Quantifying the Urban Parks' Cooling Effects on Land Surface Temperature in Istanbul Using Random Forest

Urban parks are critical in mitigating urban heat island effects, regulating temperature, and improving urban microclimates. This study investigates the cooling impact of twenty urban parks in Başakşehir, Istanbul, using remote sensing techniques, GIS-based analysis, and statistical modeling. Key environmental variables, including Land Surface Temperature (LST), Normalized Difference Vegetation Index (NDVI), Normalized Difference Built-up Index (NDBI), and Normalized Difference Water Index (NDWI), were analyzed within buffer zones ranging from 100 to 900 meters. Land use classes were classified using the Random Forest algorithm, a robust machine learning technique commonly applied in spatial analysis. The study revealed that urban parks significantly have lower LST, with the cooling effect persisting up to 500 meters (p > 0.05) and decreasing beyond 500 meters (p < 0.05). It was observed that surface temperatures varied between 33.3°C and 37.7°C in summer and between 6.8°C and 9.5°C in winter. When land use was examined, it was determined that urban areas were between 30–35% and water surfaces remained relatively low (0–2%), indicating that the contribution of water surfaces to the cooling effect of parks was limited. Among the indices evaluated, NDVI emerged as the most influential factor, positively correlating with Park Cooling Intensity (PCI, r = 0.93). Parks with larger sizes showed more substantial cooling effects, with Park Cooling Efficiency (PCE) and Park Cooling Gradient (PCG) metrics highlighting their broader environmental impact. These findings underscore the necessity of strategic urban planning to optimize park distribution and size, ensuring sustainable cooling benefits in rapidly urbanizing areas.