Development of a SGLI Adaptable Active Fire Detection Algorithm using NIR, SWIR and TIR Observation Channels

Abstract

We describe a forest fire detection algorithm adaptable to the Second-generation GLobal Imager (SGLI) sensor which is on board the upcoming GCOM-C satellite. Since the SGLI lacks a mid-infrared 4μm ‘fire channel' which is mainly used for existing satellite-based active fire observations, our fire detection approach uses near infrared and shortwave infrared reflectance channels and a thermal infrared channel which are equipped on the SGLI. Some threshold tests for fire pixels detection are made by a methodology based on the multi-spectral analysis for the reflectance channels, and based on brightness temperature difference between a fire pixel and its non-fire background for the thermal channel. In this study, we use the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data to simulate the pre-launch SGLI data and to create truth fire data. We performed a detection accuracy evaluation of our algorithm using 26 ASTER scenes. The correct detection rate of our algorithm against the truth data was approximately 14[%]. The reliable rate which is defined as a ratio of the correct detected pixels to the total detected pixels was approximately 98[%]. The fire size which is accurately detectable by our algorithm was approximately 185[m] square.