DEVELOPMENT OF THE GOCI ATMOSPHERIC CORRECTION ALGORITHM FOR LAND APPLICATIONS

초록

Although GOCI is mainly designed for ocean colour monitoring, it also has great potential for land applications by the high temporal resolution. In this study, we aim to develop the GOCI atmospheric correction (AC) algorithm over land area, which is somewhat different from the one developed for ocean area. The GOCI AC algorithm removes the effects of sun angle, Rayleigh scattering, aerosol scattering, and gas absorption to derive surface reflectance from at-sensor radiance. To represent the spatio-temporal variation of aerosol which is an influential factor for deriving surface reflectance, the spatio-temporal model of AOT was produced using AERONET data. We used GOCI L1B radiance obtained from March to October, 2016 and derived surface reflectance after applying AC algorithm. Time-series GOCI surface reflectance well showed the phenology of vegetation growth. However, GOCI hourly reflectance represented diurnal variations due to SZA change within a day. To verify the AC algorithm, in-situ measurements and MODIS products were used. When comparing GOCI surface reflectance to in-situ measurements and MODIS, a similar temporal patterns were found. However, the absolute value of surface reflectance of GOCI was higher than in-situ measurement and MODIS. The difference in the absolute value of surface reflectance was larger in NIR band than red band. And this tendency was more evident for rice paddy than grass land.