Evaluation of five algorithms for extracting soil emissivity from hyperspectral FTIR data

It is well known that soil emissivity exhibits large uncertainty in thermal infrared spectral region. In order to find a way to derive soil emissivity accurately, we examine several existed typical temperature emissivity methods (e.g. NEM, ISSTES, ADE, MMD and TES). Based on the 58 soil spectra of the ASTER Spectral Library, several sets of thermal infrared hyperspectral data were simulated to assess the applicability, stability and accuracy of these methods respectively. This work also brings some improvements of the algorithms based on the results analysis, including: a new optimal maximum emissivity has been suggested for NEM, a better empirical relationship has been discovered to substitute the original mean-minimum maximum difference relationship in MMD method, the original NEM module has been replaced by ISSTES to acquire the accurate initial value of emissivity in TES. As a conclusion, we find the ISSTES is the best. Finally, we present an example of soil emissivity extraction using five methods mentioned above with ground-based measurement hyperspectral data. The distribution of derived emissivity spectrum verifies the results of algorithm analysis.