Estimation of Building Rooftop Temperature from High Spatial Resolution Aerial Thermal Images

Abstract: This letter presents a novel technique to calculate temperatures of building rooftops and other impervious surfaces from high spatial resolution aerial thermal images. In this study, we collected aerial radiance images of 30cm spatial resolution using a FLIR Phoenix imager in long-wave and mid-wave infrared wavelengths for the city of Cedar Falls, USA to estimate building roof temperature loss. Simultaneous ground temperature measurements were made at pre-selected ground targets and roofs using 9 Fluke 561r infrared thermometers. Atmospheric correction of aerial images was performed by Empirical Line Calibration (ELC) method. The resulting ground-leaving radiances were corrected for the emissivity of different roof types and the true kinetic temperature of the building roofs was calculated. The ELC model was observed to perform better when only impervious surface targets were used for the regression. With an R2=0.71 for ELC, the method produced a root mean squared error of 0.74{\deg}C for asphalt roofs. Further, we observed that the microclimate plays a significant role while synchronizing aerial and ground measurements.