Predicting land surface temperature and examining its relationship with air pollution and urban parameters in Bengaluru: A machine learning approach

The study aims to predict land surface temperature (LST) of Bengaluru, a rapidly urbanizing city of India facing air pollution and temperature challenges. Different parameters, like air pollutants, urbanization factors, surface indices, and meteorological data, were used to develop machine learning models. The analysis revealed strong to moderate correlations between LST and most parameters, during both summer and winter seasons. The sensitivity analysis showed that LST was most sensitive to change in PM2.5 values, during both seasons, followed by SO2 (summer) and NO2, CO, and O-3 (winter). Artificial neural networks (ANN) consistently outperformed multiple linear regression (MLR), support vector regression (SVR), and random forest (RF), with higher accuracy (R-2 = 0.92 in summer, R-2 = 0.95 in winter) and lower errors (RMSE and MSE) compared to other models. The order of performance was found to be ANN > RF > SVR > MLR. The findings contribute to understanding LST relationship with different parameters. The present study can help policymakers, urban planners, and environmental agencies in implementing effective strategies to mitigate urbanization's impacts on air quality and temperature regulation. Reducing air pollutants may alleviate the heat island effect and improve overall environmental conditions, fostering a healthier and more sustainable urban environment.