The satellite’s derived thermal data has been immensely supportive to remotely access the temperature of Earth's surface materials. The downscaling techniques have been used to predict material surface temperature (MST) at higher spatial resolutions. In this research, three MODIS satellite downscaling models named (Resolution adjusted regression downscaling (RAd), Kriging Regression (KR), and Regression Tree (RegTree)) have been developed and verified for urban and rural materials of Jaipur district, India. The Kriging regression downscaling model (KR) produced least RMSE (0.61 °C) followed by RAd (0.75 °C) and RegTree (0.82 °C) models at 100 m resolution. The KR model downscaled MODIS MST data at 100 m spatial resolution, showed error below (1 °C), for more than 60% of materials and less than (2 °C) error for 75% of the materials. KR based algorithm was most suitable for both rural and mixed landscape areas due to the reconstruction of residuals at each scale. The regression based material surface temperature (MST) downscaling techniques were precisely dependent on the Earth's surface material characteristics, particularly in urban landscape. The KR methodology would facilitate the prediction of MST for numerous Earth surface materials for lands surface and environmental related applications.