Application of the TES algorithm to TIMS data acquired in HAPEX-Sahel

Abstract

In 1992 Thermal Infrared Multispectral Scanner (TIMS) data were acquired from the NASA C-130 aircraft over the Sahel, Africa as part of the Hydrological and Atmospheric Pilot Experiment in the Sahel (HAPEX). The TIMS instrument measures the radiation from the surface modified by the atmosphere in 6 channels located between 8 and 12.5 /spl mu/m. These airborne TIMS data provide a surrogate for data that will be available globally with the launch of the Advanced Spaceborne Thermal emission Reflectance Radiometer (ASTER) in 1998 on the first AM Earth Observation System platform and were used to test the Temperature Emissivity Separation (TES) Algorithm being developed for ASTER. This method relies on an empirical relationship observed between the range of emissivities and the minimum value for the 6 TIMS channels. HAPEX is an international land-surface-atmosphere observation program that was undertaken in western Niger, in the west African Sahel region. The TES algorithm was applied to scenes of about 1000 scans for both the east and west central sites with the interesting results on 2 and 4 September. The spectral behaviors of Tiger Bush, bare soil and millet sites were studied. There was up to a 10 K difference in the brightness temperature over the 6 channels for the bare soil site. Channels 1 to 3 (8.2 to 9.4 /spl mu/m) were 10 K cooler than the longer wavelength channels (9.6 to 12.5 /spl mu/m) which is characteristic of soils rich in quartz. While for tiger bush site there was less than 0.5 K difference observed. In terms of emissivity, channel 5 showed very little spatial variation in emissivity and the short wavelength channels observed substantial regions with emissivities of less than 0.8. As expected there was little spectral difference in emissivity observed for vegetated pixels, less than 0.01, and the average amplitude was 0.97. The extracted vegetation temperature was close to the air temperature. The authors results indicated a large sensitivity to the values used for the atmospheric corrections, which in the authors' case was done using MODTRAN 3.5 and nearby radiosoundings of the atmosphere.