research data_heavy haul railway

The research data is derived from landslide investigations along the Shuohuang Railway, including historical records of 1,185 landslide points provided by the Natural Resources and Planning Bureau, as well as environmental and geological parameters extracted through remote sensing and GIS analysis. The initial dataset comprises 14 influencing factors, such as elevation, slope, road and river densities, lithology, fault density, NDVI, rainfall, SPI, and TWI. Pearson correlation analysis was applied to select 10 key factors, improving model performance and reducing data dimensionality. The dataset was divided into training, validation, and testing sets (70%, 10%, 20%), and the 3-Sigma rule was used to remove outliers, ensuring the reliability of model inputs. The landslide susceptibility zoning maps generated by the model outputs were spatially analyzed using GIS, providing a scientific basis for disaster risk management in complex geological regions.

本研究数据集源自朔黄铁路沿线滑坡调查，包含由自然资源和规划局提供的1185处历史滑坡点记录，以及通过遥感与地理信息系统（GIS）分析提取的环境与地质参数。初始数据集涵盖14项影响因子，包括高程、坡度、道路与河流密度、岩性、断层密度、归一化植被指数（NDVI）、降雨量、标准化降水指数（SPI）以及地形湿度指数（TWI）等。研究采用皮尔逊相关分析筛选出10项关键影响因子，以提升模型性能并降低数据维度。数据集按照70%、10%、20%的比例划分为训练集、验证集与测试集，并通过3σ准则（3-Sigma rule）剔除异常值，保障模型输入数据的可靠性。模型输出生成的滑坡敏感性分区图经地理信息系统开展空间分析后，可为复杂地质区域的灾害风险管理提供科学依据。