College of Geoscience and Surveying Engineering, China University of Mining and Technology-Beijing, Beijing, 100083, China.
General Defense Geological Survey Department, Huaibei Mining Co., Ltd., Huaibei, 235000, China.
Environ Sci Pollut Res Int. 2024 Aug;31(36):49227-49243. doi: 10.1007/s11356-024-34456-7. Epub 2024 Jul 25.
Coal mining in regions characterized by high groundwater table markedly predisposes to surface subsidence and water accumulation, thereby engendering substantial harm to surface vegetation, soil, and hydrological resources. Developing effective methods to extract surface disturbance information aids in quantitatively assessing the comprehensive impacts of coal mining on land, ecology, and society. Due to the shortcomings of traditional indicators in reflecting mining disturbance, vegetation aboveground biomass (AGB) is introduced as the primary indicator for extracting the mining disturbance range. Taking the Huaibei Coal Base as an example, Sentinel-2 MSI imagery is firstly used to calculate spectral factors and vegetation indices. Multiple machine learning algorithms are coupled to perform remote sensing estimation and spatial inversion of vegetation AGB based on measured samples of vegetation AGB. Secondly, an Orientation Distance-AGB (OD-AGB) curve is constructed outward from the center of subsidence water areas (SWA), with the Boltzmann function used for curve fitting. According to the location of the inflection point of the curve, the boundary points of vegetation disturbance are identified, and then the disturbance range is divided. The results show that (1) the TV-SVM model, utilizing total variables and support vector machine, achieves the highest estimation accuracy, with σ and σ values of 208.47 g/m and 290.19 g/m, respectively, for the validation set. (2) Thirty-six effective disturbance areas, totaling 29.89 km, are identified; the Boltzmann function provides a good fit for the OD-AGB curve, with an R exceeding 0.8 for typical disturbance areas. (3) Analysis of general statistical laws indicates that disturbance distance conforms to the general characteristics of normal distribution, exhibiting boundedness and directional heterogeneity. The research is expected to provide scientific guidance for hierarchical zoning management, land reclamation, and ecological restoration in coal mining areas with high groundwater table.
在高地下水位地区进行采煤会显著导致地表沉降和积水，从而对地表植被、土壤和水文资源造成重大损害。开发有效的方法来提取地表扰动信息有助于定量评估采煤对土地、生态和社会的综合影响。由于传统指标在反映采矿干扰方面的局限性，引入植被地上生物量（AGB）作为提取采矿干扰范围的主要指标。以淮北煤田为例，首先利用 Sentinel-2 MSI 影像计算光谱因子和植被指数。耦合多种机器学习算法，基于植被 AGB 的实测样本进行遥感估计和空间反演。其次，从沉降水区（SWA）中心向外构建方位距离-AGB（OD-AGB）曲线，采用玻尔兹曼函数进行曲线拟合。根据曲线拐点的位置，识别植被干扰的边界点，然后划分干扰范围。结果表明：（1）利用总变量和支持向量机的 TV-SVM 模型取得了最高的估计精度，验证集的 σ 和 σ 值分别为 208.47 g/m 和 290.19 g/m。（2）确定了 36 个有效的干扰区域，总面积为 29.89 km；OD-AGB 曲线与玻尔兹曼函数拟合良好，典型干扰区域的 R 值超过 0.8。（3）一般统计规律分析表明，干扰距离符合正态分布的一般特征，具有有界性和方向异质性。本研究有望为高地下水位采煤区的分层分区管理、土地复垦和生态恢复提供科学指导。
