三维确定性模型在浅层黄土滑坡稳定性预测中的应用

浅层黄土滑坡是黄土高原广泛分布和频繁发生的地质灾害,造成了巨大的人员伤亡和财产损失.尽管二维确定性模型已被广泛用于浅层滑坡稳定性预测,但不能充分考虑岩土性质、地层结构、地下水等条件的三维空间变化,这可能与实际的斜坡稳定性不相符.因此,利用能考虑复杂斜坡环境的三维确定性模型评价滑坡稳定性,对获取更真实的评价结果以及指导滑坡防治工作具有重要意义.本文利用Scoops3D三维确定性模型评价了在浅层黄土滑坡稳定性预测中的适应性和可靠性.首先,模型计算参数敏感性的分析发现黏聚力、滑动视倾角和栅格单元重量对安全系数准确度影响较大,并用于指导获取详细的关键参数.然后,选取不同分辨率的数字高程模型(DEM)数据,利用Scoops3D模型对典型黄土沟壑中的浅层黄土滑坡稳定性进行预测,并通过详细的点状和面状滑坡分布图与预测结果的对比发现,该模型对黄土沟壑区的浅层滑坡稳定性预测准确度较高,且点状滑坡分布图可能更适合模型适应性的检验.最后,混淆矩阵法和成功率曲线法对不同分辨率数字高程模型预测结果可靠性的检验显示,该模型能够有效地预测黄土浅层滑坡的稳定性,且在高分辨率数字高程模型数据下可以获得可靠的预测精度. 

Application of a 3D deterministic model for predicting shallow loess landslide stability

In the loess plateau, shallow loess landslides are especially widespread and frequent geological disasters, causing serious casualties as well as huge property damage. Although two-dimensional deterministic models are widely applied to assess the stability of shallow landslides, they could not sufficiently consider the three-dimensional spatial variation of geotechnical property, layering configurations and groundwater. It might not conform with the actual situation of slope stability. Therefore, the three-dimensional deterministic model with considering complicated slope situation has the great significance to acquire the results that are more accordant with the actual situation. At the same time, it will exercise a profound and far-reaching influence to effectively mitigate landslide disasters. This paper takes the three-dimensional deterministic model Scoops3D to evaluate its adaptation and reliability of predicting shallow loess landslide stability. Firstly, the sensitivity analysis of model calculating parameters indicates that the most influential parameters on accuracy of safety factor are cohesive force, sliding direction of visual angle and the weight of grids, so it could guide to acquire the detail key input parameters. Then, the different resolution digital elevation models (DEMs) and geotechnical parameters are selected and used to predict the stability of shallow loess landslides in the typical gully and ridge physiographic region by using the Scoops3D. Comparing the calculated results with the detail inventory of point landslides and facial shape landslides shows that this model has a high accuracy in predicting shallow landslide stability. At the same time, the inventory of point landslides may be more suitable to model verification than facial shape landslides. Finally, the confusion matrix and the success rate curve are used to examine the reliability of predicted results that based on different resolution DEMs. The results prove that this model has a good adaptation to predict the stability of shallow loess landslides in the selected study area. Meanwhile, it could obtain reliable prediction accuracy with the high-resolution DEM data.