Use of hot spot analysis to detect underground coal fires from landsat-8 TIRS data: A case study in the Khanh Hoa coal field, North-East of Vietnam

Abstract

Underground coal fire (UCF) detection from remotely sensed data plays an important role in controlling and preventing the effects of coal fires and their environmental impact. The limitation of commonly used methods does not take into account spatial autocorrelation among observations. For solving this limitation, a method for UCF detection was proposed using hot spot analysis (HSA). Based on the radiative transfer equation (RTE), land surface temperatures (LSTs) were firstly retrieved from the Landsat-8 TIRS data. The degree of spatial clustering among these LSTs was measured using HSA. UCF areas were then delineated based on 99 percent confidence level of hot spot areas. These fires were finally validated using known UCF sites and cross-validated with the results extracted from the ASTER TIR image. It was found from a case study in the Khanh Hoa coal field (North-East of Vietnam); (i) UCFs were strongly correlated with known coal fires and were highly consistent with those obtained from the ASTER TIR data; (ii) a total fire area of 197 hectares was detected, of which the fire areas of low, medium, high and extremely high levels were 37.3, 47.3, 53.2 and 59.3 hectares respectively; (iii) these fires were mainly detected in the central area and at coal ash dump sites of the southern coal field. The results show HSA can be used to effectively detect UCFs.