Graduation Term

Summer 2026

Degree Name

Master of Science (MS)

Department

Department of Geography, Geology, and the Environment

Committee Chair

Wondwosen M. Seyoum

Committee Member

Eric W. Peterson

Committee Member

Daniel Kpienbaareh

Abstract

Phosphorus (P) pollution in Illinois watersheds has become a significant water quality concern. Recent studies have shown increasing concentrations in streams and rivers, leading to eutrophication, which damages aquatic ecosystems and threatens human health. This study examines temporal trends in total phosphorus loads across 79 monitoring stations in Illinois from 1979 to 2020 and identifies the main environmental factors that influence P load variability. Trend analyses were conducted using the nonparametric Mann-Kendall and Sen’s slope methods. GIS-based climate, hydrology, and soil property datasets were assembled for each monitoring station. These datasets were used to build machine learning models using the Generalized Linear Model, Random Forest, and Extreme Gradient Boosting (XGB) algorithms to compare their performance and determine the importance of different variables. The SHAP (Shapely Additive Explanations) analysis offered detailed insights into each model’s behavior by measuring the direct impact of individual predictors on total phosphorus load predictions.

Trend varied across all watersheds. Some watersheds showed increasing trends, while others declined; trends at major stations were not statistically significant. Among the predictive models, XGB performed best, showing the highest accuracy (R2 = 0.975) and low error metrics, outperforming Random Forest and GLM. Variable importance and SHAP analyses consistently identified drainage area as the dominant control on phosphorus load across all watersheds, accounting for most of the model influence. Annual precipitation was the second most influential variable, indicating strong climate control on phosphorus transport. Land use, soil properties, and hydrological metrics contributed minimally to the model’s performance. Spatial patterns also showed higher phosphorus loads in agricultural regions than in urban areas.

These findings suggest that watershed features and precipitation are the main factors influencing phosphorus load across Illinois. The strong influence of precipitation suggests that climate change may intensify P transport in the future. Understanding the analysis from this study can support improved nutrient management and help devise a strategy to control phosphorus levels in Illinois watersheds.

Access Type

Thesis-Open Access

Download

Share

COinS