Global warming induced by greenhouse gases is increasing wildfire frequencies and scale. Since wildfire again releases greenhouse gases(GHGs) into the air, the vicious cycle is repeated. Satellite remote sensing is a useful tool for detecting wildfire. However, estimating the GHGs emission from wildfire has not been challenged yet. Wildfires are estimated to be responsible for, on average, around 30% of global total CO emissions, 10% of methane emissions, 38% of tropospheric ozone, and over 86% of black carbon. So we need to quantify the emitted gases by biomass combustions, which can be measured by the FRP (fire radiative power) derived from the spectral characteristics of satellite sensors. This paper described the algorithm for retrieval of FRP using COMS(Communication, Ocean and Meteorological Satellite), the Korean geostationary meteorological satellite. The FRP of wildfire is retrieved by single waveband methods suitable to COMS channels. The retrieval of FRP is dependent on the emissivity of each bandwidth. So, we used MODIS NDVI through a spatio-temporal calibration for the emissivity calculations. We made sure that the FRP in wildfire pixel is much higher than its spatially and temporally neighboring pixels. For future work, we should quantify the relationships between FRP and the biomass combustion according to fuel types.