This Is AuburnElectronic Theses and Dissertations

DEVELOPMENT OF A PREDICTIVE MODEL FOR TASTE AND ODOR EPISODES IN REGIONAL DRINKING WATER RESERVOIRS

Date

2021-07-12

Author

Goodling, Peyton

Type of Degree

Master's Thesis

Department

Biosystems Engineering

Restriction Status

EMBARGOED

Restriction Type

Full

Date Available

07-12-2022

Abstract

Taste-and-odor episodes affect water quality in reservoirs throughout the world, and utilities including the City of Auburn Water Resources Department, Opelika Utilities, and Columbus Water Works have each identified having taste and odor issues in recent years and consider them high priority for resolution. These episodes are caused by high concentrations of odorous compounds, predominantly 2-methylisoborneal (MIB) and geosmin, in drinking water reservoirs. MIB and geosmin are volatile compounds that are produced by microorganisms, primarily cyanobacteria and actinobacteria, in natural water bodies. Though these compounds are not harmful, they produce musty odors in drinking water supplies that lead to distrust and complaints from consumers because humans are highly sensitive to these compounds. Both compounds are recalcitrant in traditional water treatment processes, thus activated carbon is typically used for advanced temporary treatment. The high cost of advanced treatment makes continuous treatment of raw water unreasonable for most facilities, leaving a short period between an episode and consumer complaints. To determine when these T&O episodes are most likely to occur, predictive models are needed for better water-quality management. We developed CART and multiple linear regression models for geosmin using R. One of the key advances of this work was the integration of geosmin synthase gene abundance which was determined by qPCR. Modeling of the data revealed the best model fits were built when the datasets had high (>30 ng/L) geosmin peaks shown with Auburn, whereas the current Opelika and Columbus datasets gave us limitations, display low- moderate peaks and variability, having models with lower predictive power. The inclusion of the qPCR data proves to be most effective at predicting the high geosmin levels. Sequencing of the qPCR products revealed Anabaena and Planktothrix as likely producers. Another component of this work was the evaluation of PCR primers for the MIB synthase gene in cyanobacteria and actinobacteria. Specific and efficient primers are needed for accurate quantification of the MIB synthase gene which can be incorporated into models, similar to what was done for geosmin. In evaluating MIB primers, we discovered good efficiency for 4 primer sets and good specificity for one of them. Results from select samples sent for sequencing helped in discovering the primary MIB producers for reservoir in our region.