This dissertation is comprised of three essays examining methods for surface water quality valuation. The first essay uses Zillow's ZTRAX property transaction database to investigate variation in hedonic price effects of water clarity on single-family houses throughout the United States. I consider five spatial scales and estimate models using different sample selection criteria and model specifications. The results indicate considerable spatial heterogeneity both within and across the four U.S. Census regions. However, I also find heterogeneity resulting from different types of investigator decisions, including sample selection and modelling choices. Thus, it is necessary to use practical knowledge to consider the limits of market areas and to investigate the robustness of estimation results to investigator choices. In the second essay, I integrate a coastal recreation demand model for Southern New England with property sales in Massachusetts and Rhode Island to estimate the impacts of water quality improvements for freshwater and saltwater resources throughout the region. While traditional hedonic studies for water quality reveal the direct effects of local water quality on housing prices, the spatial extent of the models is limited to 2 km from the waterfront. Integrating recreation demand models with hedonic models allows identification of additional indirect effects of water quality at recreation sites that are capitalized in housing markets through a recreational index (i.e., the potential recreational benefits for a neighborhood). However, the spatial extent of these models is currently untested. I compare the spatial extent of indirect and direct effects of water quality for multiple water quality inputs isolating spatial and temporal sources of variation. The results indicate heterogeneity in the direct effects of freshwater and saltwater clarity depending on whether the source of variation is spatial or temporal. Conflating spatial and temporal water quality variation can lead to anomalous conclusions on the impacts of coastal recreation on housing markets. In the third essay, I evaluate meta-regression models (MRMs) that are used by policy makers in calculating the potential benefits of improvements in environmental quality resulting from proposed regulatory actions. MRMs lay the foundation for benefit transfers applied by EPA to recent rulemakings under the Clean Water Act (CWA) in addition to other environmental contexts (e.g., health benefits under the Clean Air Act). While recent literature has emphasized the necessity that MRMs satisfy key theoretical conditions, including scope sensitivity and the adding up condition, existing MRMs that fulfil these properties require imposing structural model restrictions or removing relevant explanatory variables from the MRM at the expense of model fit and predictive performance. This study presents a method for defining MRMs that draws on characteristics of the given benefits transfer scenario and its relationship to the metadata while satisfying the adding up condition. This modelling framework outperforms the model used by EPA in model fit and predictive performance. Applying the results to a recent CWA rule, I find the model used by EPA substantially underestimates the potential benefits. / Doctor of Philosophy / This dissertation is comprised of three essays examining methods for surface water quality valuation. The first essay uses property sale models to evaluate spatial differences in the impacts of lake water quality on housing markets throughout the United States. I evaluate how investigator decisions related to spatial scale of the model, sample selection criteria, and model specification effects the observed results. I find variation in the impacts of water on property sale prices across the nation. However, the results in some regions are highly sensitive to investigator decisions, implying limitations to using a single model to represent the impacts. The second essay integrates a property sale model for water quality with a model that examines willingness to pay for coastal recreation in Southern New England. I evaluate the impacts of water quality at coastal recreation sites on property sales in Massachusetts and Rhode Island and compare the results to the impacts of local water quality. I find the effects of local water quality is largest on the coast but only capitalized into property sales up to 2 km from the waterfront. I do not find expected results for coastal recreation. The third essay evaluates the models used by policy makers to calculate the potential benefits of improvements in water quality resulting from regulatory action. Previous models prioritized theoretical properties of models, such as the adding up condition, at the expense of model fit and predictive performance. I develop a hybrid model approach that can be used model water quality benefits while satisfying the adding up condition with improved model performance. I apply the model to a recently proposed rulemaking and find the benefits potential benefits greatly exceed those presented in the rule making.

Stormwater management in urban areas is a major concern today. The problem of disposing of this stormwater runoff in a satisfactory manner is very difficult indeed. Both the quantity and quality aspects of the runoff must be dealt with to obtain a solution of this problem. The water quality of the runoff can vary depending upon the different land uses of the drainage basin. The quantity of the stormwater runoff also depends upon the land uses, the rainfall intensity and duration of the storm. The traditional methods available for determining the quantity of the stormwater runoff are numerous. These traditional methods and recently developed mathematical simulation models are discussed in this paper. Prediction of the water quality of stormwater runoff is in its infancy. Several of the mathematical models have the capabilities of quality simulation, however, the simulation results are usually inconsistent with actual quality data. Of the simulation models currently in use, the EPA Storm Water Management Model is one of the most comprehensive models. Application and verification of these newly developed models is limited. The EPA Model was chosen to simulate the quantity and quality of a small urban drainage area. The study area chosen was an urban commercial section of the Lake Eola drainage basin. Physical data of the study area, such as ground slopes, storm sewer sizes and locations and slopes were determined. This data was then utilized for simulations of actual rainfall events. Verification of the quantity and quality simulation results was performed with actual quantity and quality data obtained for these rainfall events. Quantity simulation was considered successful with good correlation between the simulated and actual runoff. Quality simulation was successful to a lesser degree, the conclusion being that further quality calibration of the Model was required. Correlation between actual and simulated stormwater quality was achieved to some extent. The lack of correlation was felt to be due to calibration of the Model.

The purpose of this study was to evaluate the performance of alum alone, alum in conjunction with a polymeric coagulant aid, and a primary coagulant polymer in the coagulation-flocculation-sedimentation portion of a water treatment process and to study the implications of the results with respect to future use in direct filtration. The performance of the coagulant, coagulant aid, and primary coagulant polymer was evaluated by altering the flocculation and sedimentation retention times in the pretreatIDEnt process chain. Among the water quality parameters monitored were turbidity, pH, total alkalinity, total hardness, and total solids. A brief literature review for direct filtration and coagulation-flocculation theory was also conducted. The results of this investigation indicated that reduced retention times in the pretreatment process chain treated the selected raw waters sufficiently to enable consideration of direct filtration a viable treatment alternative. Analysis of data obtained indicated the use of polyelectrolyte as a primary coagulant or coagulant aid produced a pinpoint floc which is essential to successful utilization of direct filtration and that the floc formed settled more rapidly than floc from use of aluminum sulfate alone. The use of a polyelectrolyte as a primary coagulant or coagulant aid also reduced the chemical dosages required to provide appropriate turbidity removals. / Master of Science

Nonpoint source pollution from cropland has been identified as the primary source of nitrogen and sediment, and a significant source of phosphorus in the Chesapeake Bay. These pollutants, whether from point or nonpoint sources, have been found to be the primary cause of declining water quality in the Bay. Numerous studies have indicated that, for many watersheds, a few critical areas are responsible for a disproportionate amount of the nutrient and sediment yield. Consequently, if pollution control activities are concentrated in these critical areas, then a far greater improvement in downstream water quality can be expected with limited funds. In this research a nitrogen transport model is incorporated into ANSWERS, a distributed parameter watershed model. The nitrogen model simulates nitrogen transformations of applied fertilizer and soil nitrogen in the soil. Dissolved nitrogen transport in surface runoff is modeled by assuming complete mixing of the soil surface layer and surface runoff. Sediment-bound nitrogen transport is modeled as a function of the clay content of transported sediment. The extended ANSWERS model was verified using water quality data from rainfall simulator plot studies conducted on the Prices Fork Research Farm in Blacksburg, Virginia. The four plots were 5.5 m wide by 18.3 m long with average slopes ranging from 6.2 to 11 percent. Two of the plots were tilled conventionally, and two were no-till. Simulated rainfall at an intensity of 5 cmlh was applied to the plots and runoff samples were analyzed for sediment and nitrogen. The model was then verified by comparing the simulated response with the observed data. The model predicted sediment-bound nitrogen losses within a factor of two. The model tended to overpredict dissolved nitrogen losses by a factor of five. The model shows potential as a best management practice planning tool, however, further verification of model predictions versus observed data is required. / Master of Science

This study compared physical characteristics used when selecting comparison (healthy) watersheds for the All-Forested Load Multiplier (AllForX) Approach, and examined a quantitative watershed characteristic as a selection criterion. The AllForX Approach uses a regression relationship between Virginia Stream Condition Index (VSCI) scores and AllForX values (a unit-less multiplier that is the ratio of a modeled existing sediment load divided by a modeled all-forested load condition) for an impaired watershed and several comparison watersheds to develop sediment TMDL target loads. The Generalized Watershed Loading Function (GWLF) model was used to simulate sediment loads for 20 watersheds (four impaired and 16 comparison) in the Upper James and New River basins in Virginia's Ridge and Valley physiographic region. Results suggest that within Virginia's Ridge and Valley physiographic region it may be possible to select comparison watersheds that are of a different stream order (watershed size) and lie in different river basins from the impaired watershed. Results further indicated that the topographic index (TI) distributions were not different across the modeled watersheds, indicating the watersheds are hydrologically similar. These results support selecting comparison watersheds regardless of river basin or stream order within Virginia's Ridge and Valley physiographic region. Finally, there was no statistical difference between the AllForX regressions when using the entire period of record or the two most recent VSCI data points. Therefore, for the watersheds modeled for this study, either all of the VSCI samples or the two most recent may be used in the AllForX Approach. / Master of Science

The Virginia Department of Environmental Quality (VADEQ) under the direction of the United States Environmental Protection Agency (USEPA) has listed the lower Appomattox River as impaired because it violates current water quality standards for fecal coliforms. To advance the analytical process by which various scenarios for improving water quality within the estuary are examined, an array of computer-based hydrodynamic and water quality models were investigated. The Dynamic Estuary Model (DYNHYD5), developed by USEPA, was used to simulate hydrodynamics within the lower Appomattox River. The Water Quality Analysis Simulation Program (WASP6.1), also developed by USEPA, was employed to perform water quality simulations of fecal coliforms. Also, a detailed literature review examined DYNHYD5 and WASP6.1 model theory, computer-based model solution techniques, and background hydrodynamic theory. DYNHYD5 sensitivity analysis showed that the model was most responsive to tidal heights (seaward boundary conditions) both upstream and downstream within the model network. Specific model parameters were varied during calibration until modeled water surface elevations converged on observed water surface elevations. A goodness-of-fit value of 0.749 was determined with linear regression analysis for model calibration. DYNHYD5 input parameter validation was performed with additional observations and a goodness-of-fit value of 0.829 was calculated. Through sensitivity analysis, WASP6.1 proved to be most responsive to coliform loading rates in the downstream direction and boundary concentrations in the upstream direction. With these results, WASP6.1 input parameters were calibrated against observed fecal coliform concentrations. A goodness-of-fit value of 0.573 was determined with linear regression analysis for model calibration. WASP6.1 input parameter validation was performed with additional observations and a goodness-of-fit value of 0.0002 was calculated. Model results suggest that hydrodynamic model calibration and validation can be improved with additional tidal height observations at the downstream seaward boundary. Similarly, water quality model calibration and validation can possibly be improved with the aid of detailed, time-variable coliform concentrations at the downstream seaward boundary. Therefore, it is recommended that a water quality sampling station and tidal stage recorder be installed at the confluence of the Appomattox and James Rivers to provide for further testing of estuary hydrodynamic and water quality models. / Master of Science

As one of the largest and most productive estuaries in the United States, the Chesapeake Bay is a great economic, ecological, and cultural asset to the Mid-Atlantic region. Excess nitrogen and phosphorus discharge, however, has contributed to reduced levels of dissolved oxygen in various locations throughout the Bay. In 2010, the EPA developed a Total Maximum Daily Load (TMDL) for the entire watershed that established nutrient reduction targets to achieve Bay water quality objectives. The TMDL required states in the Chesapeake Bay watershed to create implementation plans to meet nutrient reductions. Maryland and Virginia specifically established stringent point source regulatory policies designed to meet point source reduction targets. Point source control programs created financial incentives for reducing nutrient discharge beyond regulatory requirements. This thesis will examine the extent to which Maryland and Virginia wastewater treatment plants undertake operational improvements to increase nutrient removal in response to state program incentives. Through quantitative and qualitative analysis, this thesis found evidence of lowered nitrogen discharges in response to financial incentives presented by each states point source control programs at municipal wastewater treatment plants. Maryland achieves modest improvements at a subset of advanced treatment WWTPs as a result of financial incentives presented by the state's public subsidy program. Although Virginia advanced treatment plants operating within a nutrient trading program have little incentive to over-comply, there is some evidence of operational improvements at less advanced nitrogen removal plants / Master of Science

Pathogens are the most common cause of water quality impairment in Virginia. Bacteria TMDLs (Total Maximum Daily Loads) for watersheds are typically created using a modeling approach. These models require characterization of all residential, agricultural, and wildlife sources of bacteria. Wildlife bacteria source characterization is typically conducted with estimates of population and fecal coliform production. A sensitivity analysis was performed on the bacteria TMDL development process and the HSPF (Hydrological Simulation Program-FORTRAN) model to determine how wildlife population and fecal coliform density variability impacts simulated in-stream bacteria loads. The population and fecal coliform density values for seven wildlife species were sequentially varied and run through the TMDL model to analyze the changes in bacteria loads. For population density, high, median, and low values were tested, and for fecal coliform density, high and low values were tested. The analysis was conducted on three watersheds (Abrams Creek, Upper Opequon Creek, and Happy Creek), each with a different dominant land use. The results revealed that all watersheds were sensitive to the high fecal coliform densities of deer, muskrats, and raccoons. However, Happy Creek, the watershed with majority forested land use, was additionally sensitive to the high fecal coliform densities of ducks and the high population density for all species. Using the three watersheds as surrogates for comparing different land uses, the study showed the TMDL modeling process is most sensitive to changes in wildlife in watersheds dominated by forested land use. The results also demonstrated that TMDL calibration is more efficient when adjusting wildlife fecal coliform density rather than population density to match the modeled watershed with the observed water quality data. / Master of Science

Thousands of communities across the nation are exposed to health risks from contaminated drinking water. Upstream anthropogenic land covers have been linked with the degradation of source drinking water quality and likely pose a threat to a community water system's (CWS's) ability to provide safe drinking water. The goal of this study was to predict the differences in compliance with the Safe Drinking Water Act (SDWA) between CWSs based on their upstream land cover, economic situation, and system characteristics. In Central Appalachia, from 2001 to 2016, proportions of land cover in each target CWS's upstream source water watershed were weighted based on their distance to a CWS's source water intake. Violations to the SDWA at respective CWSs over the same period were modeled with their distance weighted land cover proportions, economic status of the county served, and system characteristics as covariates. The major findings were that increases in low intensity development increased the likelihood of a health-based violation, larger CWSs were less likely than smaller CWSs to obtain a monitoring and reporting violation, and CWSs that distributed purchased water were the least likely to incur either violation type. These results suggest that communities that have CWSs that are repeatedly failing to remain in compliance with the SDWA may be able to reduce public health risks associated with drinking water by purchasing from a larger CWS. Further to protect public health, community managers should consider source water protection and/or upgrading a CWS's treatment capacity prior to developing a previously undeveloped area. / Master of Science / Millions of people across the nation face health risks from contaminated drinking water. Understanding what factors influence a community water system's ability to supply safe drinking water is critical in the effort to protect public health. Land cover altered by humans has been found to pollute drinking water sources and may be linked to unsafe drinking water. This study aims to predict the differences in compliance with the Safe Drinking Water Act (SDWA) between community water systems (CWSs) based on their upstream land cover, economic situation, and system characteristics. In Central Appalachia, proportions of land cover between 2001 and 2016 were calculated for each target CWS's upstream source water watershed. Violations to the SDWA were used in a statistical analysis with land cover, economic status of the county served, and system characteristics of respective CWSs. The major findings were that increases in low intensity developed area increased the likelihood of health-based (HB) violations, larger CWSs were more likely than smaller CWSs to monitor and report their water quality, and CWSs that served purchased water were the least likely to have a HB or monitoring and reporting violation. These results suggest that purchasing drinking water from a larger CWS may allow water providers to reduce the risk to public health from unsafe drinking water. Additionally, protecting drinking source water and/or upgrading a CWS's treatment ability prior to developing a previously undeveloped area may reduce threats to drinking water safety.

A water sample concentrator utilizing a membrane adsorption/elutriation procedure for bacterial recovery was developed and field-tested. Influent sample volumes of 19.0 liters were processed by the concentrator within 10 minutes. Standard MPN and membrane filtration tests were conducted to determine the efficiency of the concentrator. In laboratory experiments, an average bacterial recovery of 48 percent was accomplished when E. Coli was the bacterium concentrated. An average recovery of 78 percent was achieved with Streptococcus faecalis. Various commercially available 142 mm diameter membrane filters were tested with the concentrator. Millipore membrane filters having a porosity of 0.45 m were found to be superior to Cox membrane filters with 0.45 and 5.0 m porosities and to zeta plus filters. The most promising technique for eluting bacteria from the membrane filter appeared to consist of mixing the filter and 125 mL of three percent bovine serum solution at pH 9.0 in an electric blender for 30 seconds and then allowing 15 minutes of contact. Miscellaneous experiments investigated the addition of multivalent cations to enhance adsorption of bacteria to the membrane filters, the use of vacuum induced flow through the concentrator and the possibility of monitoring bacterial levels through adenosine triphosphate analysis. In the field trials, the water sample concentrator detected coliform bacteria in drinking water in which no coliforms were detected in grab samples by the standard MPN method. / Master of Science