Assessment of landslide susceptibility using DBSCAN-AHD and LD-EV methods

• 作者: MAO Yi-min MWAKAPESA Deborah S LI Yi-can XU Kai-bin NANEHKARAN Yaser A ZHANG Mao-sheng
• 来源期刊: 山地科学学报（英文版）
• 年，卷（期）: 2022,（1）
• 所属期刊栏目: Mountain Hazards
• 页码范围: 184-197
• 页数: 14页
• 语种: 英文

摘要

Landslide susceptibility assessment plays a vital role in understanding landslide information in advance and taking preventive as well as control measures.The need to initially specify the number of clusters,difficulty in handling noise and quantifying rainfall data,limits the application of traditional clustering models in landslide susceptibility assessment which then lowers their performance accuracy.Thus,to overcome these limitations,this study proposed an improved clustering algorithm titled the DBSCAN-AHD algorithm,which combines the traditional DBSCAN (Density Based Spatial Clustering of Application with Noise) algorithm and an Adaptive Hausdorff Distance (AHD) for landslide susceptibility modeling.Firstly,AHD was introduced to the traditional DBSCAN to quantify rainfall.Then,the DBSCAN-AHD grouped the mapping units with similar topology and geology characteristics into subclasses without specifying the number of clusters in advance for its ability to handle noise in the data.Furthermore,the LD-EV (landslide density or eigenvalues) approach was introduced to obtain the susceptibility levels using K-means algorithm.Finally,to verify the model's performance,statistical indices and the area under the curve (AUC) were applied and compared to the traditional DBSCAN,KPSO,K-means and Hierarchical Clustering algorithms,whereby,the proposed model outperformed the others.Also,the obtained susceptibility map can provide references in taking relevant preventive and control measures.