Modeling Soil-Water Quality Interactions and Identifying Key Factors in Humid Aquaculture Areas Based on Random Forest Authors

In humid ecological areas, complex processes of material transfer and transformation occur between pond soil and water, which significantly influence water quality balance and the evolution of sediment-related environmental risks. This study selected three typical aquaculture ponds in Guigang City, Guangxi Province, and conducted high-frequency synchronized monitoring over 180 days. A total of 540 daily average datasets were obtained, covering seven key physical and chemical indicators, including soil moisture, pH, and organic matter content in the 0-10 cm bottom soil layer, as well as ammonia nitrogen, dissolved oxygen, and turbidity in the water. A coupling model of soil and water quality factors was constructed using the Random Forest regression algorithm. The model achieved a coefficient of determination (R²) of 0.87 for ammonia nitrogen concentration prediction. Mean Decrease in Impurity (MDI) and permutation importance analysis showed that soil moisture (32.4% contribution) and organic matter content (26.1% contribution) were the core influencing factors. The results indicate that nutrient release from wet soil during high-water periods is a key driver of water quality fluctuations. Based on this, the study proposes a scientific recommendation to establish a "soil-water coordinated dynamic regulation system."