Impacts of Landscape Characteristics on Waterquality

Landscape characteristics are believed to have a close linkage with in-stream nutrient level. Both empirical and process-based approaches have been used to model the relationship between landscape characteristics and non-point source (NPS) nutrient loading level in the literature. Empirical models are suitable to explicitly define the relationships between landscape characteristics and nutrient fluxes. Spatially-explicit, process-based modeling can be used to explore the process--pattern interactions in an ecosystem. Both approaches have their advantages and disadvantages. Empirical regression models have been criticized for their oversimplification, poor predictive power, and possible scientific flaws, while process-based models have been blamed for its complicated nature, poor applicability, and prohibitive data requirements. The purpose of this dissertation research is to explore the possibility of coupling these two modeling approaches to better understand the relationship between watershed landscape characteristics and in-stream nutrient loading. To this end, a simple, spatially-explicit, process-based model, IGED (Integrated Grid's Exporting and Delivery model), is developed to estimate annual average in-stream nutrient loading at the watershed scale. Then, the validated IGED model is used to examine the impacts of watershed landscape characteristics upon the NPS nutrient loading measured at stream outlets. Two major issues are focused: the effectiveness of riparian buffers in controlling nutrient discharges to streams and the impact of watershed size on the regression model predictive power when using proportions of landscape types as predictors. In the first focus research area, the process-based model IGED is used to hep construct empirical models that explicitly address the riparian buffer and nutrient load relationship. In the second focus research area, the process-based model is used as a heuristic tool to verify and explain some exploratory observations derived from empirical regression modeling. The different roles of these two modeling approaches suggest the need of coupling them in order to improve the understanding of the relationship between the watershed landscape characteristics and the NPS nutrient loading.