地理科学进展 ›› 2012, Vol. 31 ›› Issue (10): 1326-1333.doi: 10.11820/dlkxjz.2012.10.010

• 模型与方法 • 上一篇    下一篇

DEM误差对滑坡危险性评价模型的影响

包黎莉1,2, 秦承志1, 朱阿兴1, 呼雪梅1,2   

  1. 1. 中国科学院地理科学与资源研究所资源与环境信息系统国家重点实验室, 北京100101;
    2. 中国科学院研究生院, 北京100049
  • 收稿日期:2012-04-01 修回日期:2012-06-01 出版日期:2012-10-25 发布日期:2012-10-25
  • 通讯作者: 秦承志(1977-),男,副研究员,硕士生导师,中国地理学会会员,主要研究方向为数字地形分析。E-mail:qincz@lreis.ac.cn E-mail:qincz@lreis.ac.cn
  • 作者简介:包黎莉(1986-),女,硕士,中国地理学会会员,主要研究方向为数字地形分析在滑坡危险性评价模型中的应用。E-mail:baoll@lreis.ac.cn
  • 基金资助:

    中科院知识创新项目(KZCX2-YW-Q10-1-5);国家自然科学基金项目(40971235);中科院地理资源所优秀青年人才基金项目(2011RC203)

Effect of DEM Error on Landslide Susceptibility Mapping Models

BAO Lili1,2, QIN Chengzhi1, ZHU A-Xing1, HU Xuemei1,2   

  1. 1. State Key Lab of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100101, China;
    2. Graduate University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2012-04-01 Revised:2012-06-01 Online:2012-10-25 Published:2012-10-25

摘要: 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。

关键词: 不确定性, 滑坡危险性评价, 蒙特卡洛模拟, 数字高程模型(DEM), 误差

Abstract: Terrain attributes (such as slope gradient and slope aspect) computed from a gridded digital elevation model (DEM) are important input data for landslide susceptibility mapping. Elevation error in DEM can cause uncertainty in terrain attributes and further influence landslide susceptibility mapping. However little research has concerned about this issue. This paper studies this issue by a Monte Carlo simulation method for an expert knowledge-based approach to landslide susceptibility mapping and the logistic regression model, which are chosen as representatives of two main types of current landslide susceptibility mapping models, i.e. the expert- knowledge-based models and the multivariate statistical models. The study area is located in the Kaixian County of Chongqing Municipality, and belongs to the middle-upper reaches of the Yangtze River, China. The grid size of DEM is 5 m. Sequential Gaussian simulation was conducted to simulate a total of 12 DEM error fields with combinations of three error magnitudes (i.e., standard deviation values are 1, 7.5, and 15 m) and four spatial autocorrelation levels of elevation error (i.e., range values are 0, 30, 60, and 120 m). Each simulation included 100 realizations. Overall uncertainty of each landslide susceptibility mapping model associated with each simulated DEM error field was evaluated based on both a map of standard deviation of resulted landslide susceptibility from the simulation and a percentage map of classification consistency of landslide susceptibility from the simulation. The uncertainty assessment shows that the trends of the overall uncertainty of either landslide susceptibility mapping model changed with the spatial autocorrelation level of simulated DEM error are different under each simulated error magnitude. The trends of the overall uncertainty of expert knowledge-based model changed with increasing error magnitude are different under different spatial autocorrelation levels of simulated DEM error, while the overall uncertainty of logistic regression model monotonously increases as the magnitude of simulated DEM error increases. In general, the overall uncertainty of logistic regression model is more sensitive to DEM error magnitude than expert knowledge-based model.

Key words: DEM, error, landslide susceptibility mapping, Monte Carlo simulation, uncertainty