Retrieval of land surface temperature from modis mid-infrared data

This paper retrieves the Land surface temperature (LST) from MODIS mid-infrared data. Considering that the daytime mid-infrared satellite data contains both reflected radiance due to sun irradiance and emitted radiance from the surface and the atmosphere, this paper estimates the bidirectional reflectivity in mid-infrared channels firstly, and then derives the directional emissivity with the linear kernel-driven BRDF model. Finally based on the radiative transfer equations in mid-infrared channels, the LST is retrieved. The retrieved LSTs are preliminarily validated with the MODIS LST product MYD11B1. The results show that the root mean square error (RMSE) between the two estimated LST is below 1.9 K and the Bias is below 1.10 K. In addition, some in situ measurements are also used to validate the retrieved LST. The results show that the RMSE is 2.06 K and Bias is 0.73 K.