Using high-resolution UAV-borne thermal infrared imagery to detect coal fires in Majiliang mine, Datong coalfield, Northern China

Abstract

Underground coal fires occur normally under inaccessible dangerous steep hills. As a light-weight and cost-effective equipment, unmanned aerial vehicles (UAVs)-based thermal infrared (TIR) imaging technique provides a choice to safely, timely and accurately map characteristics of coal fires which is difficult to realize using the conventional technologies. The colour images captured by UAV are used to generate a map of land cover types and estimate emissivity of ground features. TIR images are adopted to retrieve land surface temperature (LST) and thus generate orthophotos, which will be further used to recognize coal fire areas. The retrieved LST at night is validated with reference LST, and the results show yielding Root Mean Square Error (RMSE) is less than 1.03 K. The accuracy rate of identified coal fire areas at night time reaches as high as 92.78%, which is higher than that of daytime on 2nd October, 2016 and 5th October, 2016. In this research, the application of UAV-borne thermal imaging demonstrates a great potential to precisely and rapidly describe features of coal fires.