. Brockmeier, W. Burton, B. 
Carlson, J. Gattis, R. Groening, J. Kopriva, N. Oleen, J. Ney, C. Simmelink, and 
J. Vondracek.  2004.  Dynamic testing of GPS receivers. Transactions of the 
ASAE 47(4): 1017-1025.  
Tian, L.  2002.  Development of a sensor-based precision herbicide application system.  
Computers & Electronics in Agriculture 36(2): 133-149. 
Trimble Engineering. 2001. Trimble 5600 data sheets. Dayton, Ohio: Trimble 
Engineering and Construction Group. Available at http://www.trimble.com . Last 
accessed July 2006. 
Trimble Surveying.  2007.  Trimble survey controller help: version 11.40, revision A.  
Sunnyvale, California: Trimble Surveying.  Available at http:// 
trl.trimble.com/docushare/dsweb/Get/Document-377255/TSCv1140_Help_ 
English.pdf.  Last accessed April 2010. 
Trooien, T. P., F. R. Lamm, L. R. Stone, M. Alam, D. H. Rogers, G. A. Clark, A. J. 
Schlegel.  2000.  Subsurface drip irrigation using livestock wastewater: dripline 
flow rates.  Applied Engineering in Agriculture 16(5): 505-508.  
Upadhyaya, S. K., G. S. Pettygrove, J. W. Oliveira, and B. R. Jahn.  2003.  An 
introduction ? global positioning system.  Available at http://www. 
calspace.ucdavis.edu/Docs/Intro-GPS-Final.pdf .  Last accessed March 2008.   
125 
 
Westphalen, M. L., B. L. Steward, and S. Han.  2003.  Topographic mapping through 
measurement of vehicle attitude.  ASAE Meeting Paper No. 031008.  St. Joseph, 
Mich.: ASAE. 
Wu, C., P. Ayers, and A. Anderson.  2005.  Influence of travel directions on the GPS 
dynamic accuracy for vehicle tracking.  ASAE Meeting Paper No. 051088.  St. 
Joseph, Mich.: ASAE.   
Yao, H. and R. L. Clark.  2000.  Development of topographic maps for precision farming 
with medium accuracy GPS receivers.  Applied Engineering in Agriculture 16(6): 
629-636.   
126 
 
APPENDIX A    
TRACTOR SPECIFICATIONS
127 
 
A.1. JOHN DEERE 6420 TRACTOR 
 
Figure A.1. John Deere 6420. 
 
Tractor Power: 
PTO  rated, kW:      70.3 
 
Engine: 
Manufacturer:       John Deere 
Fuel:        Diesel 
Aspiration:       Turbocharger with 
intercooler 
Cylinders:       4 
Displacement, L:      4.5 
Rated engine speed, RPM:     2300 
Cooling:       liquid 
Oil capacity, L:      15.9 
Hydraulic flow rate, LPM:     96 
 
Transmission: 
Type:        Infinitely Variable 
Transmission 
 
Mechanical: 
MFWD:       Yes 
 
Dimensions: 
Wheelbase, mm:      2400 
128 
 
APPENDIX B    
TOTAL-STATION SPECIFICATIONS
129 
 
B.1 TRIMBLE 5603 TOTAL STATION 
 
Figure B.1. Trimble 5603 Total Station. 
 
Total station specifications: 
Manufacturer:       Trimble 
Model:        5603 
Firmware version:      696-03.08 
 
Angle measurement: 
Accuracy (standard deviation):    3? 
Automatic level compensator:    Dual-axis compensator ?6? 
 
Distance measurement: 
Accuracy (standard deviation), prism:   ?(2 mm +2 ppm), standard; 
        ?(5 mm+2 ppm), tracking 
 
Measuring time, prism: 
Standard measurement:     2 s (0.5 Hz) 
Tracking:       0.5 s (2 Hz) 
 
Range: 
Under standard clear conditions, 1 prism:   3000 m
130 
 
B.2 DATALOGGER SPECIFICATIONS 
 
Figure B.2. Trimble TSCe datalogger. 
 
Manufacturer:       Trimble 
Model:        TSCe 
Part no.:       45185-20 
Operating system:      Microsoft Windows CE 
        Version 4.0.12 
Internal memory capacity:     16 MB 
 
131 
 
B.3 DATA COLLECTION SOFTWARE SPECIFICATIONS 
 
Figure B.3. Trimble Survey Controller. 
 
Manufacturer:       Trimble 
Model:        Trimble Survey Controller 
Software version:      11.20 ? 11.40
132 
 
APPENDIX C    
TARGETING PRISM SPECIFICATIONS 
133 
 
C.1 TRIMBLE RMT PRISM 
 
Figure C.1. Trimble RMT prism. 
 
Manufacturer:       Trimble 
Model:        RMT ATS Multi-Channel 
Part No.:       571233035 
Prism constant, mm:      10 
Power supply, V:      12 ? 35, external source 
Infrared channels:      4 
134 
 
APPENDIX D    
GUIDANCE-SYSTEM SPECIFICATIONS 
135 
 
D.1. JOHN DEERE GREENSTAR SYSTEM 
 
Figure D.1. John Deere Greenstar display. 
 
Manufacturer:       John Deere 
Model:        Greenstar Mobile Processor 
Hardware version:      PF80444 2.5 
Software/firmware version:     PF303192F 
Loader:       2.30A 
Mapping processor software version:    PF301660D 
Runtime, hr (until December, 2007):    246.8    
Correction signals available:     RTK, SF2, SF1, WAAS 
 
Serial output: 
NMEA messages:      GGA, GSA, VTG 
Baud:        19200 
Output rate:       5 Hz 
 
136 
 
D.2. JOHN DEERE STARFIRE ITC GPS RECEIVER 
 
Figure D.2. StarFire iTC GPS receiver (yellow dome). 
 
Manufacturer:       John Deere 
Model:        StarFire iTC 
Hardware part no.:      PF80732 
Software app. version:     2.52A 
Sofware loader version:     1.04 
Position receiver software version:    12.5 
Receiver runtime, hr (until December, 2007):  1337.9 
 
Features: 
Terrain compensation:     Yes 
External port:       RS232 (serial), configurable 
 
 
137 
 
D.3. TRIMBLE EZ-GUIDE GUIDANCE COMPUTER 
 
Figure D.3. Trimble EZ-Guide display and accessory GPS antenna. 
 
Manufacturer:       Trimble 
Model:        EZ-Guide Plus 
Part no.:       52001-00 
Software version no. (2007 tests):    4.10.002 Feb 22 2007 
11:53:46 
 
Serial port settings (2007 tests): 
Input:        External GPS 
Baud:        19200 bps 
Parity:        8-N-1 
Output rate:       5 Hz   
 
Supplemental information: 
 For the WAAS drive-path tests in the 2006 study, a Trimble EZ-Guide Plus with 
accessory GPS antenna was used.  The equipment did not include any form of terrain 
compensation.  Also, the EZ-Guide Plus with accessory GPS antenna was loaned to the 
department on an evaluation basis; at the end of the 2006 tests, the equipment was 
returned to the owner, and similar equipment was purchased by the department later that 
year for the 2007 tests.  No configuration information was recorded for the equipment 
bundle used in 2006.  For the WAAS drive-path tests in the 2007 study, the accessory 
GPS antenna was replaced with the John Deere StarFire iTC receiver.  Configuration 
settings in the EZ-Guide computer were adjusted accordingly.  In addition, terrain 
compensation module was available. 
138 
 
D.4. TRIMBLE EZ-STEER ASSISTED-STEERING SYSTEM WITH TERRAIN COMPENSATION 
MODULE 
 
Figure D.4. Trimble EZ-Steer assisted-steering system. 
 
Manufacturer:       Trimble 
 
EZ-Steer specifications (2007): 
Software version:      3.00.01 
Tilt:        Basic 
Filter:        Heavy3 (under Terrain 
Comp.) 
 
Vehicle setup: 
Type:        Tractor 4WD 
Steering wheel diameter, in.:     16.0 
Angle/turn, deg.:      21 
Freeplay left, in.:      1.80 
Freeplay right, in.:      1.80 
Wheelbase, in.:      95 
 
Engagement: 
Aggressiveness, percent:     120 
Minimum / maximum speed, mph:    2 / 15 
Maximum angle, deg.:     15 
Motor speed:       Auto High 
Engage on AB:      On 
Minimum number of satellites:    5 
Maximum HDOP:      3.0 
 
Features: 
External port:       RS232 (serial), configurable 
139 
 
 
APPENDIX E    
CORRECTION SIGNALS USED IN GUIDANCE-ACCURACY TESTS
140 
 
E.1 CORRECTION-SIGNAL CHARACTERISTICS 
Table E.1.  Characteristics of GPS-based correction signals used in 2006 and 2007 
tests. 
     
Correction Signal:  Wide Area Augmentation System (WAAS) 
 Type of Correction: wide-area 
 Single vs. Dual Frequency: single frequency 
 Operating Fee: free 
 Accuracy Claims: ?15 to 30 cm pass-to-pass 
  ?61 to 66cm long-term static 
 Supplemental information: Service is offered by U.S. government, designed 
     for Federal Aviation Administration (FAA) to offer 
     navigational aid to airplanes; many inexpensive 
     personal navigation devices can utilize WAAS 
 
Correction Signal: John Deere SF1 
 Type of Correction: wide-area 
 Single vs. Dual Frequency: single frequency 
 Operating Fee: free (with compatible John Deere receivers) 
 Accuracy Claims: ?33 cm pass-to-pass 
  ?76 cm static 
 Supplemental information: SF1 signal is proprietary to John Deere 
 
Correction Signal: John Deere SF2 
 Type of Correction: wide-area 
 Single vs. Dual Frequency: dual frequency 
 Operating Fee: $800 annually 
 Accuracy Claims: ?10 cm pass-to-pass 
  ?25 cm static 
 Supplemental information: SF2 signal is proprietary to John Deere 
 
Correction Signal Real-Time Kinematic (RTK) 
 Type of Correction: local-area 
 Single vs. Dual Frequency: dual frequency 
 Operating Fee: free (initial set-up costs up to $12K) 
 Accuracy Claims: ?3 cm pass-to-pass 
  ?3 cm static 
 Supplemental information: RTK procedure employs two dual-frequency 
  receivers; one is located at a fixed location (?base?), 
  while the other travels with the vehicle (?rover?); 
  most install costs come from 2nd receiver purchase 
 
141 
 
APPENDIX F    
GUIDANCE TESTING SITE ONE
142 
 
F.1 GUIDANCE TESTING SITE ONE 
 
Figure F.1. Guidance testing site one, used in 2006.  Elevation contours are in 2-ft 
(0.6-m) intervals; guidance lines are spaced 20 ft (6.1 m).  Direction of vehicle travel 
was westerly to easterly (approx. 80.1 degrees azimuth from true north; aerial 
imagery and contour data courtesy of City of Auburn GIS Department, Auburn, 
AL). 
143 
 
APPENDIX G    
GUIDANCE TESTING SITE TWO  
144 
 
G.1 GUIDANCE TESTING SITE TWO 
 
Figure G.1. Guidance testing site two, used in 2007.  Elevation contours are in 2-ft 
(0.6-m) intervals; guidance lines are spaced 40 ft (12.2 m).  Direction of vehicle 
travel was northerly to southerly (approx. 179.4 degrees azimuth from true north; 
aerial imagery and contour data courtesy of City of Auburn GIS Department, 
Auburn, AL). 
145 
 
 
Figure G.2. Guidance testing site two, used in 2007.  Elevation contours are in 2-ft 
(0.6-m) intervals; guidance lines are spaced 40 ft (12.2 m).  Direction of vehicle 
travel was easterly to westerly (approx. 271.3 degrees azimuth from true north; 
aerial imagery and contour data courtesy of City of Auburn GIS Department, 
Auburn, AL).
146 
 
APPENDIX H    
SDI TESTING APPARATUS PARTS 
147 
 
H.1. 12-VOLT PUMP 
 
Figure H.1. Shurflo diaphragm pump. 
 
Manufacturer:       Shurflo 
Model:        2088 
Type:        3 chamber diaphragm 
Flow rate, gpm:      2.8 
Electrical motor:      12VDC, intermittent-duty 
Maximum amps:      7.0 
Priming:       Self-priming to 9 ft. 
Ports:        ??-14 NPSM-Male 
Dimensions (L x W x H), in.:     7.8? x 5.0? x 4.6? 
Net weight, lb.:      5 
148 
 
H.2 PRESSURE REGULATOR 
 
Figure H.2. Watts adjustable small-pressure regulator. 
 
Manufacturer:       Watts Regulator 
Model:        26A 
Operating range, psi:      3 ? 50 
Ports:        ??, Female 
 
Materials: 
Body:        Brass 
Spring Cage:       Aluminum 
Disc:        Buna-N stainless steel 
Diaphragm:       Reinforced Buna-N 
 
H.3 DIGITAL BENCH SCALE 
 
Figure H.3. Ohaus CD-11 indicator (scale platform not shown). 
 
Manufacturer:       Ohaus 
 
Model: 
Indicator:       CD-11 
Bench scale:       B100S 
Scale capacity, kg:      100 
Resolution, kg:      0.02 
Indicator software version:     1.22 
 
149 
 
H.4 PRESSURE TRANSDUCER 
 
Figure H.4. PCB pressure transducer. 
Manufacturer:       PCB Piezotronics 
Model:        1501B02EZ50PSIG 
Output:       Amplified 0 ? 5 VDC FS 
Supply voltage:      6.5 ? 30 VDC 
Accuracy:       ? 0.25% FS 
Response time:      ? 1 ms 
Pressure port:       ?? ? 18 NPT, Male 
Electrical connection:      Solder tabs; power, signal, 
ground 
Pressure range:      0 ? 50 psig 
150 
 
H.5 DATA LOGGER 
 
Figure H.5. Campbell Scientific CR206 datalogger. 
 
Manufacturer:       Campbell Scientific 
Model:        CR206 
Operating system:      PakBus protocol 
Operating frequency of internal spread-spectrum radio: 900 MHz 
Power supply, external:     12 VDC 
Communications port:     RS232 (serial) 
 
Sensor input channels: 
Analog, 0 ? 5 V, single-ended:    5 
Pulse counter:       2 
Switched excitation:      2 
Digital I/O:       2 
 
H.6 RADIO MODEM 
 
Figure H.6. Campbell Scientific RF401 radio modem. 
 
Manufacturer:       Campbell Scientific 
Model:        RF401 
Operating frequency:      910 - 918 MHz 
I/O data rate:       9600 bps 
Power supply, external:     9 to 16 VDC 
Communications port:     RS232 (serial) 
151 
 
H.7 TOPCON X20 FIELD COMPUTER 
 
Figure H.7. X20 field computer screen shot illustrating pressure-monitoring 
software. 
 
Console: 
Processor:       1 GHz 
Memory:       512 Mb 
Operating system:      Windows XP Pro SP2 
Display size:       213 mm (8.4 in.) 
Solid state drive:      4 GB 
Mounting bracket:      RAM mount 
USB ports:       4 x USB 2.0 
Serial RS232 ports:      4 
PS2 ports:       2 
VGA ports:       1 
10/100 Base T Ethernet port:     1 
 
Pressure-monitoring software: 
Manufacturer:       Campbell Scientific 
Software:       LoggerNet 
Software version:      3.4.1 
Capabilities:       Datalogger programming 
        Real-time sensor monitoring 
152 
 
APPENDIX I    
TEST EVENT SHEDULE (2006)
153 
 
I.1. GUIDANCE-TESTING EVENT SCHEDULE OF 2006  
Table I.1. Testing event schedule of 2006. 
Test Name 
[a] 
 Date [b] Time of 
Day 
 Test Name 
[a] 
 Date [b] Time of 
Day 
         
R11  February 22 2:30 pm  R51  April 7 1:33 pm 
J11  --- ---  J51  April 7 1:59 pm 
D11  --- ---  D51  April 7 1:45 pm 
W11  February 22 2:45 pm  W51  --- --- 
         
R21  February 27 3:03 pm  R52  April 7 2:15 pm 
J21  --- ---  J52  April 7 2:36 pm 
D21  February 27 3:28 pm  D52  April 7 2:25 pm 
W21  February 27 3:16 pm  W52  --- --- 
         
R22  February 27 3:40 pm  R61  June 14 12:28 pm 
J22  --- ---  J61  --- --- 
D22  February 27 4:00 pm  D61  June 14 12:42 pm 
W22  February 27 3:50 pm  W61  --- --- 
         
R31  March 8 1:47 pm  R62  June 14 3:22 pm 
J31  --- ---  J62  June 14 3:49 pm 
D31  March 8 2:15 pm  D62  June 14 3:35 pm 
W31  March 8 2:04 pm  W62  --- --- 
         
R32  March 8 2:30 pm  R71  June 16 9:43 am 
J32  --- ---  J71  June 16 10:07 am 
D32  March 8 2:52 pm  D71  June 16 9:55 am 
W32  March 8 2:42 pm  W71  --- --- 
         
R41  March 28 T1 [c]  R72  June 16 12:21 pm 
J41  --- ---  J72  June 16 12:43 pm 
D41  March 28 T1  D72  June 16 12:32 pm 
W41  March 28 T1  W72  --- --- 
         
R42  March 28 T2      
J42  --- ---      
D42  March 28 T2      
W42  March 28 T2      
         
[a]  Test name is in the format A
xy, where A represents correction signal used (R=RTK, J=Starfire SF1, D=Starfire SF2, 
W=WAAS), x represents testing day, and y represents starting time of testing. 
[b]  Dashed lines represent absence of data for that iteration. 
[c]  Events T
1 and T2 have no time stamps. 
 
 
154 
 
APPENDIX J    
TEST EVENT SCHEDULE (2007) 
155 
 
J.1. GUIDANCE-TESTING EVENTS OF 2007 
Table J.1. Testing event schedule of 2007 for RTK and John Deere SF1 correction 
signals. 
Test Name[a] Date Time of 
Day 
 Test Name[a] Date Time of 
Day 
RS11 Mar 7 11:36 am  JS11 Mar 7 10:39 am 
--- --- ---  JW11 Mar 7 1:18 pm 
--- --- ---  JW12 Mar 7 2:02 pm 
RS21 Mar 14 8:45 am  JS21 Mar 14 9:24 am 
RS22 Mar 14 10:14 am  JS22 Mar 14 11:20 am 
RW21 Mar 14 1:09 pm  JW21 Mar 14 1:19 pm 
RW22 Mar 14 2:02 pm  JW22 Mar 14 2:13 pm 
RS31 Apr 13 10:46 am  JS31 Apr 13 10:30 am 
RS32 Apr 13 2:00 pm  JS32 Apr 13 1:43 pm 
RW31 Apr 13 12:59 pm  JW31 Apr 13 12:39 pm 
RW32 Apr 13 2:56 pm  JW32 Apr 13 2:42 pm 
RS41 
--- 
Apr 30 
--- 
8:19 am 
--- 
 JS41 
--- 
Apr 30 
--- 
8:56 am 
--- 
RS51 Jun 5 1:04 pm  JS51 Jun 5 1:37 pm 
RS52 Jun 5 3:52 pm  JS52 Jun 5 4:29 pm 
RW51 Jun 5 2:48 pm  JW51 Jun 5 2:33 pm 
--- --- ---  JW52 Jun 5 5:30 pm 
RS61 Jul 10 10:12 am  JS61 Jul 10 10:42 am 
--- --- ---  JS62 Jul 10 1:45 pm 
RW61 Jul 10 11:16 am  JW61 Jul 10 11:40 am 
--- --- ---  JW62 Jul 10 2:55 pm 
RS71 Jul 18 10:22 am  JS71 Jul 18 11:24 am 
--- --- ---  JS72 Jul 18 1:52 pm 
--- --- ---  JS73 Jul 18 3:34 pm 
RW71 Jul 18 10:36 am  JW71 Jul 18 11:38 am 
--- --- ---  JW72 Jul 18 2:08 pm 
--- --- ---  JW73 Jul 18 3:47 pm 
RS81 Jul 31 10:20 am  JS81 Jul 31 10:47 am 
RS82 Jul 31 1:52 pm  JS82 Jul 31 2:22 pm 
RS83 Jul 31 4:11 pm  JS83 Jul 31 4:38 pm 
RW81 Jul 31 11:23 am  JW81 Jul 31 11:48 am 
RW82 Jul 31 3:05 pm  JW82 Jul 31 3:29 pm 
RS91 Aug 3 11:12 am  JS91 Aug 3 11:57 am 
RW91 Aug 3 12:43 pm  JW91 Aug 3 1:23 pm 
RS101 Aug 23 7:45 am  JS101 Aug 23 8:12 am 
RS102 Aug 23 10:04 am  JS102 Aug 23 10:31 am 
RS103 Aug 23 1:16 pm  JS103 Aug 23 1:49 pm 
RW101 Aug 23 9:05 am  JW101 Aug 23 9:29 am 
RW102 Aug 23 11:04 am  JW102 Aug 23 11:24 am 
RW103 Aug 23 2:22 pm  JW103 Aug 23 2:55 pm 
[a]  Test name is in the format A
Xyz, where A represents correction signal used (R=RTK, J=Starfire SF1, D=Starfire SF2, 
W=WAAS), X represents drive path orientation (S=southward, W=westward), y represents testing day, and z represents 
testing event number. 
[b]  Dashed lines represent absence of data for that iteration.  
156 
 
Table J.2. Testing event schedule of 2007 for John Deere SF2 and WAAS correction 
signals.
 [a]  Test name is in the format AXyz, where A represents correction signal used (R=RTK, J=Starfire SF1, D=Starfire SF2, 
W=WAAS), X represents drive path orientation (S=southward, W=westward), y represents testing day, and z represents 
testing event number. 
[b]  Dashed lines represent absence of data for that iteration. 
Test Name[a] Date Time of Day  Test Name[a] Date Time of 
Day 
DS11 Mar 7 10:21 am  WS11 Mar 7 11:16 am 
DW11 Mar 7 1:06 pm  WW11 Mar 7 1:29 pm 
DW12 Mar 7 1:50 pm  WW12 Mar 7 2:19 pm 
DS21 Mar 14 9:07 am  WS21 Mar 14 9:43 am 
DS22 Mar 14 10:38 am  WS22 Mar 14 11:43 am 
DW21 Mar 14 12:58 pm  WW21 Mar 14 1:31 pm 
DW22 Mar 14 1:49 pm  WW22 Mar 14 2:26 pm 
DS31 Apr 13 10:03 am  WS31 Apr 13 11:15 am 
DS32 Apr 13 1:28 pm  WS32 Apr 13 2:16 pm 
DW31 Apr 13 12:26 pm  WW31 Apr 13 1:14 pm 
DW32 Apr 13 2:34 pm  WW32 Apr 13 3:13 pm 
DS41 Apr 30 8:37 am  WS41 Apr 30 9:13 am 
DW41 Apr 30 11:46 am  --- --- --- 
DS51 Jun 5 1:23 pm  WS51 Jun 5 1:53 pm 
DS52 Jun 5 4:13 pm  WS52 Jun 5 4:45 pm 
DW51 Jun 5 2:15 pm  WW51 Jun 5 3:14 pm 
DW52 Jun 5 5:13 pm  WW52 Jun 5 5:47 pm 
DS61 Jul 10 10:27 am  WS61 Jul 10 10:56 am 
DS62 Jul 10 1:28 pm  WS62 Jul 10 2:08 pm 
DW61 Jul 10 11:27 pm  WW61 Jul 10 11:53 am 
DW62 Jul 10 2:42 pm  WW62 Jul 10 3:12 pm 
DS71 Jul 18 10:54 am  WS71 Jul 18 11:55 am 
DS72 Jul 18 1:26 pm  WS72 Jul 18 2:31 pm 
DS73 Jul 18 3:09 pm  WS73 Jul 18 4:01 pm 
DW71 Jul 18 11:08 am  WW71 Jul 18 12:12 pm 
DW72 Jul 18 1:40 pm  WW72 Jul 18 2:52 pm 
DW73 Jul 18 3:23 pm  WW73 Jul 18 4:15 pm 
DS81 Jul 31 10:34 am  WS81 Jul 31 11:05 am 
DS82 Jul 31 2:07 pm  WS82 Jul 31 2:41 pm 
DS83 Jul 31 4:24 pm  WS83 Jul 31 4:51 pm 
DW81 Jul 31 11:35 am  WW81 Jul 31 12:12 pm 
DW82 Jul 31 3:17 pm  WW82 Jul 31 3:42 pm 
DS91 Aug 3 11:38 am  WS91 Aug 3 12:18 pm 
DW91 Aug 3 1:03 pm  WW91 Aug 3 1:47 pm 
DS101 Aug 23 7:59 am  WS101 Aug 23 8:42 am 
DS102 Aug 23 10:17 am  WS102 Aug 23 10:47 am 
DS103 Aug 23 1:34 pm  WS103 Aug 23 2:04 pm 
DW101 Aug 23 9:16 am  WW101 Aug 23 9:42 am 
DW102 Aug 23 11:14 am  WW102 Aug 23 11:35 am 
DW103 Aug 23 2:35 pm  WW103 Aug 23 3:08 pm 
157 
 
APPENDIX K    
CONCEPTS AND EQUATIONS USED IN GUIDANCE-SYSTEM 
PERFORMANCE TESTS
158 
 
K.1. CROSS-TRACK ERROR (XTE) 
 Cross-track error (XTE) refers to the shortest possible ground distance between an 
observed target position and its theoretical, or anticipated, position in relation to a 
computer-generated drive path.  In statistical terms, this value would be analogous to a 
residual value of an observation relative to a regression line.  This is the primary value on 
which other descriptive statistics are calculated, such as mean cross-track error, standard 
deviation of error, root mean square error, pass-to-pass accuracy, and year-to-year 
accuracy (see below). 
 From the example in figure K.1, three observed target positions are represented 
according to x,y planar coordinates (analogous to ?easting? and  ?northing? coordinates, 
respectively, in a gridded map coordinate system).  The three observations are plotted 
relative to a pre-determined guidance line or drive path.  The distance perpendicular to 
the guidance line from the observed position is the cross-track error (XTE) of that 
observation. 
 
Figure K.1. Depiction of cross-track error. 
159 
 
 Depending on the direction of vehicle travel, a ?signed? XTE (having a positive 
or negative value) will refer to which side of the guidance line the observation was taken.  
Mean cross-track error was calculated by the following equation: 
    g1850g3364g3037 g3404 g3117g3263uni2211 g1850g3036g3037g3015g3036g2880g2869     (1) 
where N = the number of data obtained from line j,  g1850g3115g3365 = mean XTE, and Xij = the 
distance from  point i to desired line j. 
K.2 STANDARD DEVIATION OF ERROR (SD) 
Standard deviation of error (SD) is considered as a measure of precision of a guidance 
system?s ability to maintain a vehicle?s drive path (Ehsani et al., 2002), as this value 
represents deviation from the mean XTE.  Standard deviation of error was calculated by 
the following equation: 
g1845g1830g3037 g3404 g3495 g2869g3015g2879g2869uni2211 g4666g1850g3036g3037 g3398 g1850g3364g3037g4667g2870g3015g3036g2880g2869         (2) 
where N = the number of data obtained from line j, Xij = the distance from  point i to 
desired line j, and  g1850g3364g3115g3365 = mean XTE for line j. 
K.3 ROOT MEAN SQUARE ERROR (RMSE) 
Root mean square error (RMSE) is considered as a measure of accuracy, as this value 
represents deviation of observed position from its theoretical, ideal position (Ehsani et 
al., 2002).  Root mean square error was calculated by the following equation: 
g1844g1839g1845g1831g3037 g3404 g3495g2869g3015 uni2211 g1850g3036g3037g2870g3015g3036g2880g2869       (3) 
where N = the number of data obtained from line j, and Xij = the distance from  point i to 
desired line j. 
160 
 
K.4 PASS-TO-PASS ACCURACY (P2P) 
Pass-to-pass accuracy (P2P) is regarded as the system?s accuracy in determining the 
location of the next drive path in relation to a recently-traveled drive path, typically 
within a 15- minute time frame.  Pass-to-pass accuracy was calculated by the following 
equation: 
g1842uni0032g1842g3037g3038 g3404 g1850g3364g3038 g3398 g1850g3364g3037    (4) 
where P2Pjk refers to pass-to-pass accuracy from line j to line k,  g1850g3038g3364g3364g3364 = mean XTE for line 
k, and  g1850g3115g3365 = mean XTE for line j. 
K.5 YEAR-TO-YEAR ACCURACY (Y2Y) 
 Year-to-year accuracy (Y2Y) is regarded as the system deviation in determining 
the location of a repetitious drive path, typically beyond a 15-minute time frame 
(separation of hours, days, etc.).  For a given time interval, Y2Y can best be described as 
the absolute difference between the minimum and the maximum mean signed XTE of a 
drive path (field, etc.). Year-to-year accuracy was calculated by using the following 
equation: 
g1851uni0032g1851g3037 g3404 uni007Cg3435g1839g1853g1876uni0020g1850g3364g3037g3439 g3398 g3435g1839g1861g1866uni0020g1850g3364g3037g3439uni007C  (5) 
where Y2Yj is the year-to-year accuracy on line j and  g1850g3115g3365 = mean XTE for line j. 
161 
 
APPENDIX L    
GUIDANCE ASSESSMENT OFFICE PROCESSING PROCEDURE (2006) 
162 
 
L.1. NECESSARY SOFTWARE 
? Trimble Survey Controller (on the TSCe) 
? Microsoft ActiveSync 
? Trimble Geomatics Office 
? Microsoft Excel 
L.2. PROCEDURE STEPS 
L.2.1. Uploaded data file from TSCe to Trimble Geomatics Office desktop 
software. 
L.2.2. Checked for obvious errors, measurement blunders, etc. 
L.2.3. Converted point data file into a delimited text file in the format Name-
Easting-Northing-Elevation-Code.  (Elevation and Code values were not 
required for further calculations.) 
L.2.4. Imported the text file into Microsoft Excel. 
L.2.5. In Microsoft Excel, a regression analysis was performed on the data set that 
represented the first established drive path; this data set also corresponded 
with the initial AB line, zero line, etc. in the guidance computer.  From the 
analysis, the coefficient value that pertains to slope of the model line was 
recorded.  An arc-tangent calculation was made on the coefficient, and this 
new value was converted to degrees; this newest value represented the slope 
of the model line in degrees relative to the positive x (easting) axis. 
L.2.6. Given the parameters of the regression line, an arbitrary coordinate was 
derived that fit on the line to the left of the first observed data point; 
likewise, an arbitrary coordinate was derived to the right of the last observed 
data point.  By using the swath width value, four additional parallel lines 
were derived to represent the neighboring theoretical drive paths. 
L.2.7. UTM coordinate values were normalized by subtracting out the original 
easting and northing coordinate values from the ensuing data block. 
L.2.8. The following equations were used to ?rotate? the data set to a model slope 
of zero: 
  g1850g3033 g3404 g1850g3036 g4674uni0063uni006Funi0073g4672 g3343g3117g3124g3116g4666uni0033uni0036uni0030g3398 g2016g4667g4673g4675g3398 g1851g3036 g4674uni0073uni0069uni006Eg4672 g3343g3117g3124g3116g4666uni0033uni0036uni0030g3398 g2016g4667g4673g4675  (6) 
  g1851g3033 g3404 g1851g3036 g4674uni0063uni006Funi0073g4672 g3343g3117g3124g3116g4666uni0033uni0036uni0030g3398 g2016g4667g4673g4675 g3397 g1850g3036 g4674uni0073uni0069uni006Eg4672 g3343g3117g3124g3116g4666uni0033uni0036uni0030g3398 g2016g4667g4673g4675  (7) 
where (Xi, Yi) were initial prism coordinates (Easting, Northing), (Xf , Yf) 
were modeled prism coordinates, and ? represented the degree conversion of 
the arc-tangent of the slope parameter from the regression-line equation. 
163 
 
L.2.9. As a double-check of the accuracy of the rotation function, the newly-
derived Northing coordinates for the data set from the initial drive path were 
compared to the residuals values calculated in the regression analysis; if 
both sets of numbers were similar, then this rotation model was utilized for 
every successive drive path and data set. 
L.2.10. Because the data sets pertaining to the neighboring drive paths were offset 
from the X axis by each row?s respective swath width, those Northing 
values were normalized to better reflect uniform cross-track-error values. 
L.2.11. Mean cross-track-error, standard deviation of error, root mean square error, 
pass-to-pass accuracy, and year-to-year accuracy were then calculated 
according to Appendix K. 
164 
 
APPENDIX M    
GUIDANCE ASSESSMENT OFFICE PROCESSING PROCEDURE (2007) 
165 
 
M.1. NECESSARY SOFTWARE 
? Trimble Survey Controller (on the TSCe) 
? Microsoft ActiveSync 
? Trimble Geomatics Office 
? ESRI ArcGIS Desktop Version 9 bundle 
? XTools Pro Version 5 
M.2. PROCEDURE STEPS 
M.2.1. Uploaded data file from TSCe to Trimble Geomatics Office desktop 
software. 
M.2.2. Checked for obvious errors, measurement blunders, etc. 
M.2.3. Converted point data file into ESRI-compatible shapefile format; formatted 
data included point name, easting, northing, elevation, and feature code for 
each data point. 
M.2.4. Point data were separated in ArcGIS (ArcMap, ArcCatalog) according to test 
attributes or site features.  If necessary, additional shapefiles were created for 
those separations for future reference and analyses. 
M.3. CREATING A SITE SLOPE/ASPECT MAP 
M.3.1. Topographic data were collected across the test site at 5-m-by-5-m spacing 
using a survey-grade GPS unit, resulting in point data with sub-inch vertical 
accuracy across the site.  Topographic data were uploaded from TSCe to 
Trimble Geomatics Office desktop software, checked for obvious errors, and 
converted into ESRI-compatibe shapefile format. 
M.3.2. In ArcMap, a map projection was defined for the data set (ArcToolbox / Data 
Management / Projections and Transformations / Define Projection).  Data 
were then inspected for potential outliers or projection blunders before 
continuing. 
M.3.3. A sub-set of topographic data were isolated from the large set for statistical 
validation.  This was accomplished through Geostatistical Analyst / Create 
Subsets ?; five percent of the topographic data were assigned to the ?test? 
dataset, while the remainder were assigned to the ?training? dataset. 
M.3.4. A surface raster image (digital elevation model, or DEM) was created from 
the ?training? dataset through ArcToolbox / 3D Analyst Tools / Raster 
Interpolation / Topo to Raster.  Output cell size was defined as ?1? so as to 
produce an image with a 1-square-meter resolution.  Drainage Enforcement 
was turned off (?No Enforce?). 
166 
 
M.3.5. The accuracy of the raster image was tested.  From ArcToolbox / Spatial 
Analyst Tools / Extraction / Extract Values to Points, the ?test? dataset was 
labeled as ?validation? for the raster grid.  The ?interpolate values at the point 
locations? feature was active.  After the procedure finished, interpolated 
elevation values were amended to the ?testing? shapefile.  This batch of 
numbers was then compared to the original, or measured, elevations from the 
field work by studentized t-test in Microsoft Excel; the test failed in rejecting 
the hypothesis that the numbers were not different, so the raster image was 
considered valid. 
M.3.6. Utilizing the raster image, a slope map was created through ArcToolbox / 3D 
Analyst Tools / Raster Surface / Slope.  Slope units were assigned as degrees.  
This function interpolated the extent of ?tilt? for each square meter of the 
raster image. 
M.3.7. Utilizing the raster image, an aspect map was created through ArcToolbox / 
3D Analyst Tools / Raster Surface / Aspect.  Aspect units were assigned as 
degrees. This function interpolated the direction of ?tilt? for each square 
meter of the raster image. 
M.4. ADJUSTING TARGET COORDINATES FOR PRISM TILT 
M.4.1. For each observed target position recorded within the test plot, an 
interpolated slope and aspect value was also associated.  This feature was 
accomplished through ArcToolbox / Spatial Analyst Tools / Extraction / 
Extract Values to Points; the relevant point shapefile and the relevant raster 
were analyzed.  The ?interpolate values at the point locations? feature was 
active.  After the procedure finished, interpolated slope and/or aspect values 
were appended to the shapefile attributes. 
M.4.2. For each dataset, the point names, northing/easting coordinates, derived 
slope, and derived aspect values were transferred to Microsoft Excel.    
Position re-adjustments by slope and aspect were determined for both the 
easting (X) and northing (Y) coordinates by the initial height of the target 
prism above the ground. 
M.4.3. Utilizing ArcCatalog, a new point-based shapefile was created using the re-
adjusted coordinates and identical point names. 
M.5. DETERMINING OBSERVATIONAL CROSS-TRACK ERROR 
M.5.1. Utilizing shapefile-editor tools in ArcMap, a polyline record was created by 
connecting the beginning and ending prism observations at the time of 
guidance-line establishment.  Also using the editor tools, four parallel 
polylines were created to represent the theoretical neighboring drive paths, 
167 
 
separated by a swath-width distance.  This polyline file served as the total-
station-established reference file for all subsequent tests. 
M.5.2. With the desired point file and the reference polyline file available, the Near 
function was used (ArcToolbox / Analysis Tools / Proximity / Near).  When 
this function finished, each point record had a distance to the nearest 
reference line appended to its attributes; by leaving the ?Angle? feature 
active during the procedure, a determination could be made as to which side 
of the reference line the point record was located (through positive or 
negative angle values). 
M.5.3. For ease of further calculation, the Near values were truncated to three 
decimal places (units in meters), and each value was multiplied by +1 for its 
corresponding positive angle value or -1 for its corresponding negative angle 
value.  The result was that each point record had an equivalent cross-track 
error from the desired reference line. 
M.5.4. Mean cross-track-error, standard deviation of error, root mean square error, 
pass-to-pass accuracy, and year-to-year accuracy were then calculated 
according to Appendix H. 
168 
 
APPENDIX N    
SUPPLEMENTAL 2006 XTE DATA
169 
 
Table N.1. Summary signal-correction quality statistics for testing event 1 
(February 22, 2006); tests R11 and W11. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 0 -2 -3 -6 -7 -4     
NR 22 22 36 39 34  153    
SDR 4 5 4 5 5   5   
RMSER 4 5 5 8 9    7  
P2PR -- -2 -1 -3 -1     -2 
XTEJ           
NJ           
SDJ           
RMSEJ           
P2PJ           
XTED           
ND           
SDD           
RMSED           
P2PD           
XTEW  5 -2 14 26 12     
NW  31 39 43 45  158    
SDW  7 10 8 16   15   
RMSEW  9 10 16 31    20  
P2PW  -- -7 16 12     7 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
 
170 
 
Table N.2. Summary signal-correction quality statistics for testing event 2 
(February 27, 2006); tests R21, D21, and W21. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -1 -2 -7 -6 -7 -5     
NR 25 33 43 54 65  220    
SDR 4 5 4 4 4   5   
RMSER 4 5 8 7 8    7  
P2PR -- -1 -5 1 -1     -2 
XTEJ           
NJ           
SDJ           
RMSEJ           
P2PJ           
XTED 0 14 25 27 20 19     
ND 30 33 41 49 65  218    
SDD 5 4 8 8 7   11   
RMSED 4 15 26 28 21    22  
P2PD -- 14 11 2 -7     5 
XTEW  -29 -43 -46 -33 -37     
NW  56 35 60 58  209    
SDW  20 13 17 29   22   
RMSEW  35 44 49 44    43  
P2PW  -- -14 -3 13     -1 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
171 
 
Table N.3. Summary signal-correction quality statistics for testing event 3 
(February 27, 2006); tests R22, D22, and W22. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -5 -7 -5 -5 -7 -6     
NR 32 36 40 57 70  235    
SDR 3 4 5 4 3   4   
RMSER 6 7 7 7 8    7  
P2PR -- -2 2 0 -2     0 
XTEJ           
NJ           
SDJ           
RMSEJ           
P2PJ           
XTED 20 35 37 20 23 26     
ND 31 39 45 59 70  244    
SDD 4 5 7 8 7   10   
RMSED 21 36 38 22 24    28  
P2PD -- 15 2 -17 4     1 
XTEW  -10 -17 -11 -26 -18     
NW  37 45 53 70  205    
SDW  16 20 31 32   28   
RMSEW  18 27 33 42    33  
P2PW  -- -7 6 -15     -5 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
 
172 
 
Table N.4. Summary signal-correction quality statistics for testing event 4 (March 8, 
2006); tests R31, D31, and W31. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 0 0 -2 -4 -4 -2     
NR 30 36 37 45 39  187    
SDR 3 3 4 5 5   4   
RMSER 3 3 4 6 6    5  
P2PR -- 0 -2 -2 0     -1 
XTEJ           
NJ           
SDJ           
RMSEJ           
P2PJ           
XTED 28 64 47 53 45 47     
ND 28 23 42 49 55  197    
SDD 6 9 7 6 8   12   
RMSED 29 64 47 54 46    49  
P2PD -- 36 -17 6 -8     4 
XTEW  84 0 2 -26 0     
NW  17 37 45 57  156    
SDW  10 15 22 33   40   
RMSEW  85 15 22 42    40  
P2PW  -- -84 2 -24     -35 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
173 
 
Table N.5. Summary signal-correction quality statistics for testing event 5 (March 8, 
2006); tests R32, D32, and W32. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -1 -2 -1 -3 -4 -3     
NR 31 35 45 46 58  215    
SDR 5 4 3 5 4   4   
RMSER 5 5 4 6 6    5  
P2PR -- -1 1 -2 -1     -1 
XTEJ           
NJ           
SDJ           
RMSEJ           
P2PJ           
XTED 14 47 43 40 16 32     
ND 31 34 41 50 57  213    
SDD 7 7 5 12 12   17   
RMSED 15 48 44 42 20    36  
P2PD -- 33 -7 -3 -24     0 
XTEW  5 -3 -2 -2 -1     
NW  37 46 44 52  179    
SDW  25 15 21 30   23   
RMSEW  25 15 20 29    23  
P2PW  -- -8 1 0     -2 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
174 
 
Table N.6. Summary signal-correction quality statistics for testing event 6 (March 
28, 2006); tests R41, D41, and W41. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -2 0 -5 -6 -5 -4     
NR 26 30 36 46 50  188    
SDR 5 4 4 3 5   5   
RMSER 5 4 6 6 7    6  
P2PR -- -2 -5 -1 1     -2 
XTEJ           
NJ           
SDJ           
RMSEJ           
P2PJ           
XTED 60 66 57 63 54 59     
ND 22 28 34 42 53  179    
SDD 3 5 6 8 7   8   
RMSED 60 67 57 63 55    60  
P2PD -- 6 -9 6 -9     -2 
XTEW  -13 -4 -10 -5 -8     
NW  32 35 40 51  158    
SDW  8 17 27 26   22   
RMSEW  15 17 28 26    23  
P2PW  -- 9 -6 5     3 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
175 
 
Table N.7. Summary signal-correction quality statistics for testing event 7 (March 
28, 2006); tests R42, D42, and W42. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 0 -1 -5 -3 -4 -3     
NR 23 29 34 41 50  177    
SDR 4 4 4 4 5   5   
RMSER 4 4 6 5 6    5  
P2PR -- -1 -4 2 -1     -1 
XTEJ           
NJ           
SDJ           
RMSEJ           
P2PJ           
XTED 20 53 39 34 24 33     
ND 25 28 35 42 52  182    
SDD 11 7 7 7 8   13   
RMSED 22 53 39 34 25    35  
P2PD -- 33 -14 -5 -10     1 
XTEW  -12 -1 1 -1 -2     
NW  23 35 41 49  148    
SDW  13 18 24 25   22   
RMSEW  17 18 24 25    22  
P2PW  -- 11 2 -2     4 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
176 
 
Table N.8. Summary signal-correction quality statistics for testing event 8 (April 7, 
2006); tests R51, J51, and D51. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -2 -7 -7 -8 -8 -7     
NR 34 34 46 56 73  243    
SDR 3 4 4 4 4   4   
RMSER 4 8 8 8 9    8  
P2PR -- -5 0 -1 0     -2 
XTEJ 0 6 -5 -7 -9 -4     
NJ 32 27 45 50 61  215    
SDJ 3 10 12 13 15   13   
RMSEJ 3 12 13 15 18    14  
P2PJ -- 6 -11 -2 -2     -2 
XTED 9 27 27 22 28 24     
ND 37 39 51 50 72  249    
SDD 4 8 6 6 7   9   
RMSED 9 28 27 23 29    25  
P2PD -- 18 0 -5 6     5 
XTEW           
NW           
SDW           
RMSEW           
P2PW           
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
177 
 
Table N.9. Summary signal-correction quality statistics for testing event 9 (April 7, 
2006); tests R52, J52, and D52. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -3 -8 -6 -7 -8 -7     
NR 28 33 39 46 61  207    
SDR 4 4 4 5 5   5   
RMSER 5 9 8 9 10    8  
P2PR -- -5 2 -1 -1     -1 
XTEJ -1 -30 -19 -23 -34 -23     
NJ 29 30 41 49 64  213    
SDJ 9 9 8 16 14   16   
RMSEJ 9 32 20 28 36    28  
P2PJ -- -29 11 -4 -11     -8 
XTED 42 44 50 50 50 48     
ND 30 35 44 51 57  217    
SDD 5 4 6 6 6   6   
RMSED 42 44 50 50 50    48  
P2PD -- 2 6 0 0     2 
XTEW           
NW           
SDW           
RMSEW           
P2PW           
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
178 
 
Table N.10. Summary signal-correction quality statistics for testing event 10 (June 
14, 2006); tests R61 and D61. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 0 -3 -4 -5 -5 -4     
NR 22 29 32 34 33  150    
SDR 5 6 4 5 6   5   
RMSER 4 7 6 7 8    7  
P2PR -- -3 -1 -1 0     -1 
XTEJ           
NJ           
SDJ           
RMSEJ           
P2PJ           
XTED 43 46 37 34 38 39     
ND 25 24 27 27 24  127    
SDD 4 8 10 9 15   11   
RMSED 43 46 39 35 41    41  
P2PD -- 3 -9 -3 4     -1 
XTEW           
NW           
SDW           
RMSEW           
P2PW           
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
179 
 
Table N.11. Summary signal-correction quality statistics for testing event 11 (June 
14, 2006); tests R62, J62, and D62. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -4 -5 -7 -7 -4 -5     
NR 19 33 36 36 42  166    
SDR 5 5 4 4 5   5   
RMSER 6 7 8 8 7    7  
P2PR -- -1 -2 0 3     0 
XTEJ -16 3 8 -2 -18 -5     
NJ 19 29 33 39 42  162    
SDJ 7 10 16 21 23   21   
RMSEJ 17 10 18 21 29    21  
P2PJ -- 19 5 -10 -16     0 
XTED 11 -20 -11 -15 -21 -13     
ND 20 28 34 36 39  157    
SDD 6 5 10 12 12   14   
RMSED 12 20 14 19 23    19  
P2PD -- -31 9 -4 -6     -8 
XTEW           
NW           
SDW           
RMSEW           
P2PW           
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
180 
 
Table N.12. Summary signal-correction quality statistics for testing event 12 (June 
16, 2006); tests R71, J71, and D71. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -6 -4 -4 -6 -8 -6     
NR 26 20 34 41 47  168    
SDR 4 4 5 4 4   4   
RMSER 7 6 6 7 9    7  
P2PR -- 2 0 -2 -2     0 
XTEJ 118 95 89 94 99 98     
NJ 18 24 32 29 50  153    
SDJ 8 7 8 7 12   12   
RMSEJ 118 95 90 94 100    98  
P2PJ -- -23 -6 5 5     -5 
XTED -- 57 71 68 65 65     
ND -- 28 31 37 50  146    
SDD -- 5 5 6 8   8   
RMSED -- 57 71 69 65    66  
P2PD -- 38 14 -3 -3     12 
XTEW           
NW           
SDW           
RMSEW           
P2PW           
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
181 
 
Table N.13. Summary signal-correction quality statistics for testing event 13 (June 
16, 2006); tests R72, J72, and D72. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -4 -5 -6 -7 -8 -6     
NR 17 29 34 41 37  158    
SDR 4 4 4 4 3   4   
RMSER 6 6 7 8 8    7  
P2PR -- -1 -1 -1 -1     -1 
XTEJ 176 159 144 135 117 140     
NJ 24 32 37 46 57  196    
SDJ 8 8 13 10 18   24   
RMSEJ 176 160 144 136 118    142  
P2PJ -- -17 -15 -9 -18     -15 
XTED 5 40 55 62 49 46     
ND 23 22 34 38 50  167    
SDD 4 5 14 14 15   22   
RMSED 6 40 56 63 52    51  
P2PD -- 35 15 7 -13     11 
XTEW           
NW           
SDW           
RMSEW           
P2PW           
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  Total N refers to sum of N
x datapoints. 
182 
 
Table N.14. Box-and-whisker plot data (cm) for the 2006 season. 
Testing 
Event [a] Min Q1 Med Q3 Max 
Testing 
Event [a] Min Q1 Med Q3 Max 
RTK      SF2      
R11 -17 -7 -4 0 8 D21 -10 11 20 27 47 
R21 -17 -8 -5 -2 6 D22 0 19 26 34 48 
R22 -16 -8 -6 -3 4 D31 17 40 48 54 79 
R31 -14 -5 -2 1 9 D32 -25 16 38 45 59 
R32 -15 -6 -2 1 10 D41 39 55 59 64 77 
R41 -16 -7 -4 -1 8 D42 4 24 33 42 67 
R42 -14 -6 -3 1 9 D51 2 18 24 30 44 
R51 -18 -10 -7 -4 3 D52 33 43 48 52 62 
R52 -19 -10 -7 -4 5 D61 9 34 41 46 59 
R61 -20 -7 -4 0 8 D62 -50 -22 -13 -6 20 
R62 -19 -8 -6 -2 6 D71 42 60 66 72 81 
R71 -18 -9 -6 -3 2 D72 -3 36 53 63 77 
R72 -16 -9 -7 -4 4 Summary -24 22 37 51 82 
Summary -18 -8 -5 -2 8       
      WAAS      
SF1      W11 -21 1 10 20 48 
J51 -43 -14 -3 6 26 W21 -91 -50 -37 -25 14 
J52 -62 -35 -23 -12 20 W22 -84 -37 -15 2 48 
J62 -64 -16 -2 11 33 W31 -95 -25 -2 16 105 
J71 73 89 97 106 128 W32 -62 -16 -4 13 57 
J72 69 127 139 158 186 W41 -55 -22 -13 7 44 
Summary -64 -15 5 101 186 W42 -51 -20 -4 14 48 
      Summary -84 -28 -9 10 57 
            
[a}  Testing events are listed in Appendix I. 
183 
 
APPENDIX O    
SUPPLEMENTAL 2007 XTE DATA
184 
 
Table O.1. Summary signal-correction quality statistics for north-to-south testing 
event 1 (March 7, 2007); tests RS11, JS11, DS11, and WS11. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 8 1 2 2 2 3     
NR 65 67 70 72 66  340    
SDR 6 5 6 8 8   7   
RMSER 10 5 6 8 8    8  
P2PR -- -7 1 0 0     -2 
XTEJ 30 33 43 52 52 42     
NJ 68 78 71 70 59  346    
SDJ 11 7 7 10 8   13   
RMSEJ 32 34 44 53 53    44  
P2PJ -- 3 10 9 0     6 
XTED 52 51 54 50 46 50     
ND 61 58 72 54 51  292    
SDD 4 4 7 9 6   6   
RMSED 52 51 54 51 47    51  
P2PD -- -1 3 -4 -4     -2 
XTEW -59 -55 -50 -33 -43 -47     
NW 47 66 70 70 65  318    
SDW 16 16 17 21 15   20   
RMSEW 62 57 53 39 46    51  
P2PW -- 4 5 17 -10     4 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
185 
 
Table O.2. Summary signal-correction quality statistics for east-to-west testing 
event 1 (March 7, 2007); tests JW11, DW11, and WW11. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR           
XTEJ 36 39 43 41 37 39     
NJ 34 34 26 30 19  143    
SDJ 7 9 12 16 13   12   
RMSEJ 37 40 45 44 39    41  
P2PJ -- 3 4 -2 -4     0 
XTED 72 66 59 54 38 60     
ND 29 31 31 27 14  132    
SDD 6 10 12 15 13   15   
RMSED 72 66 60 56 40    62  
P2PD -- -6 -7 -5 -16     -8 
XTEW 90 135 131 149 144 131     
NW 29 30 34 36 30  159    
SDW 5 13 21 24 26   28   
RMSEW 91 135 133 151 147    134  
P2PW -- 45 -4 18 -5     14 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
186 
 
Table O.3. Summary signal-correction quality statistics for east-to-west testing 
event 2 (March 7, 2007); tests JW12, DW12, and WW12. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR           
XTEJ 67 66 53 34 27 53     
NJ 36 35 31 25 19  146    
SDJ 14 11 13 21 18   21   
RMSEJ 68 67 54 39 32    57  
P2PJ -- -1 -13 -19 -7     -10 
XTED 126 106 70 50 25 82     
ND 33 38 28 22 23  144    
SDD 11 16 20 30 25   41   
RMSED 126 107 73 58 34    91  
P2PD -- -20 -36 -20 -25     -25 
XTEW 85 82 57 50 49 67     
NW 35 35 32 28 23  153    
SDW 5 8 22 15 19   22   
RMSEW 86 83 61 52 52    70  
P2PW -- -3 -25 -7 -1     -9 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
187 
 
Table O.4. Summary signal-correction quality statistics for north-to-south testing 
event 2 (March 14, 2007); tests RS21, JS21, DS21, and WS21. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 5 -2 -1 -2 -2 -1     
NR 68 71 73 73 64  349    
SDR 5 6 7 7 7   7   
RMSER 7 7 7 7 8    7  
P2PR -- -7 1 -1 0     -2 
XTEJ 35 51 73 76 71 61     
NJ 65 70 65 71 65  336    
SDJ 9 4 6 6 11   17   
RMSEJ 36 51 73 76 76    64  
P2PJ -- 16 22 3 -5     9 
XTED -20 -- -22 -10 -3 -13     
ND 57 -- 78 75 73  283    
SDD 5 -- 5 7 9   10   
RMSED 21 -- 22 12 9    17  
P2PD -- -- -2 12 7     4 
XTEW 6 -37 -8 -18 -11 -12     
NW 67 47 71 68 69  322    
SDW 23 14 18 17 20   23   
RMSEW 24 40 20 25 22    26  
P2PW -- -43 29 -10 7     -4 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
188 
 
Table O.5. Summary signal-correction quality statistics for north-to-south testing 
event 3 (March 14, 2007); tests RS22, JS22, DS22, and WS22. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 4 -2 -3 -3 -1 -1     
NR 58 69 68 59 53  307    
SDR 3 6 6 7 7   6   
RMSER 5 6 6 7 7    6  
P2PR -- -6 -1 0 2     -1 
XTEJ 29 26 77 4 -40 17     
NJ 54 66 29 67 36  252    
SDJ 7 4 5 11 14   33   
RMSEJ 30 27 78 12 42    37  
P2PJ -- -3 51 -73 -44     -17 
XTED 26 34 31 36 35 32     
ND 60 70 69 52 59  310    
SDD 4 6 8 9 9   8   
RMSED 26 34 32 37 36    33  
P2PD -- 8 -3 5 -1     2 
XTEW -- 37 12 6 -3 14     
NW -- 64 71 65 53  253    
SDW -- 10 9 15 16   19   
RMSEW -- 38 15 16 16    24  
P2PW -- -- -25 -6 -9     -13 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
189 
 
Table O.6. Summary signal-correction quality statistics for east-to-west testing 
event 3 (March 14, 2007); tests RW21, JW21, DW21, and WW21. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 16 17 12 9 9 13     
NR 21 30 31 25 23  130    
SDR 4 8 10 10 9   9   
RMSER 16 19 15 13 13    16  
P2PR -- 1 -5 -3 0     -2 
XTEJ 27 28 39 38 41 34     
NJ 29 30 32 23 24  138    
SDJ 4 10 8 18 12   12   
RMSEJ 28 30 39 42 42    36  
P2PJ -- 1 11 -1 3     4 
XTED -- 8 24 52 51 32     
ND -- 29 35 32 29  125    
SDD -- 5 6 14 27   27   
RMSED -- 9 24 53 57    41  
P2PD -- -- 16 28 -1     14 
XTEW 35 81 44 44 32 48     
NW 22 29 31 27 31  140    
SDW 21 36 31 47 30   38   
RMSEW 41 89 54 64 43    61  
P2PW -- 46 -37 0 -12     -1 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
190 
 
Table O.7. Summary signal-correction quality statistics for east-to-west testing 
event 4 (March 14, 2007); tests RW22, JW22, DW22, and WW22. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 18 17 16 14 14 16     
NR 25 37 33 26 33  154    
SDR 7 6 8 7 7   7   
RMSER 19 18 18 16 15    17  
P2PR -- -1 -1 -2 0     -1 
XTEJ 39 41 33 25 24 33     
NJ 34 30 36 29 29  158    
SDJ 7 6 8 9 8   10   
RMSEJ 40 42 34 26 25    34  
P2PJ -- 2 -8 -8 -1     -4 
XTED -206 -203 -112 -64 -50 -122     
ND 24 30 24 32 32  142    
SDD 27 6 10 14 16   69   
RMSED 208 203 113 65 52    140  
P2PD -- 3 91 48 14     39 
XTEW -62 -81 -66 -95 -31 -67     
NW 28 35 36 25 27  151    
SDW 38 51 29 30 48   45   
RMSEW 73 95 72 100 56    81  
P2PW -- -19 15 -29 64     8 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
191 
 
Table O.8. Summary signal-correction quality statistics for north-to-south testing 
event 4 (April 13, 2007); tests RS31, JS31, DS31, and WS31. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 4 -3 -1 0 -1 0     
NR 63 67 63 62 65  320    
SDR 5 6 6 6 6   6   
RMSER 6 6 7 6 6    6  
P2PR -- -7 2 1 -1     -1 
XTEJ 21 44 60 56 49 46     
NJ 68 64 62 64 67  325    
SDJ 6 4 8 10 8   16   
RMSEJ 21 44 61 57 49    48  
P2PJ -- 23 16 -4 -7     7 
XTED 8 71 57 52 40 46     
ND 56 60 74 69 73  332    
SDD 4 3 8 11 9   21   
RMSED 9 71 57 53 41    51  
P2PD -- 63 -14 -5 -12     8 
XTEW -10 -30 -21 -30 -63 -32     
NW 41 61 68 63 64  297    
SDW 18 26 21 9 24   27   
RMSEW 20 40 30 31 67    42  
P2PW -- -20 9 -9 -33     -13 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
192 
 
Table O.9. Summary signal-correction quality statistics for north-to-south testing 
event 5 (April 13, 2007); tests RS32, JS32, DS32, and WS32. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 6 -2 -1 1 -3 0     
NR 48 66 69 71 63  317    
SDR 6 6 7 7 7   7   
RMSER 8 7 7 7 8    7  
P2PR -- -8 1 2 -4     -2 
XTEJ 5 46 59 43 34 38     
NJ 63 70 78 70 72  353    
SDJ 8 5 8 12 13   20   
RMSEJ 9 46 59 45 36    43  
P2PJ -- 41 13 -16 -9     7 
XTED -2 -57 -31 -7 3 -19     
ND 70 65 70 57 59  321    
SDD 20 6 6 13 10   26   
RMSED 20 58 32 14 10    32  
P2PD -- -55 26 24 10     1 
XTEW -- 6 0 1 -9 -1     
NW -- 66 70 71 69  276    
SDW -- 14 27 12 14   19   
RMSEW -- 14 27 12 17    19  
P2PW -- -- -6 1 -10     -4 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
193 
 
Table O.10. Summary signal-correction quality statistics for east-to-west testing 
event 5 (April 13, 2007); tests RW31, JW31, DW31, and WW31. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 11 17 18 14 9 14     
NR 35 40 36 34 25  170    
SDR 7 10 9 8 9   9   
RMSER 13 19 20 16 13    17  
P2PR -- 6 1 -4 -5     0 
XTEJ 29 29 27 23 16 26     
NJ 44 39 36 35 29  183    
SDJ 10 8 9 8 11   10   
RMSEJ 31 30 29 25 20    28  
P2PJ -- 0 -2 -4 -7     -3 
XTED 46 48 36 30 25 37     
ND 39 40 41 44 35  199    
SDD 8 5 10 9 8   12   
RMSED 47 49 37 32 26    39  
P2PD -- 2 -12 -6 -5     -5 
XTEW -184 -155 -131 -79 -84 -130     
NW 36 32 35 29 28  160    
SDW 15 10 11 12 33   44   
RMSEW 184 156 132 80 90    137  
P2PW -- 29 24 52 -5     25 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
 
194 
 
Table O.11. Summary signal-correction quality statistics for east-to-west testing 
event 6 (April 13, 2007); tests RW32, JW32, DW32, and WW32. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 16 18 15 13 11 15     
NR 33 37 36 17 33  156    
SDR 7 7 7 10 9   8   
RMSER 17 19 16 17 14    17  
P2PR -- 2 -3 -2 -2     -1 
XTEJ -31 -12 -7 5 3 -8     
NJ 33 35 36 39 30  173    
SDJ 16 8 7 8 18   18   
RMSEJ 35 15 10 10 18    19  
P2PJ -- 19 5 12 -2     8 
XTED -40 -40 -43 -48 -41 -43     
ND 38 27 32 33 34  164    
SDD 8 6 8 8 10   10   
RMSED 41 41 44 49 42    44  
P2PD -- 0 -3 -5 7     0 
XTEW -68 40 31 19 -6 4     
NW 34 33 38 33 32  170    
SDW 20 29 9 23 35   46   
RMSEW 71 49 32 30 35    46  
P2PW -- 108 -9 -12 -25     16 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
195 
 
Table O.12. Summary signal-correction quality statistics for north-to-south testing 
event 6 (April 30, 2007); tests RS41, JS41, DS41, and WS41. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 5 0 -1 0 -2 1     
NR 65 68 66 62 61  322    
SDR 6 6 4 5 7   6   
RMSER 8 6 4 5 8    6  
P2PR -- -5 -1 1 -2     -2 
XTEJ 63 54 54 50 45 53     
NJ 56 66 71 71 65  329    
SDJ 6 6 6 8 7   9   
RMSEJ 63 55 54 51 45    54  
P2PJ -- -9 0 -4 -5     -4 
XTED 53 36 38 38 31 39     
ND 62 72 61 72 58  325    
SDD 6 6 4 7 6   9   
RMSED 53 37 38 38 38    40  
P2PD -- -17 2 0 -7     -6 
XTEW -8 -33 -23 -27 -15 -21     
NW 67 67 60 57 64  315    
SDW 12 12 12 9 14   15   
RMSEW 14 35 25 28 20    26  
P2PW -- -25 10 -4 12     -2 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
196 
 
Table O.13. Summary signal-correction quality statistics for east-to-west testing 
event 7 (April 30, 2007); test DW41. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR           
XTEJ           
NJ           
SDJ           
RMSEJ           
P2PJ           
XTED -12 -44 -11 -3 -26 -17     
ND 35 26 34 35 26  156    
SDD 9 22 23 25 31   26   
RMSED 15 49 25 24 40    31  
P2PD -- -32 33 8 -23     -4 
XTEW           
NW           
SDW           
RMSEW           
P2PW           
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
197 
 
Table O.14. Summary signal-correction quality statistics for north-to-south testing 
event 7 (June 5, 2007); tests RS51, JS51, DS51, and WS51. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -2 -8 -8 -7 -8 -6     
NR 55 51 64 57 59  286    
SDR 7 8 7 6 7   7   
RMSER 7 11 10 10 11    10  
P2PR -- -6 0 1 -1     -2 
XTEJ -14 -29 -26 -26 -29 -25     
NJ 55 60 58 58 50  281    
SDJ 7 7 5 6 8   9   
RMSEJ 15 30 26 26 30    26  
P2PJ -- -15 3 0 -3     -4 
XTED -27 -47 -17 -9 -15 -23     
ND 65 53 60 59 7  294    
SDD 6 4 5 8 9   15   
RMSED 28 47 18 13 17    27  
P2PD -- -20 30 8 -6     3 
XTEW -20 -44 -18 -20 -34 -27     
NW 46 51 56 55 50  258    
SDW 19 20 20 17 15   21   
RMSEW 28 48 26 26 37    34  
P2PW -- -24 26 -2 -14     -4 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
198 
 
Table O.15. Summary signal-correction quality statistics for north-to-south testing 
event 8 (June 5, 2007); tests RS52, JS52, DS52, and WS52. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 0 -13 -10 -8 -11 -9     
NR 47 45 62 52 56  262    
SDR 6 7 8 4 7   8   
RMSER 6 15 12 9 13    12  
P2PR -- -13 3 2 -3     -3 
XTEJ -66 -76 -52 -57 -69 -64     
NJ 48 65 66 61 52  292    
SDJ 6 4 8 11 7   12   
RMSEJ 67 76 53 58 69    65  
P2PJ -- -10 24 -5 -12     -1 
XTED 49 33 26 29 28 32     
ND 50 65 65 59 55  294    
SDD 7 13 10 8 11   13   
RMSED 49 36 28 30 30    35  
P2PD -- -16 -7 3 -1     -5 
XTEW 27 -5 -11 -7 2 -2     
NW 20 54 55 57 56  242    
SDW 10 15 12 15 15   17   
RMSEW 29 16 17 17 15    18  
P2PW -- -32 -6 4 9     -6 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
199 
 
Table O.16. Summary signal-correction quality statistics for east-to-west testing 
event 8 (June 5, 2007); tests RW51, JW51, DW51, and WW51. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 4 4 1 3 14 3     
NR 37 37 36 29 7  146    
SDR 8 6 9 15 21   11   
RMSER 9 7 9 15 24    11  
P2PR -- 0 -3 -2 11     2 
XTEJ 10 36 28 24 26 25     
NJ 39 40 38 32 21  170    
SDJ 10 7 8 17 22   15   
RMSEJ 14 36 29 29 34    29  
P2PJ -- 26 -8 -4 2     4 
XTED 15 35 18 16 5 18     
ND 44 30 30 29 24  157    
SDD 10 8 10 16 12   14   
RMSED 18 36 21 22 12    23  
P2PD -- 20 -17 -2 -11     -2 
XTEW 80 34 59 41 48 54     
NW 32 33 28 18 9  120    
SDW 12 9 16 25 5   23   
RMSEW 84 35 61 48 49    59  
P2PW -- -46 25 -18 7     -8 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
200 
 
Table O.17. Summary signal-correction quality statistics for east-to-west testing 
event 9 (June 5, 2007); tests JW52, DW52, and WW52. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR           
XTEJ 29 28 20 22 28 25     
NJ 34 42 40 31 30  177    
SDJ 8 8 7 20 12   12   
RMSEJ 30 29 21 30 30    28  
P2PJ -- -1 -8 2 6     0 
XTED 31 -1 -70 -53 -17 -20     
ND 38 38 36 30 26  168    
SDD 5 8 19 34 36   44   
RMSED 31 8 72 63 40    48  
P2PD -- -32 -69 17 36     -12 
XTEW -98 -54 -80 -32 -40 -64     
NW 29 23 26 14 31  123    
SDW 16 20 26 21 26   33   
RMSEW 99 57 84 38 48    72  
P2PW -- 44 -26 48 -8     58 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
201 
 
Table O.18. Summary signal-correction quality statistics for north-to-south testing 
event 9 (July 10, 2007); tests RS61, JS61, DS61, and WS61. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -2 -2 -4 -3 -3 -3     
NR 51 55 54 64 63  287    
SDR 10 6 4 7 8   7   
RMSER 10 6 6 7 9    8  
P2PR -- 0 -2 1 0     0 
XTEJ 16 -5 0 -3 -9 -1     
NJ 46 60 64 63 56  289    
SDJ 6 6 5 7 9   10   
RMSEJ 16 8 5 8 12    10  
P2PJ -- -21 5 -3 -6     -6 
XTED -15 -43 -28 -12 -8 -21     
ND 54 65 63 62 62  306    
SDD 6 5 4 6 12   15   
RMSED 16 43 28 13 14    26  
P2PD -- -28 15 16 4     2 
XTEW 12 -17 -21 -3 -7 -9     
NW 34 63 57 47 59  260    
SDW 12 11 12 17 15   17   
RMSEW 16 20 24 17 16    19  
P2PW -- -29 -4 18 -4     -5 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
202 
 
Table O.19. Summary signal-correction quality statistics for north-to-south testing 
event 10 (July 10, 2007); tests JS62, DS62, and WS62. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR           
XTEJ -- -23 -18 -16 -8 -16     
NJ -- 60 59 64 58  241    
SDJ -- 9 7 9 7   10   
RMSEJ -- 25 19 19 11    19  
P2PJ -- -- 5 0 8     4 
XTED -27 -12 -13 -4 1 -10     
ND 50 61 60 63 60  294    
SDD 9 10 7 7 7   12   
RMSED 28 15 15 8 7    16  
P2PD -- 15 -1 9 3     6 
XTEW 13 -18 -8 -22 -12 -10     
NW 52 56 63 62 62  295    
SDW 10 13 12 17 13   18   
RMSEW 16 22 16 28 18    20  
P2PW -- -31 10 -14 10     -6 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
203 
 
Table O.20. Summary signal-correction quality statistics for east-to-west testing 
event 10 (July 10, 2007); tests RW61, JW61, DW61, and WW61. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -2 -3 -4 -4 -3 -3     
NR 31 33 33 36 31  164    
SDR 5 7 9 8 7   7   
RMSER 6 8 10 8 4    8  
P2PR -- -1 -1 0 1     0 
XTEJ 51 20 15 4 6 20     
NJ 38 34 37 33 34  176    
SDJ 5 5 9 11 10   19   
RMSEJ 51 21 18 12 12    28  
P2PJ -- -31 -5 -9 2     -11 
XTED 30 19 -5 -20 -22 0     
ND 35 36 35 39 36  181    
SDD 17 6 7 12 13   24   
RMSED 34 20 9 23 25    24  
P2PD -- -11 -24 -15 -2     -13 
XTEW 65 55 24 51 37 50     
NW 36 42 20 25 27  150    
SDW 12 17 16 29 13   22   
RMSEW 66 58 28 58 40    54  
P2PW -- -10 -31 27 -14     -7 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
204 
 
Table O.21. Summary signal-correction quality statistics for east-to-west testing 
event 11 (July 10, 2007); tests JW62, DW62, and WW62. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR --          
XTEJ 27 16 27 16 15 20     
NJ 38 40 37 37 30  182    
SDJ 6 8 8 6 10   9   
RMSEJ 28 18 28 18 18    22  
P2PJ -- -11 11 -11 -1     -3 
XTED 16 13 12 5 -9 8     
ND 40 33 40 40 34  187    
SDD 6 7 9 9 6   12   
RMSED 17 14 15 10 11    14  
P2PD -- -3 -1 -7 -14     -6 
XTEW -3 21 4 15 8 9     
NW 32 34 34 23 30  153    
SDW 8 6 11 17 15   14   
RMSEW 9 22 11 23 17    17  
P2PW -- 24 -17 11 -7     3 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
205 
 
Table O.22. Summary signal-correction quality statistics for north-to-south testing 
event 11 (July 18, 2007); tests RS71, JS71, DS71, and WS71. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 4 -2 0 1 -2 0     
NR 56 61 61 58 46  282    
SDR 5 6 6 7 7   6   
RMSER 6 6 6 7 7    6  
P2PR -- -6 2 1 -3     -2 
XTEJ -47 -36 -44 -37 -31 -39     
NJ 58 65 70 65 44  302    
SDJ 8 8 6 10 9   10   
RMSEJ 48 37 44 38 32    41  
P2PJ -- 11 -8 7 6     4 
XTED -22 -25 -15 -17 -11 -19     
ND 60 69 68 67 43  307    
SDD 5 5 7 8 7   8   
RMSED 23 26 16 19 13    20  
P2PD -- -3 10 -2 6     3 
XTEW 25 9 12 15 24 16     
NW 34 61 63 60 40  258    
SDW 8 15 8 10 13   13   
RMSEW 26 18 14 18 28    20  
P2PW -- -16 3 3 9     0 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
206 
 
Table O.23. Summary signal-correction quality statistics for north-to-south testing 
event 12 (July 18, 2007); tests JS72, DS72, and WS72. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR           
XTEJ -12 -39 -49 -40 -36 -40     
NJ 12 66 68 63 42  251    
SDJ 5 4 6 7 10   10   
RMSEJ 13 39 50 40 37    41  
P2PJ -- -27 -10 9 4     -6 
XTED 9 20 5 -4 -3 6     
ND 59 65 63 60 46  293    
SDD 6 7 6 7 8   11   
RMSED 11 21 8 8 9    13  
P2PD -- 11 -15 -9 1     -3 
XTEW -- 18 -9 -6 -9 -1     
NW -- 51 56 53 37  197    
SDW -- 15 9 13 11   17   
RMSEW -- 24 12 14 14    17  
P2PW -- -- -27 3 -3     -9 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
207 
 
Table O.24. Summary signal-correction quality statistics for north-to-south testing 
event 13 (July 18, 2007); tests JS73, DS73, and WS73. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR           
XTEJ -41 -41 -44 -45 -47 -43     
NJ 62 63 64 61 37  287    
SDJ 5 11 5 9 7   8   
RMSEJ 42 42 44 46 48    44  
P2PJ -- 0 -3 -1 -2     -2 
XTED 17 12 20 26 24 20     
ND 60 62 65 61 38  286    
SDD 12 4 6 8 8   10   
RMSED 21 13 21 28 26    22  
P2PD -- -5 8 6 -2     2 
XTEW -94 -88 -76 -69 -71 -80     
NW 57 68 67 61 42  295    
SDW 15 13 14 15 14   17   
RMSEW 95 89 78 70 72    82  
P2PW -- 6 12 7 -2     6 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
208 
 
Table O.25. Summary signal-correction quality statistics for east-to-west testing 
event 12 (July 18, 2007); tests RW71, JW71, DW71, and WW71. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -3 -5 -6 -6 -13 -2     
NR 36 39 25 22 21  175    
SDR 7 9 13 13 12   14   
RMSER 7 10 14 14 17    14  
P2PR -- -2 -1 0 -7     -2 
XTEJ 32 36 32 27 28 32     
NJ 39 36 28 14 13  130    
SDJ 10 8 11 12 7   10   
RMSEJ 33 37 34 29 28    34  
P2PJ -- 4 -4 -5 1     -1 
XTED 12 16 12 7 8 12     
ND 39 41 33 21 18  152    
SDD 7 9 17 16 16   13   
RMSED 14 19 20 17 17    18  
P2PD -- 4 -4 -5 1     -1 
XTEW 78 52 38 37 38 54     
NW 32 29 21 11 15  108    
SDW 18 25 20 30 26   28   
RMSEW 81 57 43 47 46    61  
P2PW -- -26 -14 -1 1     -10 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
209 
 
Table O.26. Summary signal-correction quality statistics for east-to-west testing 
event 13 (July 18, 2007); tests JW72, DW72, and WW72. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR           
XTEJ 61 81 77 73 68 72     
NJ 35 34 32 32 33  166    
SDJ 15 6 19 17 27   19   
RMSEJ 63 81 79 75 73    74  
P2PJ -- 20 -4 -4 -5     2 
XTED 44 48 67 45 69 53     
ND 34 43 21 32 25  155    
SDD 7 8 25 13 19   17   
RMSED 45 49 71 47 72    55  
P2PD -- 4 19 -22 24     6 
XTEW 39 74 60 55 60 58     
NW 34 35 28 30 32  159    
SDW 7 18 14 20 22   21   
RMSEW 39 76 61 58 64    61  
P2PW -- 35 -14 -5 5     5 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
210 
 
Table O.27. Summary signal-correction quality statistics for east-to-west testing 
event 14 (July 18, 2007); tests JW73, DW73, and WW73. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER           
NR           
SDR           
RMSER           
P2PR           
XTEJ 49 49 56 45 37 47     
NJ 39 37 26 29 30  161    
SDJ 7 12 15 6 10   12   
RMSEJ 49 51 58 45 38    49  
P2PJ -- 0 7 -11 -8     -3 
XTED -3 -5 -9 -12 -6 -7     
ND 37 37 30 33 29  166    
SDD 8 10 12 16 17   13   
RMSED 8 11 14 20 17    15  
P2PD -- -2 -4 -3 6     -1 
XTEW 37 25 26 25 20 27     
NW 34 26 26 28 23  137    
SDW 8 14 17 16 7   14   
RMSEW 38 28 31 30 21    31  
P2PW -- -12 1 -1 -5     -4 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
211 
 
Table O.28. Summary signal-correction quality statistics for north-to-south testing 
event 14 (July 31, 2007); tests RS81, JS81, DS81, and WS81. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 6 -1 -1 2 0 1     
NR 51 55 62 60 55  283    
SDR 4 5 7 7 7   7   
RMSER 7 5 7 7 7    7  
P2PR -- -7 0 3 -2     -2 
XTEJ 52 33 40 38 41 41     
NJ 54 60 61 60 58  293    
SDJ 4 5 6 7 7   9   
RMSEJ 52 33 41 38 42    41  
P2PJ -- -19 7 -2 3     -3 
XTED 18 16 19 22 21 19     
ND 55 60 61 57 55  288    
SDD 6 4 7 6 7   6   
RMSED 19 16 20 23 22    20  
P2PD -- -2 3 3 -1     1 
XTEW 91 62 49 85 97 77     
NW 50 61 53 57 58  279    
SDW 7 6 9 11 18   21   
RMSEW 91 62 50 85 99    80  
P2PW -- -29 -13 36 12     2 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
212 
 
Table O.29. Summary signal-correction quality statistics for north-to-south testing 
event 15 (July 31, 2007); tests RS82, JS82, DS82, and WS82. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 3 -2 -2 1 0 0     
NR 53 64 65 66 64  312    
SDR 5 4 7 7 7   6   
RMSER 5 4 7 7 7    6  
P2PR -- -5 0 3 -1     -1 
XTEJ 75 59 71 73 78 71     
NJ 57 62 68 64 62  313    
SDJ 5 4 6 8 10   9   
RMSEJ 75 59 71 74 78    72  
P2PJ -- -16 12 2 5     1 
XTED -1 3 1 4 13 3     
ND 59 62 67 65 47  300    
SDD 8 5 7 8 10   9   
RMSED 8 6 7 9 17    10  
P2PD -- 4 -2 3 9     4 
XTEW 0 -36 -24 -29 -33 -24     
NW 60 59 58 66 51  294    
SDW 19 13 9 15 19   20   
RMSEW 18 39 26 32 38    31  
P2PW -- -36 12 -5 -4     -8 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
213 
 
Table O.30. Summary signal-correction quality statistics for north-to-south testing 
event 16 (July 31, 2007); tests RS83, JS83, DS83, and WS83. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 2 -4 -3 1 1 0     
NR 60 61 63 59 60  303    
SDR 5 5 5 5 6   6   
RMSER 6 6 5 5 6    6  
P2PR -- -6 1 4 0     0 
XTEJ 85 82 95 86 86 87     
NJ 59 60 51 59 60  289    
SDJ 7 10 12 12 12   11   
RMSEJ 86 83 96 87 87    88  
P2PJ -- -3 13 -9 0     0 
XTED -7 -10 3 -2 -13 -6     
ND 58 61 65 63 59  306    
SDD 8 4 7 7 8   9   
RMSED 11 11 8 7 15    11  
P2PD -- -3 13 -5 -11     -2 
XTEW 46 24 38 57 55 44     
NW 57 58 63 54 59  291    
SDW 13 5 13 14 12   17   
RMSEW 48 24 40 59 56    47  
P2PW -- -22 14 19 -2     2 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
214 
 
Table O.31. Summary signal-correction quality statistics for east-to-west testing 
event 15 (July 31, 2007); tests RW81, JW81, DW81, and WW81. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -3 -4 0 -6 -11 -5     
NR 37 37 35 40 22  171    
SDR 6 8 9 10 9   9   
RMSER 6 9 8 12 14    10  
P2PR -- -1 4 -6 -5     -2 
XTEJ 18 14 18 12 11 15     
NJ 36 37 
 
39 30 22  164    
SDJ 6 6 9 8 7   8   
RMSEJ 19 16 20 14 13    17  
P2PJ -- -4 4 -6 -1     -2 
XTED -2 10 17 9 10 9     
ND 36 37 34 38 27  172    
SDD 6 6 9 11 8   10   
RMSED 6 12 19 15 12    13  
P2PD -- 12 7 -8 1     3 
XTEW 67 88 87 102 103 89     
NW 32 31 31 30 27  151    
SDW 8 12 14 26 19   22   
RMSEW 67 89 88 106 105    92  
P2PW -- 21 -1 15 1     9 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
215 
 
Table O.32. Summary signal-correction quality statistics for east-to-west testing 
event 16 (July 31, 2007); tests RW82, JW82, DW82, and WW82. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -3 -1 -3 -8 -7 -4     
NR 39 39 39 36 22  175    
SDR 6 6 11 12 8   9   
RMSER 7 6 11 15 10    10  
P2PR -- 2 -2 -5 1     -1 
XTEJ -31 0 4 -6 -7 -7     
NJ 36 41 41 31 33  182    
SDJ 11 7 10 23 19   19   
RMSEJ 33 7 11 23 20    20  
P2PJ -- 31 4 -10 -1     6 
XTED 0 4 -3 -8 -10 -3     
ND 36 38 40 35 26  175    
SDD 8 5 10 11 8   10   
RMSED 8 7 10 14 13    10  
P2PD -- 4 -7 -5 -2     -2 
XTEW 33 113 118 113 122 98     
NW 34 34 30 28 25  151    
SDW 11 11 29 12 20   39   
RMSEW 35 113 122 114 123    105  
P2PW -- 80 5 -5 9     22 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
216 
 
Table O.33. Summary signal-correction quality statistics for north-to-south testing 
event 17 (August 3, 2007); tests RS91, JS91, DS91, and WS91. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 2 -7 -5 -7 -8 -5     
NR 54 52 58 53 41  258    
SDR 7 6 6 6 5   7   
RMSER 7 9 8 9 10    9  
P2PR -- -9 2 -2 -1     -2 
XTEJ -25 -15 3 4 20 -2     
NJ 48 43 51 46 47  235    
SDJ 7 14 16 23 19   23   
RMSEJ 26 20 17 23 27    23  
P2PJ -- 10 18 1 16     9 
XTED -2 -16 -21 -21 -33 -19     
ND 45 42 46 49 51  233    
SDD 5 19 17 19 21   20   
RMSED 5 24 27 28 39    28  
P2PD -- -14 -5 0 -12     -8 
XTEW -56 -45 -77 -58 -53 -59     
NW 20 45 54 57 55  231    
SDW 12 27 20 20 24   25   
RMSEW 57 52 80 61 58    64  
P2PW -- 11 -32 19 5     1 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
217 
 
Table O.34. Summary signal-correction quality statistics for east-to-west testing 
event 17 (August 3, 2007); tests RW91, JW91, DW91, and WW91. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 35 28 23 29 32 29     
NR 31 33 31 13 26  134    
SDR 12 13 24 38 17   20   
RMSER 37 30 32 47 36    35  
P2PR -- -7 -5 6 3     -1 
XTEJ -54 -47 -42 -41 -33 -43     
NJ 19 32 30 25 21  127    
SDJ 14 11 16 13 14   15   
RMSEJ 56 48 45 43 36    46  
P2PJ -- 7 5 1 8     5 
XTED 120 -74 44 82 17 37     
ND 27 32 37 35 25  156    
SDD 14 17 32 40 18   71   
RMSED 121 76 55 91 25    80  
P2PD -- -194 118 38 -65     -26 
XTEW 11 5 -5 35 22 13     
NW 33 27 23 20 24  127    
SDW 8 15 15 13 19   19   
RMSEW 13 15 15 37 28    22  
P2PW -- -6 -10 40 -13     3 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
218 
 
Table O.35. Summary signal-correction quality statistics for north-to-south testing 
event 18 (August 23, 2007); tests RS101, JS101, DS101, and WS101. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 3 -4 -3 -3 -6 -3     
NR 44 51 45 62 58  260    
SDR 6 6 7 6 7   7   
RMSER 6 7 7 7 9    7  
P2PR -- -7 1 0 -3     -2 
XTEJ -56 -- -51 -52 -59 -54     
NJ 55 -- 69 64 56  244    
SDJ 4 -- 6 6 9   7   
RMSEJ 56 - 51 52 60    55  
P2PJ -- -- 5 -1 -7     -1 
XTED -8 -40 -50 -42 -38 -41     
ND 13 52 58 62 60  245    
SDD 2 9 7 7 9   12   
RMSED 8 41 50 43 39    42  
P2PD -- -32 -10 8 4     -8 
XTEW 112 95 95 112 129 109     
NW 51 55 63 61 63  293    
SDW 5 6 7 13 13   16   
RMSEW 113 95 95 113 129    110  
P2PW -- -17 0 17 17     4 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
219 
 
Table O.36. Summary signal-correction quality statistics for north-to-south testing 
event 19 (August 23, 2007); tests RS102, JS102, DS102, and WS102. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 21 1 -3 -6 -10 0     
NR 55 56 67 66 61  305    
SDR 6 19 8 9 12   15   
RMSER 22 18 8 10 15    15  
P2PR -- -20 -4 -3 -4     -8 
XTEJ 41 53 44 28 20 37     
NJ 54 57 59 61 51  282    
SDJ 9 6 9 7 10   14   
RMSEJ 42 53 45 28 22    40  
P2PJ -- 12 -9 -16 -8     -5 
XTED 36 15 8 16 14 18     
ND 61 63 66 68 62  320    
SDD 7 12 7 8 10   13   
RMSED 37 19 10 18 18    22  
P2PD -- -21 -7 8 -2     -6 
XTEW 23 19 17 18 20 19     
NW 25 53 49 63 60  250    
SDW 35 19 20 11 15   19   
RMSEW 41 27 26 21 25    27  
P2PW -- -4 -2 1 2     -1 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
220 
 
Table O.37. Summary signal-correction quality statistics for north-to-south testing 
event 20 (August 23, 2007); tests RS103, JS103, DS103, and WS103. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -3 -4 -6 -6 -4 -5     
NR 46 55 57 50 46  254    
SDR 3 6 5 6 6   6   
RMSER 4 8 8 8 7    7  
P2PR -- -1 -2 0 2     0 
XTEJ -27 -20 -18 -20 -37 -24     
NJ 43 53 52 54 52  254    
SDJ 6 5 7 8 7   10   
RMSEJ 27 21 19 22 38    26  
P2PJ -- 7 2 -2 -17     -2 
XTED -24 -24 -12 -12 -12 -17     
ND 52 55 55 54 54  270    
SDD 6 2 9 9 10   10   
RMSED 25 25 14 15 16    20  
P2PD -- 0 12 0 0     3 
XTEW -133 -222 -228 -235 -209 -206     
NW 53 57 57 59 51  277    
SDW 17 12 23 24 19   42   
RMSEW 134 222 229 237 210    211  
P2PW -- -89 -6 -7 26     -19 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
221 
 
Table O.38. Summary signal-correction quality statistics for east-to-west testing 
event 18 (August 3, 2007); tests RW101, JW101, DW101, and WW101. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER -1 -5 -5 -4 -1 -3     
NR 44 33 35 31 18  161    
SDR 10 8 8 11 13   10   
RMSER 10 9 9 11 12    10  
P2PR -- -4 0 1 3     0 
XTEJ -54 -41 -42 -49 -41 -46     
NJ 40 39 38 36 33  186    
SDJ 8 9 9 11 19   13   
RMSEJ 55 42 43 50 45    47  
P2PJ -- 13 -1 -7 8     3 
XTED 9 -34 -77 -100 -86 -55     
ND 39 41 34 34 35  183    
SDD 6 10 10 17 19   42   
RMSED 11 36 77 101 88    69  
P2PD -- -43 -43 -23 14     -24 
XTEW 33 65 50 46 81 55     
NW 27 25 34 29 28  143    
SDW 15 12 16 20 9   22   
RMSEW 37 66 52 50 81    59  
P2PW -- 32 -15 -4 35     12 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
222 
 
Table O.39. Summary signal-correction quality statistics for east-to-west testing 
event 19 (August 3, 2007); tests RW102, JW102, DW102, and WW102. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 2 2 -2 -2 -6 -1     
NR 36 39 38 36 32  181    
SDR 6 6 10 11 12   10   
RMSER 6 7 10 11 14    10  
P2PR -- 0 -4 0 -4     -2 
XTEJ -- 0 -15 -20 -19 -14     
NJ -- 28 36 37 33  134    
SDJ -- 7 9 14 14   14   
RMSEJ -- 7 17 24 24    20  
P2PJ -- -- -15 -5 1     -6 
XTED -3 -20 -14 -12 -7 -11     
ND 36 40 38 38 32  184    
SDD 5 7 10 12 18   12   
RMSED 5 21 17 16 16    17  
P2PD -- -17 6 2 5     -1 
XTEW 52 81 86 77 68 73     
NW 33 28 31 38 30  160    
SDW 15 9 12 10 17   18   
RMSEW 54 82 87 78 70    75  
P2PW -- 29 5 -9 -9     4 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
223 
 
Table O.40. Summary signal-correction quality statistics for east-to-west testing 
event 20 (August 3, 2007); tests RW103, JW103, DW103, and WW103. 
Signal[a] Row 1[b] Row 2[b] Row 3[b] Row 4[b] Row 5[b] XTE[c] N[d] SD[c] RMSE[c] P2P[c] 
XTER 0 0 -2 -5 -3 -2     
NR 37 34 35 26 23  155    
SDR 7 8 16 18 13   13   
RMSER 7 8 16 19 13    13  
P2PR -- 0 -2 -3 2     -1 
XTEJ 2 -10 3 4 7 1     
NJ 35 30 35 27 20  147    
SDJ 6 12 12 18 15   13   
RMSEJ 6 15 12 18 16    14  
P2PJ -- -12 13 1 3     1 
XTED 2 16 15 11 9 11     
ND 39 37 38 22 22  158    
SDD 7 10 10 15 14   12   
RMSED 8 19 18 18 16    16  
P2PD -- 14 -1 -4 -2     2 
XTEW 35 -30 -83 -110 -134 -59     
NW 33 30 23 29 26  141    
SDW 12 12 25 25 42   67   
RMSEW 37 33 87 112 140    89  
P2PW -- -65 -53 -27 -24     -42 
[a]  Cross-track error by signal (XTE
x) refers to the correction signals evaluated during the test, 
marked by the subscript R for RTK, J for John Deere SF1, D for John Deere SF2, and W for 
WAAS.  Nx refers to the number of total-station recorded positions available from each path (row). 
[b]  Values represent average XTE
x (cm), SDx (cm), and RMSEx (cm) for this row given Nx 
datapoints.  Negative or positive XTEx values represent cross-track error to the right or left of the 
reference line, respectively.  P2Px values are not dependent upon Nx, however.  
[c]  XTE, SD, and RMSE refer to average values across all five rows with all N
x datapoints for this 
event.  P2P refers to average of four P2Px values available. 
[d]  N refers to sum of N
x datapoints. 
224 
 
Table O.41.  Box-and-whisker plot data for the 2007 season (RTK correction). 
Testing 
Event [a] Min Q1 Med Q3 Max 
Testing 
Event [a] Min Q1 Med Q3 Max 
southward      westward      
RS11 -15 -2 3 8 21 RW21 -9 6 14 19 33 
RS21 -19 -6 -1 4 18 RW22 -3 11 16 21 29 
RS22 -20 -6 0 4 11 RW31 -11 8 14 21 36 
RS31 -17 -4 0 4 12 RW32 -9 11 15 20 35 
RS32 -22 -5 0 5 21 RW51 -26 -3 4 10 34 
RS41 -18 -4 0 4 16 RW61 -20 -8 -3 2 16 
RS51 -22 -12 -6 -2 12 RW71 -39 -11 -5 3 12 
RS52 -30 -14 -9 -3 12 RW81 -29 -11 -4 2 14 
RS61 -22 -8 -2 2 17 RW82 -32 -11 -4 2 16 
RS71 -17 -4 0 5 15 RW91 -29 17 31 44 72 
RS81 -17 -3 1 6 18 RW101 -33 -10 -3 5 14 
RS82 -17 -4 0 4 15 RW102 -27 -8 1 6 19 
RS83 -14 -5 -1 4 14 RW103 -40 -8 -1 7 32 
RS91 -22 -10 -5 0 13 Summary -33 -5 4 13 42 
RS101 -21 -8 -3 1 14       
RS102 -41 -10 -2 12 38       
RS103 -18 -8 -4 -1 10       
Summary -22 -7 -1 3 19       
            
[a}  Testing events are listed in Appendix J. 
225 
 
Table O.42.  Box-and-whisker plot data for the 2007 season (SF1 correction). 
Testing 
Event [a] Min Q1 Med Q3 Max 
Testing 
Event [a] Min Q1 Med Q3 Max 
southward      westward      
JS11 12 32 43 52 73 JW11 7 32 40 47 66 
JS21 17 49 67 75 95 JW12 -9 40 58 69 90 
JS22 -52 2 23 30 86 JW21 -2 26 33 43 59 
JS31 11 39 48 58 81 JW22 4 26 34 40 53 
JS32 -9 29 44 52 87 JW31 -1 20 26 33 49 
JS41 29 47 53 58 77 JW32 -53 -17 -7 3 27 
JS51 -49 -31 -25 -19 -1 JW51 -21 14 27 36 54 
JS52 -102 -73 -66 -54 -31 JW52 -9 20 26 32 51 
JS61 -26 -7 -1 4 24 JW61 -25 8 17 29 61 
JS62 -45 -23 -16 -10 10 JW62 -6 14 21 27 42 
JS71 -62 -46 -41 -33 -15 JW71 1 28 34 39 50 
JS72 -65 -47 -40 -35 -6 JW72 20 62 78 85 100 
JS73 -64 -49 -43 -38 -23 JW73 11 40 48 55 78 
JS81 18 34 41 47 64 JW81 -7 10 15 20 34 
JS82 46 65 72 78 94 JW82 -56 -18 -2 6 21 
JS83 58 78 87 94 117 JW91 -83 -52 -43 -34 -10 
JS91 -56 -22 -5 18 50 JW101 -81 -54 -44 -37 -17 
JS101 -85 -59 -53 -49 -37 JW102 -55 -22 -14 -4 13 
JS102 0 27 38 49 70 JW103 -37 -6 2 10 29 
JS103 -50 -28 -23 -17 -4 Summary -57 1 22 38 94 
Summary -102 -31 21 52 117       
            
[a}  Testing events are listed in Appendix J. 
  
226 
 
Table O.43.  Box-and-whisker plot data for the 2007 season (SF2 correction). 
Testing 
Event [a] Min Q1 Med Q3 Max 
Testing 
Event [a] Min Q1 Med Q3 Max 
southward      westward      
DS11 34 47 51 55 66 DW11 14 54 64 72 83 
DS21 -44 -21 -15 -6 15 DW12 -24 59 87 116 145 
DS22 10 26 31 38 53 DW21 -67 13 29 55 77 
DS31 -2 36 51 63 79 DW22 -239 -200 -106 -58 -30 
DS32 -71 -36 -20 2 30 DW31 5 28 37 47 65 
DS41 19 33 38 44 61 DW32 -64 -49 -42 -38 13 
DS51 -54 -30 -21 -13 7 DW41 -87 -31 -14 4 21 
DS52 -2 23 32 41 63 DW51 -24 7 19 27 50 
DS61 -53 -32 -20 -10 14 DW52 -134 -62 -5 14 42 
DS62 -41 -18 -9 -2 21 DW61 -57 -17 -5 20 58 
DS71 -36 -25 -19 -14 3 DW62 -22 0 10 16 37 
DS72 -22 -2 6 14 34 DW71 -25 6 14 21 37 
DS73 -6 13 19 26 45 DW72 10 43 50 58 97 
DS81 2 15 19 23 34 DW73 -46 -13 -4 3 14 
DS82 -18 -2 3 9 26 DW81 -16 1 9 15 33 
DS83 -29 -12 -6 1 16 DW82 -31 -10 -2 4 23 
DS91 -78 -33 -14 -4 13 DW91 -111 -7 44 100 144 
DS101 -72 -48 -42 -36 -5 DW101 -166 -86 -70 -28 20 
DS102 -15 9 16 26 48 DW102 -46 -20 -11 -4 22 
DS103 -44 -25 -17 -10 9 DW103 -26 3 12 20 30 
Summary -78 -16 3 27 79 Summary -84 -13 7 27 95 
            
[a}  Testing events are listed in Appendix J. 
227 
 
Table O.44.  Box-and-whisker plot data for the 2007 season (WAAS correction). 
Testing 
Event [a] Min Q1 Med Q3 Max 
Testing 
Event [a] Min Q1 Med Q3 Max 
southward      westward      
WS11 -105 -61 -50 -33 10 WW11 56 112 138 151 175 
WS21 -76 -28 -13 5 38 WW12 9 54 72 85 95 
WS22 -45 3 12 27 47 WW21 -72 18 44 74 138 
WS31 -98 -49 -31 -17 35 WW22 -172 -99 -69 -38 41 
WS32 -40 -11 -4 10 73 WW31 -205 -162 -136 -91 -1 
WS41 -59 -32 -21 -10 18 WW32 -104 -22 17 35 85 
WS51 -85 -41 -28 -14 28 WW51 -8 37 52 71 98 
WS52 -54 -15 -6 10 44 WW52 -143 -96 -55 -32 -17 
WS61 -50 -22 -10 1 36 WW61 -16 36 52 64 110 
WS62 -53 -21 -12 2 29 WW62 -31 -1 10 21 38 
WS71 -13 7 16 24 51 WW71 -30 39 60 70 115 
WS72 -39 -11 -2 4 43 WW72 -5 40 62 73 103 
WS73 -111 -94 -79 -70 -32 WW73 -17 19 30 37 56 
WS81 11 59 80 92 135 WW81 37 71 92 103 147 
WS82 -77 -37 -25 -12 20 WW82 14 68 115 125 145 
WS83 -5 28 47 57 81 WW91 -30 2 12 22 52 
WS91 -119 -77 -60 -44 7 WW101 -2 41 57 70 94 
WS101 71 96 106 119 153 WW102 18 63 76 85 112 
WS102 -31 6 20 29 69 WW103 -202 -99 -74 -15 67 
WS103 -307 -236 -219 -189 -97 Summary -112 10 44 77 175 
Summary -119 -33 -10 16 102       
            
[a}  Testing events are listed in Appendix J. 
 
 
228 
 
APPENDIX P    
SDI PRODUCT SPECIFICATIONS AND LABORATORY DATA
229 
 
P.1 DRIPLINE PRODUCT SPECIFICATIONS 
 The dripline product evaluated for this study was Netafim? DripNET PC? 
pressure-compensating product (Netafim USA, 2010).  Manufacturer emitter data are 
presented in figure P.1-1.  Manufacturer product performance, as a function of flow rate 
versus operating pressure, is presented in figure P.1-2. 
Figure P.1-1. Dripline emitter data provided by manufacturer. 
 
 
Figure P.1-2. Dripline product performance provided by manufacturer. 
 
230 
 
P.2 HORIZONTAL (0% SLOPE) TESTS 
 A section of new Netafim? DripNET PC? dripline with 20 contiguous emitters 
was selected for testing.  To establish a baseline measurement, the dripline was 
conditioned for up to one hour at an operating pressure of 83 kPa; after this conditioning 
period, emitter discharge was recorded for a five-minute period.  Three replications 
occurred in this manner, and the five-minute emitter discharge values are recorded in 
Table P.2. 
Table P.2. Five-minute emitter discharge (mL) from horizontal dripline, 83 kPa test. 
 Rep 1  Rep 2  Rep 3 
 89  91  91 
 91  94  92 
 84  86  85 
 94  96  92 
 86  88  89 
 86  87  86 
 85  85  85 
 92  93  95 
 88  90  90 
 83  84  85 
 84  86  87 
 81  82  84 
 85  88  88 
 90  92  91 
 91  94  95 
 89  90  91 
 92  91  94 
 88  89  89 
 87  88  88 
 89  89  91 
Cumul. 1754  1783  1788 
Mean 88  89  89 
SD 3.4  3.6  3.4 
CV, % 3.9  4.1  3.8 
  
231 
 
P.3 COMPARISON OF SLOPE (0% VS. 6% RISE) 
 The section of new Netafim? DripNET PC? dripline with 20 contiguous emitters 
used to generate data for Table P.2 was selected for re-testing.  The dripline was oriented 
horizontally (0% slope) and conditioned for up to one hour at an operating pressure of 
117 kPa; after this conditioning period, emitter discharge was recorded for a five-minute 
period.  The dripline was then oriented on a slope (6% rise) and conditioned for up to one 
hour at an average operating pressure of 117 kPa; after this conditioning period, emitter 
discharge was recorded for a five-minute period.  One replication for each slope 
occurred, and the five-minute emitter discharge values are recorded in Table P.3. 
Table P.3. Five-minute emitter discharge (mL) from dripline on two slopes, 117 kPa 
test. 
 0% 
slope 
 6% 
slope 
 
 88  89  
 93  92  
 85  87  
 92  94  
 86  88  
 85  87  
 85  84  
 91  92  
 90  91  
 84  83  
 85  86  
 81  84  
 86  88  
 90  92  
 92  89  
 89  88  
 90  83  
 86  86  
 86  86  
 85  85  
Cumul. 1749  1754  
Mean 87  88  
SD 3.2  3.2  
CV, % 3.7  3.7  
 
232 
 
P.4 OPERATING-PRESSURE TESTS, SIX-PERCENT SLOPE RISE 
 A section of new Netafim? DripNET PC? dripline with 20 contiguous emitters 
was selected for testing.  The dripline was oriented on a slope (6% rise) and conditioned 
for at least 15 minutes at an initial minimum operating pressure of 48 kPa; after this 
conditioning period, emitter discharge was recorded for a five-minute period.  While 
discharge continued, the operating pressure was increased to the next desired level, re-
conditioned for at least 15 minutes, and another five-minute data collection occurred.  
This method was continued for the next two operating pressures chosen.  Three 
replications occurred in this manner, and the five-minute emitter discharge values are 
recorded in Tables P.4-1 and P.4-2. 
233 
 
Table P.4-1. Five-minute emitter discharge (mL) across 48-kPa and 83-kPa 
operating pressures, six-percent slope rise. 
 
48 kPa, 
R1 
48 kPa, 
R2 
48 kPa, 
R3 
83 kPa, 
R1 
83 kPa, 
R2 
83 kPa, 
R3 
 85 85 84 85 84 84 
 87 88 88 89 87 88 
 93 93 93 94 92 93 
 82 83 85 86 82 84 
 85 87 86 88 85 85 
 86 85 86 86 84 85 
 95 94 93 95 91 95 
 85 86 86 84 84 85 
 90 88 90 90 87 89 
 93 92 92 90 91 91 
 85 86 86 86 87 86 
 86 84 84 85 83 82 
 84 88 87 83 84 85 
 91 91 90 89 89 85 
 91 90 90 88 88 88 
 85 90 89 85 86 87 
 91 94 95 89 92 92 
 79 82 83 78 81 81 
 87 89 90 85 88 90 
 82 83 85 81 82 85 
Cumulative 1742 1758 1762 1736 1727 1740 
Mean 87 88 88 87 86 87 
SD 4.2 3.7 3.4 4.0 3.4 3.7 
CV, % 4.8 4.2 3.9 4.6 4.0 4.3 
234 
 
Table P.4-2. Five-minute emitter discharge (mL) across 117-kPa and 138-kPa 
operating pressures, six-percent slope rise. 
 
117 kPa, 
R1 
117 kPa, 
R2 
117 kPa, 
R3 
138 kPa,  
R1 
138 kPa, 
R2 
138 kPa, 
R3 
 83 82 82 82 82 81 
 88 87 87 89 88 88 
 93 92 91 92 91 91 
 85 84 84 85 85 85 
 86 85 85 86 85 84 
 85 85 84 84 85 84 
 95 93 93 95 93 93 
 85 85 82 84 85 84 
 90 88 88 90 88 88 
 92 90 89 91 90 90 
 88 86 85 86 85 86 
 86 82 83 85 83 82 
 84 86 84 86 86 83 
 91 90 89 90 90 89 
 90 90 89 90 90 89 
 85 88 84 86 87 85 
 88 94 93 89 93 93 
 79 81 83 83 83 83 
 85 90 91 87 80 91 
 82 86 85 83 87 87 
Cumulative 1740 1744 1731 1743 1736 1736 
Mean 87 87 87 87 87 87 
SD 4.0 3.7 3.6 3.5 3.6 3.6 
CV, % 4.6 4.2 4.1 4.0 4.1 4.2 
 
235 
 
P.5 EMITTER DISCHARGE VERSUS SDI TESTING APPARATUS ESTIMATES 
 A section of new Netafim? DripNET PC? dripline with 100 contiguous emitters 
was selected for testing.  The dripline was oriented horizontally and conditioned for at 
least 15 minutes at an initial minimum operating pressure of 48 kPa; after this 
conditioning period, emitter discharge was recorded for a five-minute period; in addition, 
the mass of the supply vessel from the SDI-testing apparatus was recorded at the 
beginning and ending of the five-minute period.  While discharge continued, the 
operating pressure was increased to the next desired level, re-conditioned for at least 15 
minutes, and another five-minute data collection occurred.  This method was continued 
for the next two operating pressures chosen.  Three replications occurred in this manner, 
and the five-minute emitter discharge values and apparatus discharge equivalent values 
are recorded in the Tables P.5-1 to P.5-12. 
236 
 
Table P.5-1. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 48-
kPa operating pressure, replication #1. 
48 kPa, Rep1           
 89 91 85 79 85 83 81 85 87 89 
 86 86 83 88 93 82 84 80 86 87 
 85 90 81 85 85 85 85 90 86 88 
 91 84 83 83 75 82 83 82 83 93 
 87 89 85 92 84 80 80 91 90 87 
 92 86 87 84 88 87 86 86 82 91 
 88 82 78 84 84 82 80 81 88 86 
 89 92 86 83 82 83 82 88 89 86 
 76 75 85 78 84 86 83 89 85 85 
 88 80 94 81 81 82 86 81 90 81 
Emitter mean 85          
Emitter SD 4.0          
Emitter CV, % 4.7          
Cumulative output 8505          
 
     Start ? finish mass, kg:  20.00 ? 11.34 = 8.66 kg; estimable to 87 mL / emitter 
Table P.5-2. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 48-
kpa  operating pressure, replication #2. 
 48 kPa, Rep2 
 87 88 85 82 . 85 81 86 89 90 
 88 87 83 90 95 77 86 83 84 86 
 82 91 . 85 87 86 86 90 86 90 
 90 85 86 83 75 83 86 88 85 92 
 84 92 82 91 86 80 . 93 87 90 
 90 87 86 84 89 87 87 85 85 90 
 90 86 80 84 83 84 80 83 88 85 
 77 94 86 82 84 87 83 88 88 92 
 83 81 82 81 85 87 84 88 83 88 
 88 80 96 83 85 84 87 82 90 80 Emitter mean 86 
         Emitter SD 3.9 
         Emitter CV, % 4.5 
         Cumulative output 8538 
          
Start ? finish mass, kg:  19.00 ? 10.40 = 8.60 kg; estimable to 86 mL / emitter 
237 
 
Table P.5-3. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 48-
kpa operating pressure, replication #3. 
48 kPa, Rep3           
 88 90 88 82 86 84 82 86 88 90 
 89 87 66 90 95 82 84 82 85 84 
 87 93 . 86 90 89 85 90 87 91 
 90 87 87 83 80 85 86 87 85 91 
 86 94 80 92 88 68 79 92 90 88 
 93 88 87 87 90 88 87 85 89 91 
 93 86 82 88 88 85 83 84 88 88 
 90 95 87 84 85 86 84 88 86 93 
 86 80 83 80 83 86 84 85 81 88 
 95 81 88 83 86 84 88 81 88 80 Emitter mean 86 
         Emitter SD 4.6 
         Emitter CV, % 5.3 
         Cumulative output 8614 
          
Start ? finish mass, kg:  23.00 ? 14.46 = 8.54 kg; estimable to 85 mL / emitter 
Table P.5-4. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 83-
kPa operating pressure, replication #1. 
83 kPa, Rep1           
 86 85 87 81 86 80 82 82 83 86 
 87 85 . 88 93 81 . 80 . 83 
 87 91 82 84 90 84 81 83 87 86 
 87 91 85 81 82 84 81 86 82 90 
 90 92 86 89 . . 75 90 88 83 
 92 86 90 86 90 88 88 85 89 88 
 88 86 81 82 82 85 83 85 88 83 
 91 94 85 84 84 85 81 88 90 91 
 82 82 85 77 81 85 83 87 90 90 
 83 82 90 82 87 84 86 82 90 79 Emitter mean 85 
         Emitter SD 3.7 
         Emitter CV, % 4.4 
         Cumulative output 8117 
          
Start ? finish mass, kg:  23.00 ? 14.26 = 8.74 kg; estimable to 87 mL / emitter 
 
238 
 
Table P.5-5. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 83-
kPa operating pressure, replication #2. 
83 kPa, Rep2           
 93 91 88 83 89 86 82 85 90 94 
 91 89 86 91 98 79 87 81 90 87 
 89 92 . 86 88 89 84 91 94 92 
 89 87 88 85 78 85 83 88 89 92 
 85 94 85 95 89 77 82 94 89 89 
 92 88 . 87 93 85 83 85 86 90 
 93 86 81 87 86 84 81 82 87 85 
 90 94 87 76 86 85 82 87 89 91 
 84 . 83 81 84 87 85 87 78 88 
 94 79 87 82 88 84 88 80 90 79 Emitter mean 87 
         Emitter SD 4.5 
         Emitter CV, % 5.1 
         Cumulative output 8424 
          
Start ? finish mass, kg:  22.00 ? 13.40 = 8.60 kg; estimable to 86 mL / emitter 
 
Table P.5-6. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 83-
kPa operating pressure, replication #3. 
83 kPa, Rep3           
 79 94 85 87 91 82 82 85 88 92 
 95 93 90 94 99 78 . 80 85 85 
 90 97 80 90 91 84 80 84 90 90 
 92 89 90 86 83 77 80 82 89 93 
 87 91 91 99 89 . 77 91 89 85 
 92 87 85 86 90 87 85 85 86 90 
 93 85 80 86 85 83 82 84 85 87 
 79 93 84 82 . 87 81 88 89 92 
 85 77 84 83 84 83 79 85 85 89 
 93 81 88 82 87 80 85 82 89 90 Emitter mean 86 
         Emitter SD 4.9 
         Emitter CV, % 5.7 
         Cumulative output 8389 
          
Start ? finish mass, kg:  22.00 ? 13.54 = 8.46 kg; estimable to 85 mL / emitter 
 
239 
 
Table P.5-7. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 
117-kPa operating pressure, replication #1. 
117 kPa, Rep1           
 98 89 90 82 88 87 83 85 92 96 
 91 87 85 91 94 78 76 81 81 88 
 84 90 82 89 86 89 83 92 88 91 
 84 87 87 77 81 89 89 90 83 93 
 76 87 89 91 85 79 81 92 90 90 
 88 83 79 87 90 86 88 87 89 93 
 96 86 81 84 . 82 82 83 89 85 
 88 91 . 78 81 90 80 91 93 96 
 . . 85 83 84 86 81 86 85 89 
 90 83 90 83 92 82 87 82 89 78 Emitter mean 86 
         Emitter SD 4.8 
         Emitter CV, % 5.5 
         Cumulative output 8297 
          
Start ? finish mass, kg:  21.00 ? 12.38 = 8.62 kg; estimable to 86 mL / emitter 
 
Table P.5-8. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 
117-kPa operating pressure, replication #2. 
117 kPa, Rep2           
 97 87 90 84 90 87 85 86 89 96 
 89 87 88 91 96 83 84 82 87 88 
 86 92 76 87 87 86 82 92 92 90 
 86 84 86 77 78 87 85 89 86 93 
 86 90 90 91 78 . 80 92 90 89 
 90 86 75 87 90 86 87 80 87 89 
 96 88 81 86 81 78 78 79 87 82 
 94 95 88 85 83 79 82 86 91 91 
 81 85 83 84 86 83 81 . 84 81 
 92 81 91 81 90 80 85 82 90 . Emitter mean 86 
         Emitter SD 4.8 
         Emitter CV, % 5.6 
         Cumulative output 8360 
          
Start ? finish mass, kg:  19.00 ? 10.38 = 8.62 kg; estimable to 86 mL / emitter 
  
240 
 
Table P.5-9. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 
117-kPa operating pressure, replication #3. 
117 kPa, Rep3           
 . 92 89 86 . 85 83 85 88 95 
 91 90 90 93 99 86 88 80 89 84 
 86 93 84 87 89 88 80 90 91 89 
 88 87 88 86 83 87 82 87 85 93 
 87 93 91 91 88 . 81 90 90 88 
 90 86 87 86 90 81 85 81 86 89 
 94 86 81 86 85 84 79 80 87 82 
 90 92 87 85 85 85 83 87 91 92 
 82 62 82 83 85 87 82 84 85 88 
 91 80 87 81 88 82 86 82 89 76 Emitter mean 86 
         Emitter SD 4.7 
         Emitter CV, % 5.5 
         Cumulative output 8379 
          
Start ? finish mass, kg:  20.02 ? 11.38 = 8.64 kg; estimable to 86 mL / emitter 
 
Table P.5-10. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 
138-kPa operating pressure, replication #1. 
138 kPa, Rep1           
 99 90 88 85 85 89 90 90 94 98 
 92 89 89 94 97 90 92 87 93 87 
 89 93 75 85 89 92 86 96 97 95 
 87 87 90 87 80 93 90 94 90 97 
 . 94 92 96 89 . 86 97 94 93 
 90 84 87 85 90 86 89 79 88 90 
 94 85 81 85 85 87 77 85 90 88 
 78 91 86 87 84 81 86 83 92 93 
 77 83 87 84 86 88 79 86 87 88 
 89 80 87 82 90 83 86 81 90 80 Emitter mean 88 
         Emitter SD 5.1 
         Emitter CV, % 5.8 
         Cumulative output 8630 
          
Start ? finish mass, kg:  23.00 ? 14.30 = 8.70 kg; estimable to 87 mL / emitter 
 
241 
 
Table P.5-11. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 
138-kPa operating pressure, replication #2. 
138 kPa, Rep2           
 . 90 92 85 92 85 85 85 90 94 
 93 91 90 81 95 85 88 77 88 84 
 89 95 85 89 89 85 75 89 92 91 
 85 89 90 86 86 85 82 87 85 91 
 84 93 92 95 89 . 77 86 88 89 
 91 81 87 87 91 84 85 87 88 88 
 96 87 82 85 84 75 83 80 90 84 
 92 86 86 86 86 87 83 85 91 90 
 85 81 86 86 87 89 81 86 87 85 
 90 80 86 82 90 83 88 82 90 80 Emitter mean 87 
         Emitter SD 4.3 
         Emitter CV, % 4.9 
         Cumulative output 8499 
          
Start ? finish mass, kg:  23.00 ? 14.36 = 8.64 kg; estimable to 86 mL / emitter 
 
Table P.5-12. Five-minute emitter discharge (mL) versus apparatus estimate (kg); 
138-kPa operating pressure, replication #3. 
138 kPa, Rep3           
 . 90 92 85 92 85 85 85 90 94 
 93 91 90 81 95 85 88 77 88 84 
 89 95 85 89 89 85 75 89 92 91 
 85 89 90 86 86 85 82 87 85 91 
 84 93 92 95 89 . 77 86 88 89 
 91 81 87 87 91 84 85 87 88 88 
 96 87 82 85 84 75 83 80 90 84 
 92 86 86 86 86 87 83 85 91 90 
 85 81 86 86 87 89 81 86 87 85 
 90 80 86 82 90 83 88 82 90 80 Emitter mean 86 
         Emitter SD 3.8 
         Emitter CV, % 4.4 
         Cumulative output 8438 
          
Start ? finish mass, kg:  21.02 ? 12.36 = 8.66 kg; estimable to 87 mL / emitter 
242 
 
Table P.5-13.Emitter exponent calculations as referenced in ASAE S553 (2008). 
     Sums 
Operating Pressure (kPa), p 48 83 117 138 386 
Mean Emitter Discharge (L h-1)a, q 1.028 1.032 1.032 1.044 4.136 
Log (p) 1.681 1.919 2.068 2.140 7.808 
Log (q) 0.012 0.014 0.014 0.019 0.058 
[Log (p)] x [Log (q)] 0.020 0.026 0.028 0.040 0.115 
[Log (p)]2 2.827 3.683 4.277 4.579 15.366 
      
a Mean emitter discharge refers to average discharge from three replications of 100 contiguous emitters. 
 
Emitter exponent = uni2211g4670g3013g3042g3034uni0020g4666g3043g4667g3400g3013g3042g3034uni0020g4666g3044g4667g4671g2879
g3117
g3289g4670uni2211g3013g3042g3034uni0020g4666g3043g4667g4671g4670uni2211g3013g3042g3034uni0020g4666g3044g4667g4671
uni2211g4670g3013g3042g3034uni0020g4666g3043g4667g4671g3118g2879g3117g3289g4670uni2211g3013g3042g3034uni0020g4666g3043g4667g4671g3118  
where n=number of pressure levels tested; therefore, emitter exponent = 
uni0030uni002Euni0031uni0031uni0035g3398uni0020g2869g2872 g3400 uni0037uni002Euni0038uni0030uni0038g3400 uni0030uni002Euni0030uni0035uni0038
uni0031uni0035uni002Euni0033uni0036uni0036g3398 g2869g2872uni0020g3400 uni0037uni002Euni0038uni0030uni0038g2870  
equals 0.011. 
243 
 
APPENDIX Q    
SDI FIELD DATA
244 
 
Q.1. TEST SITE LOCATION 
 The SDI study was conducted at the Tennessee Valley Research and Extension 
Center (TVREC), Belle Mina, Alabama.  The test plot was referred to by station 
personnel as Field 52, located in the southwest corner of the property, as is indicated in 
Figure Q.1. 
  
Figure Q.1.  Aerial imagery of Tennessee Valley Research and Extension Center 
(TVREC), Belle Mina, Alabama (2006 data; imagery courtesy of Aerial 
Photography Field Office, United States Department of Agriculture). 
 
245 
 
Q.2 ATTRIBUTES OF TEST SITE 
 Field 52 is equipped with a subsurface-drip irrigation (SDI) system, optimized for 
the seasonal growth of cotton.  Long-term interdisciplinary research concerns the overall 
lint-yield production under optimal-watering and drought-stress conditions; therefore, the 
SDI system was designed such that the watering ?zones? corresponded with the treatment 
plots within the randomized-complete block design established in 2003.  Because the 
irrigation treatments have been unaltered since 2003, the system zones that correspond to 
optimal watering treatment have been used more often than the non-supplemented zones. 
 The treatment plots are approx. 380m in length and 6m in width.  SDI laterals 
were horizontally centered in each treatment plot at 2.03-m spacing and at a depth of 
0.3m.  For each irrigation-treatment plot, the three center-most laterals were targeted for 
this study.  Treatment plots relating to supplemental irrigation usage are illustrated in 
Figure Q.2-1. 
246 
 
  
Figure Q.2-1.  Irrigation treatment assignment for Field 52. 
 Topographic data were collected with a survey-grade GPS receiver, sub-inch 
accuracy.  The data were converted to a digital elevation model (DEM) in ArcGIS V.9 
software.  From the DEM, a slope-classification map was produced for Field 52.  Three 
categories of slope classification were determined: 0 - 2%, 2 - 4%, and greater than 4%  
rise.  With the slope map and the irrigation treatment map, dripline-sampling locations 
were then derived as a 3x2 randomized complete block design with three replications (18 
total sampling locations), illustrated in Figure Q.2-2.  Sampling locations with site-
number designations are represented in Figure Q.2-3. 
  
247 
 
   
Figure Q.2-2.  Sampling locations for Field 52, based on irrigation treatment 
(system use frequency) and surface slope attributes. 
  
248 
 
  
Figure Q.2-3.  Dripline-sampling locations in Field 52. 
249 
 
Q.3 FIELD DATA 
 The following tables represent SDI test-sample discharges for each target 
operating pressure at each sampling location.  In addition, actual dripline slope is 
presented, calculated by measuring difference in elevation of each end of the 12.2-m 
portion of drip tape with a survey-grade GPS receiver with centimeter accuracy. 
Table Q.3-1.  SDI-apparatus sample-discharge values for Sites 1-4. 
    
Site 1          1.3% slope 
 48 kpa------------------19.52 kg - 17.78 kg = 1.74 kg 
 83 kpa------------------16.48 kg - 14.76 kg = 1.72 kg 
 117 kpa----------------13.58 kg - 11.88 kg = 1.70 kg 
 138 kpa----------------11.26 kg -   9.54 kg = 1.72 kg 
Site 2          1.2% slope 
 48 kpa------------------18.36 kg - 16.72 kg = 1.64 kg 
 83 kpa------------------15.80 kg - 14.06 kg = 1.74 kg 
 117 kpa----------------12.98 kg - 11.00 kg = 1.98 kg 
 138 kpa----------------10.08 kg -   7.94 kg = 2.14 kg 
Site 3          0.8% slope 
 48 kpa------------------19.76 kg - 18.04 kg = 1.72 kg 
 83 kpa------------------17.04 kg - 15.28 kg = 1.76 kg 
 117 kpa----------------14.22 kg - 12.48 kg = 1.74 kg 
 138 kpa----------------11.82 kg - 10.04 kg = 1.78 kg 
Site 4          1.7% slope 
 48 kpa------------------20.72 kg - 19.00 kg = 1.72 kg 
 83 kpa------------------17.78 kg - 16.00 kg = 1.78 kg 
 117 kpa----------------15.12 kg - 13.32 kg = 1.80 kg 
 138 kpa----------------12.68 kg - 10.88 kg = 1.80 kg 
    
 
250 
 
Table Q.3-2.  SDI-apparatus sample-discharge values for Sites 5-8. 
    
Site 5          0.3% slope 
 48 kpa------------------20.90 kg - 19.36 kg = 1.54 kg 
 83 kpa------------------18.06 kg - 16.40 kg = 1.66 kg 
 117 kpa----------------15.84 kg - 14.12 kg = 1.72 kg 
 138 kpa----------------13.82 kg - 12.08 kg = 1.74 kg 
Site 6          0.6% slope 
 48 kpa------------------19.84 kg - 18.30 kg = 1.54 kg 
 83 kpa------------------17.70 kg - 16.06 kg = 1.74 kg 
 117 kpa----------------15.16 kg - 13.44 kg = 1.72 kg 
 138 kpa----------------12.80 kg - 11.04 kg = 1.76 kg 
Site 7          0.4% slope 
 48 kpa------------------19.26 kg - 17.54 kg = 1.72 kg 
 83 kpa------------------16.50 kg - 14.80 kg = 1.70 kg 
 117 kpa----------------14.00 kg - 12.30 kg = 1.70 kg 
 138 kpa----------------11.58 kg -   9.88 kg = 1.70 kg 
Site 8          1.8% slope 
 48 kpa------------------14.94 kg - 13.36 kg = 1.58 kg 
 83 kpa------------------12.42 kg - 10.78 kg = 1.64 kg 
 117 kpa----------------10.02 kg -   8.26 kg = 1.76 kg 
 138 kpa----------------  7.20 kg -   5.46 kg = 1.74 kg 
    
 
251 
 
Table Q.3-3.  SDI-apparatus sample-discharge values for Sites 9-12. 
    
Site 9          1.9% slope 
 48 kpa------------------19.18 kg - 18.36 kg = 0.82 kg 
 83 kpa------------------17.76 kg - 16.88 kg = 0.88 kg 
 117 kpa----------------16.38 kg - 15.46 kg = 0.92 kg 
 138 kpa----------------15.16 kg - 14.20 kg = 0.96 kg 
Site 10         0.5% slope 
 48 kpa------------------18.42 kg - 16.80 kg = 1.62 kg 
 83 kpa------------------15.94 kg - 14.18 kg = 1.76 kg 
 117 kpa----------------13.34 kg - 11.56 kg = 1.78 kg 
 138 kpa----------------10.84 kg -   9.04 kg = 1.80 kg 
Site 11         0.4% slope 
 48 kpa------------------20.10 kg - 18.24 kg = 1.86 kg 
 83 kpa------------------17.46 kg - 15.56 kg = 1.90 kg 
 117 kpa----------------14.78 kg - 12.86 kg = 1.92 kg 
 138 kpa----------------12.04 kg - 10.12 kg = 1.92 kg 
Site 12         0.5% slope 
 48 kpa------------------21.04 kg - 19.58 kg = 1.46 kg 
 83 kpa------------------18.90 kg - 17.22 kg = 1.68 kg 
 117 kpa----------------16.48 kg - 14.74 kg = 1.74 kg 
 138 kpa----------------14.20 kg - 12.42 kg = 1.78 kg 
    
 
252 
 
Table Q.3-4.  SDI-apparatus sample-discharge values for Sites 13-16. 
    
Site 13         1.3% slope 
 48 kpa------------------18.36 kg - 16.70 kg = 1.66 kg 
 83 kpa------------------16.02 kg - 14.32 kg = 1.70 kg 
 117 kpa----------------13.54 kg - 11.86 kg = 1.68 kg 
 138 kpa----------------11.32 kg -   9.64 kg = 1.68 kg 
Site 14         0.7% slope 
 48 kpa------------------20.86 kg - 19.22 kg = 1.64 kg 
 83 kpa------------------18.52 kg - 16.90 kg = 1.62 kg 
 117 kpa----------------16.16 kg - 14.44 kg = 1.72 kg 
 138 kpa----------------13.78 kg - 12.08 kg = 1.70 kg 
Site 15         0.7% slope 
 48 kpa------------------17.78 kg - 16.04 kg = 1.74 kg 
 83 kpa------------------15.42 kg - 13.62 kg = 1.80 kg 
 117 kpa----------------12.96 kg - 11.22 kg = 1.74 kg 
 138 kpa----------------10.36 kg -   8.62 kg = 1.74 kg 
Site 16         0.8% slope 
 48 kpa------------------19.40 kg - 17.62 kg = 1.78 kg 
 83 kpa------------------16.74 kg - 14.86 kg = 1.88 kg 
 117 kpa----------------13.02 kg - 11.26 kg = 1.76 kg 
 138 kpa----------------10.74 kg -   8.98 kg = 1.76 kg 
    
 
253 
 
Table Q.3-5.  SDI-apparatus sample-discharge values for Sites 17 and 18. 
    
Site 17         1.1% slope 
 48 kpa------------------18.78 kg - 16.98 kg = 1.80 kg 
 83 kpa------------------16.14 kg - 14.32 kg = 1.82 kg 
 117 kpa----------------13.38 kg - 11.70 kg = 1.68 kg 
 138 kpa----------------11.20 kg -   9.38 kg = 1.82 kg 
Site 18         1.9% slope 
 48 kpa------------------19.34 kg - 17.80 kg = 1.54 kg 
 83 kpa------------------16.80 kg - 15.08 kg = 1.72 kg 
 117 kpa----------------14.16 kg - 12.42 kg = 1.74 kg 
 138 kpa----------------11.68 kg -   9.96 kg = 1.72 kg