Open AccessEvaluation of Landslide Susceptibility Based on Logistic Regression Model
Author(s) -
Zhen Du,
Biao Zhang,
Hong Hu,
Junji Bao,
Wenbin Li
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.179
H-Index - 26
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/440/5/052004
Subject(s) - landslide , logistic regression , collinearity , elevation (ballistics) , lithology , curvature , geology , support vector machine , regression analysis , statistics , geomorphology , computer science , mathematics , machine learning , geometry , paleontology
A lot of methods can be used for landslide susceptibility evaluation, such as support vector machine model, artificial neural network, etc. These models have good modeling effect, but often have the problem of low modeling efficiency. Hence, this paper proposes a simple and effective model of landslide susceptibility evaluation - Logistic regression model. The Ningdu county of Jiangxi province in China, with 297 recorded landslides, was used as study case. The 6 environmental factors including elevation, slope, profile curvature, distance to rivers, lithology and NDVI were extracted in this study. The analysis showed that the significance of Profile curvature was greater than 0.05, and there was a collinearity problem, so it was excluded. After the establishment of the factor evaluation system, the prediction rate curve is used to evaluate the accuracy of the model. The results show that the AUC value of the prediction rate curve of logistic regression model is 0.864, indicating that the evaluation accuracy of logistic regression model is high and the modeling is reasonable. In addition, landslides in the study area are mainly distributed along both sides of the rivers, and elevation and lithology play a major role in the occurrence of landslides.