Prediction of river water temperature using the Extreme Gradient Boosting – tropical river system of India

The physical, biological, and chemical properties of a river are directly influenced by its river water temperature (RWT), which also controls the survival and fitness of all aquatic organisms. Machine Learning (ML) gained popularity because of its ability to model complex and nonlinearities between RWT and its predictors compared to process-based models that require large data. The present study demonstrates a new ML approach, Extreme Gradient Boosting (XGBoost), to predict accurate RWT estimates with the most appropriate form of AT. Further, the proposed XGBoost results are compared with the Support Vector Regressor (SVR) model. The proposed modelling framework's effectiveness is demonstrated with a tropical river system of India, Tunga-Bhadra River, as a case study. Results indicate that the XGBoost results are better than SVR for RWT prediction. The study demonstrates how ML methods can be used to generate accurate RWT predictions in river water quality modelling.