113 / 2023-05-31 16:45:01

Underground Coal Fire Locating Based on InSAR Technique and Elastic Dislocation Model

Underground coal fire areas locating, Elastic dislocation model(EDM), DS-InSAR, ground subsidence, DBSCAN, GA-PSO

The underground coal fires is extremely harmful to the global ecosystem which brings large amounts of greenhouse gases and causes massive ground subsidence. Xinjiang is one of the area which is severely affected by coal fires in the world. After years of extinguishing, there are still many unextinguished coal fires area. To extinguish underground coal fires, it is crucial to detect and determine their spatial location. Traditional thermal infrared technology for locating the coal fire areas, however, are ground-based. This paper presents a deformation-based method for locating the underground space caused by coal fire burning using Interferometric Synthetic Aperture Radar (InSAR) techniques and the elastic dislocation model (EDM). As the undergound coal fire space is often a approximate cuboid-shaped void and eight critical geometric parameters (i.e., two central geodetic coordinates ,length, width, height, depth, azimuth angle, inclined angle) are used to define this underground space. The proposed method efficiently estimateds the eight geometric parameters from InSAR observations. Therefore, firstly the InSAR-derived LOS surface deformation data is filtered by the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, which can identify continuous areas of surface subsidence. Secondly, it applies the elastic dislocation model (EDM), a widely used model for earthquake-induced ground deformation prediction, to construct a functional relationship between the geometric parameters and the surface deformation. Next, the genetic algorithm-particle swarm optimization (GA-PSO) is employed to determine parameters of underground coal fire areas. Finally, the proposed method was tested with both simulated and one real data sets in Miquan, Xinjiang. The results demonstrate that the estimated geometric parameters of the underground coal fire areas are accurate and compatible overall, with averaged relative errors of approximately 2.39% and 13.08% being observed for the simulated and the real data experiments, respectively. Due to the advantages of the InSAR teconology and EDM, the proposed method provides a practical method for economically locating the underground coal fire areas in large-scale.