Multi-sensor data fusion for the detection of underground coal fires

The spontaneous combustion of coal causes widespread underground coal fires in several countries, amongst which is China. These coal fires cause serio us environmental, economic and safety problems. In northern China, the coal fires occur within a wide region stretching 5000 km east-west and 750 km n orth-south. Remote sensing therefore provides an ideal tool for monitoring this environmental hazard over such a large and remote area. As part of a r esearch project to detect, measure, monitor and extinguish these coal fires , this paper describes a remote-sensing-based multi-sensor data-fusion meth odology for detecting the underground fires. The methodology is based on fu sing a variety of satellite-based image types (optical, thermal, microwave) together with airborne data (optical and thermal infrared) and ancillary d ata sources such as geological and topographic maps. The results of the rem ote-sensing data fusion are presented, using pixel-based, feature-based and decision-based fusion approaches.