Analysis of Optimal Buffer Distance for Linear Hazard Factors in Landslide Susceptibility Prediction

Lu Fang, Qian Wang, Jianping Yue and Yin Xing ()
Additional contact information
Lu Fang: School of Earth Science and Engineering, Hohai University, 8 Focheng West Road, Nanjing 211100, China
Qian Wang: Shandong Gold Group Penglai Mining Co., Ltd., Yantai 265621, China
Jianping Yue: School of Earth Science and Engineering, Hohai University, 8 Focheng West Road, Nanjing 211100, China
Yin Xing: School of Geography Science and Geomatics Engineering, Suzhou University of Science and Technology, Suzhou 215009, China

Sustainability, 2023, vol. 15, issue 13, 1-17

Abstract: A linear hazard-causing factor is the environmental element of landslide susceptibility prediction, and the setting of buffer distance of a linear hazard-causing factor has an important influence on the accuracy of landslide susceptibility prediction based on machine learning algorithms. A geographic information system (GIS) has generally been accepted in the correlation analysis between linear hazard-causing factors and landslides; the most common are statistical models based on buffer zone analysis and superposition analysis for linear causative factor distances and landslide counts. However, there is a problem in the process of model building: the buffer distance that is used to build the statistical model and its statistical results can appropriately reflect the correlation between the linear disaster-causing factors and landslides. To solve this problem, a statistical model of landslide density and distance of linear disaster-causing factors under different single-loop buffer distances was established based on Pearson’s method with 12 environmental factors, such as elevation, topographic relief, and distance from the water system and road, in Ruijin City, Jiangxi Province to obtain the most relevant single-loop buffer distance linear disaster-causing factor combinations; random forest (RF) machine learning models were then used to predict landslide susceptibility. Finally, the Kappa coefficient and the distribution characteristics of the susceptibility index were used to investigate the modeling laws. The analysis results indicate that the prediction accuracy of the most correlated single-loop buffer distance combination reaches 96.65%, the error rate of non-landslide points is 4.2%, and the error of landslide points is 11.3%, which is higher than the same single-loop buffer distance combination, confirming the reasonableness of the method of using correlation to obtain the linear disaster-causing factor buffer distance.

Keywords: landslide; linear hazard factors; buffer; correlation; optimal distance (search for similar items in EconPapers)
JEL-codes: O13 Q Q0 Q2 Q3 Q5 Q56 (search for similar items in EconPapers)
Date: 2023
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (1)

Downloads: (external link)
https://www.mdpi.com/2071-1050/15/13/10180/pdf (application/pdf)
https://www.mdpi.com/2071-1050/15/13/10180/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jsusta:v:15:y:2023:i:13:p:10180-:d:1180440

Access Statistics for this article

Sustainability is currently edited by Ms. Alexandra Wu

More articles in Sustainability from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().