A Remote Sensing Approach to Spatiotemporal Analysis of Land Surface Temperature in Response to Land Use/Land Cover Change via Cloud Base and Machine Learning Methods, Case Study: Sari Metropolis, Iran

چکیده مقاله

This study examines the evolution of land use and land cover (LULC) patterns and their impacts on urban heat development, offering insights critical for city planning. This research develops a predictive model connecting LULC shifts to trends in land surface temperature (LST) over 35 years (1985–2020) in Sari metropolis, Iran. Utilizing Google Earth Engine (GEE), we calculated LST and generated LULC maps across eight intervals. A Convolutional Neural Network (CNN) was employed to examine the LULC-LST relationship, benchmarking its performance against traditional machine learning models like Support Vector Machine (SVM) and Random Forest (RF). The results revealed a significant substantial urban expansion, where urban areas grew from 24.61% in 1985 to 53.4% in 2020, largely replacing forests and rangelands. Forest cover experienced a drastic reduction, declining from 48.78 to 9.29% over the study period. This type of unsustainable urbanization contributed to a maximum temperature increase of 2 °C, with the highest temperature rises in the areas where forests were cleared for development. However, the regions where focused on increasing vegetation experienced minimal temperature increase. In terms of modeling accuracy, CNN model outperformed other methods with an accuracy of 92.03%, significantly higher than SVM (85.4%), RF (86.3%), and linear regression (69.8%). The CNN’s superior performance lies in its ability to incorporate additional LULC indices (Normalized Difference Vegetation Index, Normalized Difference Built-up Index, and Normalized Difference Water Index), enabling extraction of complex spatial features and significantly enhancing predictive accuracy. Ultimately, suggested deep learning (DL) approach provided a reliable tool for monitoring LULC shifts and forecasting urban heat islands that contribute to key insights for sustainable city planning and heat mitigation.