Enhancing Temporal Analysis of Jodhpur's Land Surface Temperature: A Comparative Study of XGBoost and Fine-Tuned XGBoost Model

Recent adverse climatological conditions paves the way for studying, understanding and predicting capabilities of it. Being a pivotal parameter of climate, Land Surface Temperature (LST) is the ideal and substantial choice for this study. By utilizing the analytical capabilities of remotely sensed LST data and Geographic Information System (GIS), this study utilizes the XGBoost regressor to predict temperatures in Western Rajasthan city, i.e., Jodhpur from 2003 – 2022 on the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor based data. This LST data is retrieved from Aqua and Terra satellite mounted MODIS sensor. Through meticulous preprocessing and fine-tuning, the model's accuracy is significantly enhanced. The findings not only provide valuable insights for climate prediction applications but also contribute to the field of temporal data analysis in machine learning. This unique and novel temporal predictive analysis is founding work for the development of heat mitigation strategies to be developed by urban planners, dwellers, policy makers and stakeholders.