Correlation-based temperature and emissivity separation algorithm

Based on analyzing the relationship between the atmospheric downward radiance and surface emissivity, this paper proposes a correlation criterion to optimize surface temperature during the process of temperature and emissivity separation from thermal infrared hyperspectral data, and puts forward the correlation-based temperature and emissivity separation algorithm (CBTES). The algorithm uses the correlation between the atmospheric downward radiance and surface emissivity to optimize surface temperature, and obtains surface emissivity with this temperature. The accuracy of CBTES was evaluated by the simulated thermal infrared hyperspectral data. The simulated results show that the CBTES can achieve high accuracy of temperature and emissivity inversion. CBTES has been compared with the iterative spectrally smooth temperature/emissivity separation (ISSTES), and the comparison results show that they have relative accuracy. Besides, CBTES is insensitive to the instrumental random noise and the change of atmospheric downward radiance during the measurements. As regards the nonisothermal pixel, its radiometric temperature changes slowly with the wavenumber when its emissivity is defined as r-emissivity. The CBTES can be used to derive the equivalent temperature of nonisothermal pixel in a narrow spectral region when we assumed that the radiometric temperature is invariable in the narrow spectral region. The derived equivalent temperatures in multi-spectral regions in 714–1250 cm−1 can characterize the change trend of nonisothermal pixel’s radiometric temperature.