Predicting Spatiotemporal Dynamics of Land Use Influenced Thermal Patterns Using Remote Sensing-based Machine Learning Algorithms

Urbanization in arid regions significantly influences regional thermal environments and ecosystem services. This study investigates the transformation of natural landscapes, analyzing spatiotemporal land use changes and thermal dynamics over 1992–2023, with projections through 2043. Using Landsat imagery and Artificial Neural Network (ANN) modeling, results revealed that urban expansion increased by 171.93% (303.18 km² to 825.11 km²), while natural barren lands decreased by 18.50% (2 961.67 km² to 2 413.55 km²). Surface temperatures in converted areas reached 44.62°C by 2023, with a negative correlation between vegetation and temperature ( r = −0.290) and positive correlation between built-up areas and temperature ( r = 0.340). ANN-based projections under a business-as-usual scenario indicate continued natural land conversion, with urban areas expected to occupy 42.23% of the study area by 2043, potentially exposing 43.1% of converted lands to extreme temperatures (> 45°C). Under a mitigation scenario incorporating strategic urban planning interventions, the expansion of high-temperature zones could be significantly reduced. The percentage of urban areas experiencing severe heat stress (land surface temperature [LST] > 45°C) is predicted to increase from 12.3% in 2023 to 28.6% and 43.1% in 2033 and 2043, respectively, under the business-as-usual scenario. Ground-truth validation using meteorological station data confirmed the accuracy of LST estimates, with an RMSE of 0.695 and correlation of 0.83. These findings provide crucial insights for sustainable urban expansion in arid regions, supporting evidence-based strategies for maintaining ecosystem services while accommodating urban growth. The study highlights the need for context-specific interventions including reflective surfaces, strategic green corridors, and heat-resistant urban design to mitigate extreme heat risks, particularly for vulnerable populations. The methodological approach and findings contribute to global urban sustainability goals, specifically supporting SDG 11 (Sustainable Cities and Communities) and SDG 13 (Climate Action).