Retrieval of Component Temperatures of Leaf, Sunlit and Shaded Soil in Maize Canopy Based on Airborne Thermal Infrared Multiangular Observations

Land surface temperature is the key parameter to determine land surface energy budget. Due to land surface rugged geometric structure and different components, pixels in the remote sensing image measured from space are often non-isothermal. Several approaches of decomposing temperatures have been developed recent years. These approaches only focus on two components, soil and leaf. In fact ground-based measurements indicate temperatures of sunlit and shaded leaf are nearly the same while temperatures of sunlit and shaded soil are different significantly. In order to retrieve three component temperatures including leaf, sunlit soil and shaded soil, an analytical parameterized model named FR97 was modified in this paper. The modified FR97 model was calibrated by 4SAIL (Scattering by Arbitrarily Inclined Leaves) model. Bayesian algorithm was used to solve the inversion problem. Finally the inversion algorithm was verified by the airborne observed datasets of Wide-angle Infrared Dual mode line / area Array Scanner (WIDAS) over dense maize area during the Heihe Watershed Allied Telemetry Experimental Research (HiWATER) campaign. The results indicate that the modified FR97 model improves a lot for hotspot effect than original FR97 model. And more accurate component temperatures were retrieved especially for sunlit soil component (RMSE = 1.378°C).