Contrasting Annual and Summer Phosphorus Export Using a Hybrid Bayesian Watershed Model

• Published in: Water resources research 2023-01, Vol.59 (1), p.n/a
• Main Authors: Karimi, K., Miller, J. W., Sankarasubramanian, A., Obenour, D. R.
• Format: Article
• Language: English
• Subjects: Agricultural land; Agriculture; Algae; Algal blooms; Annual; Anthropogenic factors; Bayesian analysis; Bayesian modeling; Bayesian theory; Climate and land use; Climate change; Climate models; Coastal waters; Dissolved oxygen; Environmental degradation; Eutrophication; Human influences; Interannual variability; Lakes; Land use; Loading rate; Mathematical models; Modelling; Nutrient loading; Nutrient pollution; Nutrient retention; Nutrient sources; Nutrients; Oxygen; Phosphorus; Phosphorus removal; Precipitation; Probability theory; Retention; Rivers; seasonal export; Seasons; Statistical analysis; Stream pollution; Streams; Summer; Urban development; Urbanization; Water management; Water quality; Water quality management; watershed; Watersheds
• ISSN: 0043-1397; 1944-7973
• DOI: 10.1029/2022WR033088

Nutrient pollution is a widespread environmental problem that degrades water quality worldwide. Addressing this issue calls for characterizing nutrient sources and retention rates, especially in seasons when water quality problems are most severe. Hybrid (statistical‐mechanistic) watershed models have been used to quantify nutrient loading from various source categories. However, these models are generally developed for long‐term average conditions, limiting their ability to assess temporal drivers of nutrient loading. They also have not been calibrated for season‐specific estimates of loading and retention rates. To address these issues, we developed a hybrid watershed model that incorporates interannual variability in land use and precipitation as temporal drivers of phosphorus loading and transport. We calibrate the hybrid watershed model within a Bayesian hierarchical framework on both an annual and summer basis over a multi‐decadal period (1982–2017). For our study area in the North Carolina Piedmont region (USA), we find that urban lands developed before 1980 are the largest contributor of phosphorus (per unit area), especially under dry conditions. Seasonally, summer phosphorus export rates are generally found to be lower than corresponding annual rates (kg/ha/mo), while in‐stream retention is found to be elevated in summer. In addition, we find that precipitation has a substantially larger influence on phosphorus export from agricultural lands than other source types, especially in summer, and that antecedent (May) precipitation significantly influences summer phosphorus export. Overall, our approach provides a data‐driven and probabilistic line of evidence to support watershed phosphorus management across different sources and seasons. Plain Language Summary Excessive nutrients (nitrogen and phosphorus) can cause algal blooms, low dissolved oxygen, and other water quality problems for lakes, streams, and coastal waters. Thus, models that quantify the anthropogenic sources of nutrients are needed for managing water quality. Most modeling efforts focus on relatively short periods, but changes in land use and climate typically occur over decadal time scales. Here, we develop models to account for how annual and summer phosphorus loads are changing from year to year. We use a statistical (Bayesian) framework to combine prior knowledge of phosphorus loading rates with data from our study area to estimate contributions from different sources. We find th