Quantifying the Role of Climate and Watershed Characteristics on Surface Water Quality in Southeast Atlantic region of the US

Abstract

Surface water quality is controlled by multiple variables, including climatic, hydrological, and anthropogenic variables. These variables can operate at various temporal and spatial scales (e.g., local to regional watersheds). Therefore, accurate modeling of surface water quality is often complicated due to the multitude of pollutant sources and the complex evolution and interaction between hydrologic and landscape characteristics. This challenge has become formidable, particularly with the rapid increase in urbanization and agricultural activities. Effective conservation practices can significantly improve the surface water quality. However, employing these practices at appropriate locations within a watershed requires a clear understanding of how and why water quality constituents differ in time and space and which variables significantly contribute to water quality degradation. Additionally, it remains unclear how and to what degree different precipitation datasets and watershed characteristics affect the long-term average water quality within watersheds, especially across large areas. Furthermore, there is a lack of understanding of whether increasing the complexity of watershed characteristics (e.g., accounting for biogeochemical hotspots and the distance of pollutant sources from the streams) can enhance the modeling performance. This study attempts to address these questions and to improve the understanding of the potential influence of climate and watershed characteristics on surface water quality.

Specifically, the overall objectives of this study are: (a) to quantify the precipitation uncertainty in streamflow and water quality simulations by using different high-resolution precipitation products in a hydrological model (i.e., SWAT); (b) to quantify the relationships between watershed characteristics (i.e., topography, land use/cover, soils, and climate) and the mean water quality in streams across several watersheds; (c) to develop machine learning algorithms to spatially predict stream water quality based on a set of climate and watershed characteristics; and (d) to reassess the relationship between landscape/land use alterations (e.g., urbanization) and the long-term stream water quality using different landscape/land use representation metrics. This study uses a combination of hydrological and statistical modeling frameworks to quantify and forecast stream water-quality responses regarding changes in land use, climate, soil, and land management.