Socioeconomic drivers of riverine biochemical oxygen demand: Insights from interpretable machine learning

Biochemical oxygen demand (BOD₅) is a comprehensive indicator for assessing organic pollution and water quality in rivers. Socioeconomic factors, such as regional population pressure, urbanization intensity, and land use, affect riverine BOD₅ by altering aquatic environmental conditions. However, the underlying driving mechanism remains insufficiently understood. This study combined interpretable machine learning (IML) with partial least squares structural equation modeling (PLS-SEM) to develop an analytical framework for the factors influencing riverine BOD₅. Our findings indicated that variations in urbanization intensity and socioeconomic patterns across the watershed significantly influence river water quality characteristics, with BOD₅ showing a strong correlation with these water quality parameters. The developed Random Forest model confirmed the substantial contribution of nitrogen and phosphorus compounds to predicting riverine BOD₅, identifying TKN as the critical driver. Furthermore, path analysis using PLS-SEM revealed that socioeconomic factors exert an indirect influence on BOD₅ concentrations by mediating the discharge of nitrogen and phosphorus nutrients into surface waters. In highly urbanized areas, riverine BOD₅ was strongly influenced by socioeconomic activities, whereas in less urbanized regions, it is governed more by natural processes than by socioeconomic drivers. This study elucidates the impact of socioeconomic development on riverine BOD₅, providing scientific insights for deeply understanding the driving mechanisms of river pollution and fostering sustainable environmental and socioeconomic development.