Skip to Main Content
Surface emissivity estimation is one of the most important aspects for land surface temperature estimation from remotely sensed data. Since different surface types are always combined in one single pixel, the emissivity is complicate to estimate. In this study, a modified NDVI threshold method is used to estimate the emissivities of band 4 and 5 from FY3A/VIRR data. The NDVI thresholds have been determined to separate bare soil, partially vegetated soil and fully vegetated land. Then a regression model that links emissivities in band 4 and 5 to reflectances in band 1-2 and 6-9 and NDVI is built to calculate emissivities in band 4 and 5 of bare soil. The regression relations also have been used to estimate the partially vegetated soil emissivity, and a constant emissivity have been used for fully vegetated area. Using 300*300 pixels around Heihe watershed, emissivity of each pixel in band 4 and 5 have been estimate, and a simple analysis show that the proposed algorithm can be used to estimate the emissivities of band4 and 5 from FY3A/VIRR data.