Comprehensive Water Quality Assessment using Ensemble Machine Learning in a Tropical Lake, Kerala, SW India

ABSTRACT This study intends to evaluate and characterize the water quality of Vellayani Lake (VL), a tropical freshwater body in Kerala, Southwest India. Comprehensive analyses of physico-chemical parameters were conducted in surface water samples (n=13) during post-monsoon (January), pre-monsoon (May) and monsoon (July) periods. Key parameters viz., temperature, pH, Electrical Conductivity, Total Dissolved Solids, Dissolved Oxygen and nutrient concentrations were measured. The Water Quality Index (WQI) classification was employed to assess water quality. WQI values ranged from ‘Excellent’ to ‘Poor’, with a general decline of quality during post-monsoon. Additionally, a Random Forest based ensemble machine learning model was applied to validate the WQI results, achieving high accuracy (R2 = 0.96) and identified phosphate, Dissolved Oxygen, Electrical Conductivity and Total Dissolved Solids as key predictors. This integrative approach provides a comprehensive understanding of the lake’s water quality dynamics, emphasizing the need for spatially targeted interventions to reduce pollutant loads and stabilize the lake’s ecological function.