Modeling the climate of urban areas is of interest for studying urban heat islands (UHIs). Reliable assessment of the primary causes of UHIs and the efficacy of various heat mitigation strategies requires accurate prediction of urban temperatures and realistic representation of land surface physical characteristics in models. In this study, we expand the capabilities of the Weather Research and Forecasting (WRF) model by implementing high-resolution, real-time satellite observations of green vegetation fraction (GVF) and albedo. Satellite-based GVF and albedo replace constant values that are assumed for urban pixels in the default version of WRF. Simulations of urban meteorology in Los Angeles using the improved model show marked improvements relative to the default model. The largest improvements are for nocturnal air temperatures, with a reduction in root-mean-square deviation between simulations and observations from 3.8 to 1.9°C. Utilizing the improved model, we quantify relationships between surface and 2 m air temperatures versus urban fraction, GVF, albedo, distance from the ocean, and elevation. Distance from the ocean is found to be the main contributor to variations in temperatures around Los Angeles. After conditionally sampling pixels to minimize the influence of distance from the ocean and elevation, we find that variations in GVF and urban fraction are responsible for up to 58 and 27% of the variance in temperatures. The satellite-supported meteorological modeling framework reported here can be used for studying UHIs in other cities and can serve as a foundation for testing the efficacy of various heat mitigation strategies.