A comparison of four methods for extracting Land Surface Emissivity and Temperature in the Thermal Infrared Hyperspectral Data

Land Surface Temperature (LST) and Land Surface Emissivity (LSE) are two important physical propertiesof Earth apos;s surface. LST retrieval plays a valuable role in environmental studies. Therefore, in order toestimate LST accurately, it is necessary to obtain LSEs. The HyTES (Hyperspectral Thermal EmissionSpectrometer) instrument has 256 spectral bands covering the thermal infrared (TIR) spectral range. Due toa large number of narrow bandwidths of HyTES, this sensor can produce the accurate LST and LSEs. Themain goal of this paper is to evaluate the accuracy of LSE and LST retrieval methods from HyTES data andto improve the accuracy of the Normalization (NOR) method. For this purpose, four different methods havebeen considered to retrieve LSE: (i) the Reference Channel method (REF), (ii) the Emissivity Normalizationmethod, (iii) the Alpha emissivity method (AlPHA) and (iv) a method to improve the NOR algorithm. Thefirst three methods have been widely used with thermal multispectral data in other researches. Thesemethods were used with HyTES hyperspectral data in this paper; the fourth is a new method that improvedthe accuracy of the NOR method. The results of quality assessment show that the emissivity RMSEs of theREF, NOR, ALPHA methods and the new proposed method are 0.021, 0.815, 0.034 and 0.0201,respectively. Also, LST RMSEs of the REF and NOR methods are less than 1.5 K.