Improving Watershed-Scale Understanding of Land Use/Cover Impacts on Hydrologic and Biogeochemical Dynamics through Process-based Modeling

Abstract

Understanding the impacts of land use and land cover (LULC) changes on hydrologic and biogeochemical processes is essential for sustainable watershed management, particularly in ecologically sensitive and climatically dynamic regions like the eastern United States (U.S.). Different LULC types, such as forests, urban areas, wetlands, and croplands, exhibit distinct hydrologic and biogeochemical behaviors, making it imperative to improve process-based models to capture their influence accurately. This dissertation advances watershed-scale hydrologic and biogeochemical modeling by enhancing the representation of intra-watershed processes in the Soil and Water Assessment Tool (SWAT), with a focus on improving model accuracy across diverse LULC systems through targeted modifications in process representation, calibration techniques, and spatial data integration. The research comprises four interrelated studies. The first study evaluates the applicability of the SWAT-Carbon (SWAT-C) model in simulating both terrestrial and aquatic carbon fluxes in a forested watershed that serves as a drinking water source for the City of Mobile City in the Southeast U.S. Using remote sensing and in-situ datasets, the model was calibrated to capture dominant pathways of dissolved organic carbon (DOC) transport. The results showed that extensive forest areas could be a major source of DOC, posing a significant risk to drinking water quality. Additionally, the study evaluated three management scenarios, including forest conversion, forest litter raking, and increased crop residue removal, for their potential to reduce DOC exports. This chapter highlighted the strong influence of forests and their management strategies on DOC dynamics and demonstrated the capability of the SWAT-C model in simulating vertical and lateral carbon fluxes within forested watersheds. The second study introduces National Land Cover Dataset-Imperviousness (NLCD-Imp), a Quantum Geographical Information System (QGIS)-based plugin developed to enhance LULC maps by integrating detailed impervious fractions, directly connected impervious areas (DCIAs), and tree canopy cover (TCC). Applied to the Noonday Creek watershed in metropolitan Atlanta, Georgia, this tool demonstrated that default SWAT urban classifications in SWAT may introduce uncertainty in representing urban characteristics, and in simulating hydrology, and water quality predictions. The SWAT simulations using NLCD-Imp showed increased surface runoff, decreased evapotranspiration, and up to 2 to 4 times higher nitrogen and phosphorus loads in highly urbanized areas. By improving urban LULC representation in SWAT, NLCD-Imp reduces uncertainties in hydrological and water quality simulations, facilitating more precise urban water resource management. The freely available NLCD-Imp plugin, with slight modifications, can also be utilized with any watershed-scale model that utilizes the NLCD LULC map. The third study developed and evaluated SWAT-WetQual, a coupled modeling framework that integrates SWAT with the process-based Wetland Watet Quality Model (WetQual) model to simulate detailed wetland biogeochemistry processes. The SWAT-WetQual model, validated in the Greensboro watershed, a wetland-rich basin in the mid-Atlantic U.S. in the states of Maryland, captured complex sediment, nutrient, and carbon cycling processes beyond the capabilities of SWAT’s default wetland module. Comparative scenario analyses revealed that wetlands reduced sediment (12%), nitrate (68%), total nitrogen (54%), and total organic carbon (24%) loads. Seasonal analyses further underscored the critical role of wetlands in mitigating pollutant exports during high-flow and agricultural activity periods. The newly developed SWAT-WetQual model offers a valuable tool for researchers, watershed planners, and decision-makers to better understand the cumulative effects of wetlands on watershed-scale water quality. The fourth study applies the SWAT-WetQual framework alongside NLCD-Imp enhanced LULC maps to assess the long-term impacts of LULC change on hydrology and water quality trends in the St. Andrew Bay watershed, Florida. A shift toward wetter conditions and increased water quality loads over time were observed, particularly in subbasins experiencing forest-to-rangeland conversions. The model demonstrated that wetlands consistently mitigated pollutant exports, reducing nitrogen and total organic carbon (TOC) loads by 13–18% compared to scenarios without wetlands. These findings emphasize the importance of wetlands in buffering the effects of LULC change despite notable hydrologic biogeochemical responses to such changes. Collectively, this dissertation demonstrates that improving intra-watershed process representation, particularly forest carbon cycling, urban impervious surface dynamics, and wetland biogeochemistry, enhances SWAT's predictive capacity across complex landscapes. Beyond methodological advancements, the research provides actionable insights for reducing nutrient and organic carbon loading and informs targeted watershed management strategies, underscoring the importance of spatially nuanced modeling for resilient water resource planning.