Urbanization in arid regions simultaneously intensifies air pollution and urban heat stress, yet their coupled interactions remain poorly understood. This study investigates the spatiotemporal relationships between land surface temperature (LST) and atmospheric pollutants in Urumqi, a representative arid-zone megacity of Northwest China, during 2019-2024 by integrating multi-source satellite retrievals and numerical simulations. MODIS LST and Sentinel-5P column pollutant data (O3, NO2, SO2, CO, CH2O) were analyzed using trend statistics and spatial autocorrelation methods. Results show that daytime LST exhibited a decreasing trend, whereas nighttime LST increased significantly, reflecting an enhanced nocturnal urban heat island. Pollutants displayed distinct seasonal cycles, with O3 increasing and NO2 declining over time. Bivariate Moran's I and Local Indicators of Spatial Association (LISA) revealed that O3 exhibited the strongest positive spatial coupling with LST, especially at night, whereas NO2 and SO2 showed weaker and seasonally variable correlations. To explore mechanisms, WRF and WRF-Chem experiments for February 2023 indicate that ozone and co-emitted pollutants enhanced near-surface warming, leading to daytime temperature differences of about 0.4-1.6 degrees C and to ozone-linked nocturnal warming when O3 exceeds its multi-day mean. These results highlight ozone as a dominant driver of pollutant-temperature feedbacks in arid urban ecosystems and provide a quantitative basis for jointly managing air quality and urban heat risks in arid basin cities.