Assessing the impact of rainfall, topography, and human disturbances on nutrient levels using integrated machine...

Nutrient pollution caused by excessive total nitrogen (TN) and total phosphorus (TP) is a significant environmental challenge globally, threatening water quality and ecosystem health. This study investigates the interplay between rainfall, topography, and human disturbances in shaping TN and TP dynamics, aiming to identify critical factors driving nutrient pollution. Using integrated machine learning and statistical approaches, including Self-Organizing Maps (SOMs), Generalized Additive Models (GAMs), and logistic regression, the study reveals that rainfall acts as a dominant dilution factor, reducing nutrient concentrations and explaining 72.8% of TN variance and 55.6% of TP variance. Conversely, steep slopes and human-disturbed land use are positively correlated with nutrient levels but are moderated by rainfall's mitigating effect. Logistic regression further confirms rainfall's predictive significance, achieving high classification accuracy for TN (AUC = 0.82) and TP (AUC = 0.76). The identification of Critical Source Areas (CSAs) highlights priority regions for intervention, demonstrating the utility of the proposed framework for targeting water quality improvement measures. By integrating spatial and statistical modeling, this study provides actionable solutions and contributes to advancing water quality management strategies through enhanced understanding of natural and anthropogenic influences on nutrient pollution.