Attributing Sources of Surface Water Pollutants in the Maumee River Basin Using Network Modeling
Abstract
Despite broad public attention, surface water nutrient pollution from the Maumee River basin continues to be the largest contributor to the eutrophication of Lake Erie. A major impediment to nutrient management policy is that pollution sources are highly scattered across a complex stream network. These sources include inorganic fertilizers and manures applied to crop fields, runoff from intensive livestock farms, eroded soil, and wastewater treatment plants. While prior work has engaged in basin-scale attributions of pollutant sources, source attribution at finer spatial and temporal scales remains a major challenge, particularly for environmental policy and enforcement. This study addresses this challenge by developing a novel network modeling approach for subbasin-scale nutrient source attribution. We construct the network model for streamflow nutrient transport using USGS flowline data ( NHDPlus) and streamflow measurements. The nodes of the network represent sources, monitoring sites, and tributary junctions, and the edges represent stream segments. This model simulates the evolution of pollutant concentrations, given the contributions of source nodes. High-frequency (daily) water quality measurements at both mainstream and tributaries of the river network provided data for different forms of phosphorus (P) and nitrogen (N). We solve for the contributions of source nodes with Approximate Bayesian Computation, which generates the posterior distribution of source contributions based on prior distributions. We define the prior distributions of source nodes based on existing research on source types, as well as physical processes triggering runoff and soil erosion, such as rainfall and snow melting. We first conduct source attribution using Approximate Bayesian Computation in a synthetic stream network. The results demonstrate that our model can correctly and efficiently attribute pollutant sources. When applied to the Maumee River Basin, our model can attribute sources at HUC-10-scale. Our preliminary results suggest that the contributions of different regions to the overall pollution significantly varies in time and pollutant types. Further analysis of facilities and croplands associated with source nodes enables source attributions at a finer spatial scale.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFM.H14A..05W