Updating Soil Hydraulic Properties under Changing Land Use/Cover for Improved Hydrologic Predictions

Abstract

Land use/cover (LULC) change, especially forest to urban transition, can alter the soil hydraulic properties, including soil hydraulic conductivity, even though the soil texture and series may remain the same. Soil hydraulic properties have a big influence on hydrologic processes. Watershed models are commonly used to project the potential alterations in the hydrologic regime of streams in response to ongoing or expected LULC changes. Soil related hydrologic parameters (such as hydraulic conductivity) required by these models are typically derived from soil databases. Unfortunately, when LULC changes, these soil parameters are often retained at their existing values. This is because of the lack of knowledge in quantifying changes in values of these parameters under different LULC conditions. Analyzing these soil parameters either in the field or in the laboratory is time consuming and costly. Further, scaling up from such small scales is not easy. Alternatively, pedotransfer functions, which are algorithms that describe soil-water relationships based on basic soil properties, can be used to analyze existing databases of measured soil hydraulic data. Soil hydraulic properties are seldom investigated directly under LULC changes; however some information on changes in bulk density is available. Changes in bulk density can be used as an input parameter for pedotransfer functions to derive changes in soil hydraulic conductivity to be used in watershed modeling. In practice, these functions often prove to be good predictors for updating soil hydraulic properties. This study aims to overcome this challenge using pedotransfer functions for updating soil hydraulic parameters under changing LULC by making use of soil maps in conjunction with historic aerial photos. The methodology was tested in two watersheds in Northwest Georgia with the Soil and Water Assessment Tool (SWAT). Both watersheds have seen significant urbanization (formerly forest dominated) over the past two decades. Sensitivity analysis revealed that curve number and soil properties were the most sensitive parameters on flow generation. The model performance was evaluated by defining two periods which are describes as reference and testing periods. The results showed that changes in LULC and its alteration to soil properties affect model performances. Overall discharge simulations of the watersheds were similar, but improvement was observed in high flows when changed soil parameters were used.