Data Optimization Based on Ordinary Kriging for Radon Detection to Identify Spontaneous Combustion Areas

Abstract

In order to optimize the data analysis of radon detection, the ordinary Kriging interpolation pertaining to Geostatistics was introduced to avoid the interpolation distortion caused by the minimum curvature fitting method currently in effect. The radon detection data was collected from the Zhangling fire area of the Gushuyuan Coal Mine, the isolated outliers caused by human factors in the raw data were deleted through boxplot, and the concentration intervals that characterizes the spontaneous combustion area were obtained eventually. The statistical characteristics of the normal distribution for radon detection data were obtained through exploratory analysis and normal test and the data conversion was sorted out according to the statistical characteristics. Then, the variogram model for converted data was constructed to interpolate as the ordinary Kriging. The locations of suspected fire sources delineated by ordinary Kriging were contrasted with the locations by minimum curvature fitting method and then the accuracy was verified through drills. The results show that compared with the 4 suspected fire areas and 6 high temp fire sources delineated by the minimum curvature method, 2 areas and 4 sources were fixed with the ordinary Kriging. The borehole results for high temp fire source locations are consistent with the ordinary Kriging inferences, while excluding 2 additional locations delineated by the minimum curvature fitting, which proves that the ordinary Kriging can improve the accuracy of the judgements for fire source locations and can be a guide for the prevention of coal spontaneous combustion.

Keywords

• isotopic radon detection
• fire source detection
• geostatistics
• ordinary kriging
• data optimization