论文部分内容阅读
基于数字高程模型(DEM)计算得到的坡度、坡向等地形属性是滑坡危险性评价模型的重要输入数据,DEM误差会导致地形属性计算结果不确定性,进而影响滑坡危险性评价模型的结果。本文选择基于专家知识的滑坡危险性评价模型和逻辑斯第回归模型,采用蒙特卡洛模拟方法,研究DEM误差所导致的滑坡危险性评价模型结果不确定性。研究区位于长江中上游的重庆开县,采用5 m分辨率的DEM,以序贯高斯模拟方法模拟了不同大小(误差标准差为1 m、7.5 m、15 m)和空间自相关性(变程为0 m、30 m、60 m、120 m)的12类DEM误差场参与滑坡危险性评价。每次模拟包括100个实现,通过对每次模拟分别计算滑坡危险性评价结果的标准差图层和分类一致性百分比图层,用以评价结果不确定性。评价结果表明,在不同的DEM精度下,两个滑坡危险性评价模型所得结果的总体不确定性随空间自相关程度的变化趋势并不相同。当DEM空间自相关性程度不同时,基于专家知识的滑坡危险性评价模型的评价结果总体不确定随着DEM误差增加而呈现不同的变化趋势,而逻辑斯第回归模型的评价结果总体不确定性随着DEM误差大小增加而单调增加。从评价结果总体不确定性角度而言,总体上逻辑斯第回归模型比基于专家知识的滑坡危险性评价模型更加依赖于DEM数据质量。
Landform attributes such as slope and aspect based on digital elevation model (DEM) are important input data of landslide risk assessment model. DEM error will lead to uncertainty of terrain attribute calculation results, which will affect the result of landslide hazard assessment model. In this paper, the landslide risk assessment model based on expert knowledge and Logistic regression model are selected. The Monte Carlo simulation method is used to study the uncertainty of landslide risk assessment model caused by DEM error. The study area is located in Kaixian County, Chongqing, the middle and upper reaches of the Yangtze River. The DEM with 5 m resolution was used to simulate the spatial autocorrelation (with a range of error of 1 m, 7.5 m and 15 m) Which are 0 m, 30 m, 60 m and 120 m, are involved in landslide hazard assessment. Each simulation includes 100 implementations, and the standard deviation layer and the classification consistency percentage layer of landslide risk assessment results are calculated separately for each simulation to evaluate the uncertainty of the results. The evaluation results show that under different DEM accuracies, the overall uncertainty of the results obtained by the two landslide risk assessment models varies with the degree of spatial autocorrelation. When the degree of spatial autocorrelation of DEM is different, the overall uncertainty of evaluation results of landslide risk assessment model based on expert knowledge shows different trends with the increase of DEM error, while the overall uncertainty of the evaluation results of Logistic regression model Monotonically increase with increasing DEM error size. In terms of the overall uncertainty of the evaluation results, the Logistic regression model generally relies more on the DEM data quality than the landslide risk assessment model based on expert knowledge.