Remote-sensing-based coal-fire detection with low-resolution MODIS data

Remote-sensing imagery is often used for detecting and monitoring coal fires. The Landsat7 Enhanced Thematic Mapper Plus (ETM+) sensor and its predecessors of the Landsat family were frequently utilized for that purpose. With Landsat5 quickly approaching the end of its lifetime and the partial malfunction of Landsat7 in 2003, other potential sensors, including Moderate Resolution Imaging Spectroradiometer (MODIS), merit investigation. One kilometer MODIS data were successfully acquired and analyzed to detect coal fires in China during one summer and two winter night scenes. Band ratios of MODIS bands 20/32 enhanced subpixel-sized hot spots over background values, and an automated thermal anomaly algorithm was an asset in extracting potential coal-fire locations. For areas with known subsurface fires, between 0% and 17% were correctly detected in the three images. Areas with surface fires had success rates of 42% to 49%. These results indicate that MODIS is potentially useful for monitoring large areas for newly developing surface coal fires. Most subsurface coal fires, however, remain undetected.