Assessing Controls on Nutrient Loading at the Watershed Scale through Data-Driven Modeling

No Thumbnail Available




Journal Title

Series/Report No.

UNC-WRRI;496;WRRI Project ; 20-07-W

Journal ISSN

Volume Title




Anthropogenic nutrient loading is a critical driver of water quality throughout North Carolina and much of the world. Nutrient (nitrogen and phosphorus) loading has increased over the last century due to fertilization of crops and green spaces, as well as waste from humans, pets, and livestock. The most salient outcome of nutrient loading is eutrophication of lakes and coastal waters, often leading to harmful algal blooms and hypoxia, which jeopardize water supplies, wildlife habitats, and public recreation. While sources of nutrients have been identified and many control measures have been proposed, there remains a need to quantitatively assess these sources and controls, particularly at the watershed scale. In this project, we develop a modern, data-driven approach to characterizing nutrient sources and control strategies, using the Falls and Jordan Lake watersheds of North Carolina as our study area. The approach leverages large databases of water quality, hydro-meteorology, and watershed attributes, which have been developed by federal, state, and local governments over the last few decades. The approach also advances a “hybrid” watershed model that integrates a mechanistic representation of nutrient fate and transport within a Bayesian framework, so that prior knowledge of loading and transport rates is updated through data-driven inference with quantified uncertainties. As an integral part of this effort, we develop a comprehensive geospatial database on watershed development, buffers, and stormwater management regions. We then use the Bayesian hybrid modeling approach to assess nitrogen and phosphorus export from lands with different forms of development and stormwater management. Results generally support the hypothesis that stormwater management has been effective. Streamside vegetated buffers are associated with approximately 36% and 37% reductions in TN and TP, respectively; and post-construction SCMs are associated with 43% and 52% reductions in TN and TP, respectively. Finally, we apply the model to assess hypothetical urban growth scenarios. These future scenarios indicate that stormwater management (SCMs and buffers) will substantially reduce (but not fully compensate for) the impacts of continuing urban development.