Using characteristic spectral bands of OMIS1 imaging spectrometer to retrieve urban land surface temperature

Abstract

According to the reflectance spectra of various materials, this paper mainly studies the relations between inherent features indicated by different combinations of characteristic spectral bands and surface temperature for five land surface cover types in the Shanghai area. On the basis of above‐analysis results, we used the built optimum regression models to retrieve surface temperature, and then the precision and feasibility of the correlation method were evaluated. Results show that a clear and high correlativity exists between characteristic spectral bands and surface temperature. The maximum error is about 2.0 K by using correlation models to estimate urban land surface temperature, and main error of retrieved temperature results from mistakenly classification and existent mixture pixels. This indicates that characteristic spectral bands can be used as a modulating parameter for the estimation of surface temperature.