Abstract:
Against the backdrop of continued adjustment in China’s coal production layout and regional capacity optimization, the spatiotemporal distribution and causative characteristics of coal mine water hazard accidents have shown new changes. Based on national coal production statistics, coal mine water accident data, and accident investigation report texts from 2014 to 2025, this study integrates kernel density estimation, gravity-center migration, standard deviational ellipse analysis, and text mining to analyze the spatiotemporal evolution and multidimensional causative characteristics of coal mine water accidents during the optimization of the coal production pattern from both macro-spatiotemporal and micro-textual perspectives. The results show that national raw coal output generally rebounded from 2014 to 2025, whereas the number of coal mine water hazard accidents, fatalities, the accident rate per 100 million tons of raw coal, and the fatality rate per 100 million tons of raw coal declined overall. Spatially, national planned coal mining areas exhibit an overall pattern of “dense in the north and sparse in the south”, while coal mine water accidents show a more pronounced multicenter dispersed pattern. Although the two are correlated at the macro level, their local spatial correspondence is not fully consistent. After integrating the spatial background of coal production with accident occurrence density, high-value coupling zones are mainly concentrated in the major coal-producing area at the Shanxi–Shaanxi–Inner Mongolia junction and the southwestern coal mining area at the Yunnan–Guizhou–Sichuan junction. Further analysis based on gravity-center migration and standard deviational ellipses indicates that the centers of both national coal production and coal mine water hazard accidents shifted northwestward overall. Coal production continued to concentrate in resource-rich areas, but the spatial evolution of mine water hazard accidents did not follow a simple “capacity transfer–accident transfer” process. Instead, it reflects the spatial reorganization of risk elements under the new production pattern, encompassing the hazard of causative factors, the sensitivity of disaster-forming environments, the exposure of disaster-bearing systems, and overall system vulnerability. Text mining further shows that coal mine water accidents can be classified by dominant water source, among which old goaf water accidents account for the largest proportion, indicating that water accumulated in historically mined-out areas remains an important risk source for current coal mine water hazard accidents. Clear differences also exist in the dominant water-inrush source types associated with different enterprise types. Furthermore, high-weight keywords in accident investigation reports are mainly distributed across three dimensions: geological environment, on-site operations, and safety management, indicating that accident formation presents evident multidimensional causative characteristics. The findings can provide support for the investigation of causative factors of coal mine water hazards, regional risk identification, and the optimization of prevention and control strategies.