多源卫星热点产品的融合与交互检验

卫星遥感是快速获取地面火灾信息的有效手段，不同频次不同空间分辨率的各类卫星热异常产品记录了大量火情信息，然而在大范围空间尺度上很难对其进行地面火灾监测有效性的验证，影响着多源卫星火情监测融合产品的实际应用。本文研究以贵州地区为例，运用较高空间分辨率的中分辨率成像光谱仪(moderate resolution imaging spectroradiometer, MODIS)和可见光红外成像仪(visible infrared imaging radiometer suite, VIIRS)监测热点对两个先进的静止气象卫星葵花成像仪(advanced himawari imager, AHI)监测热点产品进行检验和评估，分别量化两个AHI热点算法的侦测性能。结果表明：在8 h时间窗内，两个AHI热点算法相对于MODIS和VIIRS热点的误分率在39.2%～71.7%,漏分率在52.7%～68.5%,而相对于地面调查火灾的误分率为25.5%和33.3%。随着格点内极轨卫星热点像元数的增加，两个AHI算法侦测到的概率迅速增大，当5 km×5 km格点上出现4个MODIS和10个VIIRS以上...

Study on Intercomparison and Combination of Multi-sources Satellites Active Fire Products

Satellite remote sensing is the effective means to monitor fire occurred on wild ground. Plenty of fire information are contained in active fire products derived from satellite sensors with different temporal frequencies and spatial resolutions, while their practical application are suspectable because validations are difficult to execute on a large scale area. Two AHI (Advanced Himawari Imager) fire algorithm products across Guizhou were compared with the thermal hotspots from higher resolution LEO (Low Earth Orbiting) satellites: MODIS (Moderate Resolution Imaging Spectroradiometer) and VIIRS (Visible infrared Imaging Radiometer Suite). Detection capabilities of the two AHI algorithms were evaluated too based on the number of LEO hotspots with and without a matching AHI pixel within a 5×5 km grid. For an 8-hour window from 01:00 to 08:00 UTC, the Commission errors of two AHI algorithms are 39.2%-71.7% and omission errors are 52.7%-68.5% with respect to MODIS and VIIRS detections, while the commission errors are 25.5% and 33.3% with ground truth fires. The detection probabilities of two AHI algorithms are increasing rapidly with the increase of LEO hotspots, showing an 80% detection rate when a minimum of 4 MODIS hotspots or when 10 VIIRS hotspots occur within 5×5 km grid. Therefor two methods of multi-satellite hotspots composite are applied in Guizhou province to combine the best traits of LEO and geostationary satellite data. One is based on maximum gridded FRP (Fire Radiant Power) value and the other is based on total weight calculated from hotspots detected by geostationary and polar-orbiting satellites. The spatial distribution of gridded fire probability in each month during fire occur frequently are then obtained to support fire prevention and management.