XingHuan Visible and Uncooled Multispectral Infrared Camera for Wildfire Detection: Algorithm Description and Initial Validation

Advancements in wildfire detection using meteorological satellite data have progressed significantly since the 1980 s, driven by improved instruments and analytical algorithms. In May 2024, the XingHuan satellite, designed for wildfire monitoring, was deployed into orbit. It carries a visible-panchromatic camera and a multispectral infrared imaging system, featuring a vanadium oxide (VOx) array. The system provides a 60 m spatial resolution for visible- panchromatic imaging and 120 m for both mid-wave infrared (MWIR) and long-wave infrared (LWIR) imaging, all within a 1.5 kg payload. The camera's high resolution, five spectral bands, and time-delay integration (TDI) method significantly improve the signal-to-noise ratio (SNR) for detecting small wildfires. However, accurate radiometric quantification of the data depends on precise temperature measurement and thermal control system of the camera, which complicates fire pixel identification. This study proposes a novel approach, the Fire dual-band local contrast method (Fire-DBLCM) for fire detection, using uncalibrated raw data from the XingHuan satellite. The Fire-DBLCM method selects an optimal background based on fire's spectral properties, improving detection sensitivity. Inspired by the human visual system (HVS), this method circumvents the conventional fire detection process, such as radiometric calibration, atmospheric correction, cloud and water pixel classification, and absolute thresholding. Integrating both mid-wave and long-wave spectral bands further enhances the SNR compared to single band-dependant initial data and conventional local contrast method (LCM) approaches. We evaluate the Fire-DBLCM's performance by comparing its results with data from medium-resolution satellite sensors (MODIS and VIIRS) and high-resolution satellites (Landsat-8/9 and Sentinel-2). The findings indicate that the Fire-DBLCM algorithm identifies smaller wildfires more effectively than medium-resolution satellites and detects residual heat signatures more sensitively than high-resolution satellites. The smallest fires identified were $< $ 700 m 2 during the day and $< $ 100 m 2 at night, establishing a new benchmark for uncooled infrared satellite constellations with high spatiotemporal resolution.