基于MaxEnt和斜坡单元的崩滑灾害易发性评价——以云南省孟连县为例

EVALUATION OF LANDSLIDE SUSCEPTIBILITY BASED ON MAXENT AND SLOPE UNIT——TAKING MENGLIAN COUNTY OF YUNNAN PROVINCE AS AN EXAMPLE

为验证基于MaxEnt和斜坡单元崩滑灾害易发性评价模型的可行性与准确性,以云南省孟连县为例,选取高程、坡度、坡向、距断层距离、距公路距离、距河流距离、降水量、工程地质岩组、归一化植被指数(NDVI)9个评价因子,基于MaxEnt模型,对划分的800个斜坡单元进行易发性评价,将评价结果与基于栅格单元的评价结果进行对比。结果表明:(1)孟连县崩滑灾害极低易发区面积占37.44%;低易发区占29.08%;中易发区占17.67%;高易发区占11.26%;极高易发区占4.54%。(2)针对因子贡献率,基于斜坡单元和栅格单元的评价结果具有一定相似性,均说明人类工程活动和地质构造是孟连县崩滑灾害的主要诱因。(3)基于斜坡单元和栅格单元的评价结果AUC值分别为0.858和0.872,z统计量计算结果表明两者结果精确度相同,但基于斜坡单元更符合现实调查的实际意义。证明基于MaxEnt模型和斜坡单元的崩滑灾害易发性评价模型具有较好的可行性与准确性。

In order to verify the feasibility and accuracy of the evaluation model of landslide susceptibility based on MaxEnt and slope unit, taking Menglian County of Yunnan Province as an example, nine evaluation factors of elevation, slope, aspect, distance from fault, distance from highway, distance from river, precipitation, engineering geological rock group and normalized difference vegetation index( NDVI) were selected. Based on the MaxEnt model, the susceptibility of 800 slope units was evaluated, and the evaluation results were compared with the evaluation results based on grid unit.The results show that :(1) The extremely low susceptibility area of landslide disaster in Menglian County accounts for 37.44 %;low-prone areas accounted for 29.08 %;medium-prone areas accounted for 17.67 %;high-prone areas accounted for 11.26 %;the extremely high-prone areas accounted for 4.54 %.(2) According to the factor contribution rate, the evaluation results based on slope unit and grid unit have certain similarity, which shows that human engineering activities and geological structure are the main causes of landslide disaster in Menglian County.(3) The AUC values of the evaluation results based on the slope unit and the grid unit are 0.858 and 0.872, respectively. The calculation results of the z statistic show that the accuracy of the two results is the same, but the slope unit is more in line with the practical significance of the actual survey. It is proved that the evaluation model of landslide hazard susceptibility based on MaxEnt model and slope unit has good feasibility and accuracy.