Enhancing Urban Heat Island Analysis through Indices-Based Prediction of Land Surface Temperature in Khon Kaen City, Thailand

Urban Heat Island (UHI) presents significant environmental concerns, exacerbated by urbanization and climate change, resulting in elevated temperatures in urban areas compared to rural ones. This study focuses on Khon Kaen City, Thailand, to improve UHI analysis through machine learning models, leveraging satellite-derived indices to predict Land Surface Temperature (LST). Employing eXtreme Gradient Boost Regressor (XGBR) and Random Forest Regressor (RFR), we used Normalized Difference Built-Up Index (NDBI), Enhanced Built-up and Bareness Index (EBBI), and Soil-Adjusted Vegetation Index (SAVI) from Landsat 8 images for the years 2013, 2018, and 2023. Our results demonstrate that XGBR outperforms RFR in estimating LST, with higher R2 values and lower mean absolute error (MAE), mean squared error (MSE), and root mean squared error (RMSE). Specifically, XGBR showed R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> values of 0.7788, 0.8039, and 0.7574 for 2013, 2018, and 2023, respectively. The study emphasizes the crucial influence of NDBI a nd E BBI on predicting LST, showcasing the necessity of including building index variables in UHI research. This method provides important information for urban planners and policy makers. It can facilitate more precise evaluations of trends in UHI and help develop sustainable urban strategies to counteract the effects of UHI in Khon Kaen City.