Predictive modeling of land surface temperature dynamics in Damascus, Syria using google earth engine: a remote sensing and random forest approach

Estimating Land Surface Temperature (LST) is essential for analyzing urban heat island (UHI) effects and promoting sustainable urban development. This research applies a machine learning methodology using the Random Forest (RF) algorithm on the Google Earth Engine (GEE) platform to predict LST variations in Damascus, Syria, while investigating the environmental factors influencing these changes. High-resolution time-series data from 2022 and 2023 were collected from Landsat-9, MODIS, and Sentinel-2, covering multiple seasons to ensure a comprehensive analysis. To enhance model accuracy, key spectral indices were incorporated, including the Normalized Difference Vegetation Index (NDVI), Normalized Difference Built-up Index (NDBI), and Normalized Difference Water Index (NDWI), which were selected due to their strong correlation with land cover characteristics affecting thermal behavior. Additionally, environmental variables such as albedo, aerosol absorption, precipitation, and elevation were integrated to refine LST predictions. The RF model demonstrated high predictive accuracy R2 = 0.87 in spring, with NDVI and elevation emerging as dominant variables influencing LST variability. While precipitation was included as a factor, its direct impact on LST predictions was found to be minimal, suggesting a lesser role in short-term temperature variations. Observations revealed peak summer temperatures averaging 40.7 degrees C, emphasizing the UHI effect in densely urbanized zones with minimal vegetation cover. Correlation analyses highlighted strong relationships between albedo and RGB imagery (r = 0.91), LST and albedo (r = 0.52), NDBI (r = 0.54), and elevation (r = 0.52), confirming the significant role of built-up intensity and land surface properties in shaping temperature distribution patterns. These insights contribute to a deeper understanding of urban thermal dynamics, offering crucial information for sustainable urban planning. In the context of Damascus, targeted urban greening initiatives, such as increasing vegetation cover in high-risk UHI zones and integrating reflective roofing materials, are recommended to mitigate thermal stress and enhance climate resilience. This framework provides policymakers with a robust scientific basis for informed decision-making, promoting sustainable urban adaptation strategies in semi-arid environments.