Phosphorus and Nitrogen Transport in the Binational Great Lakes Basin Estimated Using SPARROW Watershed Models

Abstract

Eutrophication problems in the Great Lakes are caused by excessive nutrient inputs (primarily phosphorus, P, and nitrogen, N) from various sources throughout its basin. In developing protection and restoration plans, it is important to know where and from what sources the nutrients originate. As part of a binational effort, Midcontinent SPARROW (SPAtially Referenced Regression On Watershed attributes) models were developed and used to estimate P and N loading from throughout the entire basin based on nutrient inputs similar to 2002; previous SPARROW models only estimated U.S. contributions. The new models have a higher resolution (~2‐km2 catchments) enabling improved descriptions of where nutrients originate and the sources at various spatial scales. The models were developed using harmonized geospatial datasets describing the stream network, nutrient sources, and environmental characteristics affecting P and N delivery. The models were calibrated using loads from sites estimated with ratio estimator and regression techniques and additional statistical approaches to reduce spatial correlation in the residuals and have all monitoring sites equally influence model development. SPARROW results, along with interlake transfers and direct atmospheric inputs, were used to quantify the entire P and N input to each lake and describe the importance of each nutrient source. Model results can be used to compare loading and yields from various tributaries and jurisdictions.

Nitrogen and Phosphorus Sources and Delivery from the Mississippi/Atchafalaya River Basin: An Update Using 2012 SPARROW Models

“…The BRE technique was shown to produce more accurate and less biased long‐term mean‐annual loads than the regression techniques previously used for less conservatively transported constituents, such as P that was believed to be underestimated (Lee et al 2016). Robertson et al (2019) also showed that using the modified BRE technique for sites without trends (some sites included in this study) improved load estimation and that the use of those loads improved SPARROW model accuracy. Robertson et al (2019) also showed that most calibration sites were in areas without point sources, and the improved (increased) P loads in these areas resulted in nonpoint sources being more important and point sources (WWTPs) being less important.…”

“…Robertson et al (2019) also showed that using the modified BRE technique for sites without trends (some sites included in this study) improved load estimation and that the use of those loads improved SPARROW model accuracy. Robertson et al (2019) also showed that most calibration sites were in areas without point sources, and the improved (increased) P loads in these areas resulted in nonpoint sources being more important and point sources (WWTPs) being less important.…”

“…To reduce the effects of a nonuniform distribution of calibration sites, including very closely spaced sites in some areas with nonindependent residuals, potentially biasing the SPARROW model, Robertson et al (2019) thinned the calibration sites by choosing only one site per 12‐digit hydrologic unit code (Seaber et al 1987). In this study, a statistical approach (Pearson correlation tests on sites within 5 km) was used to decide when to thin closely spaced calibration sites.…”

“…The use of conditioned loads can underestimate the residuals compared to when all model‐simulated loads are used and can reduce the influence the downstream sites have on the model (Qian et al 2005; Wellen et al 2014). To address the potential unequal effect caused by nested basins, a statistical algorithm was developed that computed weights for each residual dependent on the fraction of its upstream drainage area that is nested (described in Robertson and Saad 2019; Robertson et al 2019). These weights are then used in subsequent re‐estimations of the coefficients in the model using weighted nonlinear least squares regression (Schwarz et al 2006, eq.…”

Key Components and Contrasts in the Nitrogen Budget Across a U.S.‐Canadian Transboundary Watershed

“…The transboundary nature of the watershed has complicated efforts to trace N pollutant sources in air and waters and to develop effective nutrient management plans. However, construction of a transboundary total N budget allows us to integrate information from all sectors, compare different management practices across the border (Robertson et al, 2019), and link these activities to the environmental outcomes. An informal partnership formed in 2016 between scientists and stakeholders in the United States and Canada to study N budgets and sustainability for the transboundary watershed.…”