Urbanized areas experience a highly dynamic thermal environment that results in significant microclimatic variations. Detecting these variations is challenging from the urban climate perspective because weather station networks commonly lack the necessary spatial density within urban environments. The present study addresses this issue by exploring how to estimate these microclimatic variations in relation to the built environment attributes. For that purpose, it utilizes satellite-based remote sensing data sources and terrestrial thermal observations. Land surface temperature (LST) data from Terra and Aqua Moderate Resolution Imaging Spectroradiometer (MODIS) sensors are obtained for 15 years (2000–2015). Obtained LST data are compared with near-surface air temperatures (Tair) at seven locations. These locations are selected from areas with various built environment characteristics. Thermal data obtained for these locations are analyzed for two daytime and two nighttime periods within a diurnal cycle. The agreement between LSTs and Tair is initially investigated at these temporal intervals. The findings indicate a good agreement between the two data sets at all investigated time periods. The agreement is particularly stronger for the nighttime conditions. Alternative ways for deriving Tair are explored using the LST data and auxiliary information, including building-related attributes from selected locations, solar zenith angle, and black sky albedo data. One daytime and one nighttime model for deriving Tair is derived via multiple regression inquiries. The models’ performance shows an adjusted R2 of 0.916 with a root mean squared error (RMSE) of 2.31 K for daytime and an adjusted R2 of 0.979 with an RMSE of 1.27 K for nighttime assessments.
Örs et al. (Fri,) studied this question.