Spatial Landslide Hazard Prediction Using Rainfall Probability and a Logistic Regression Model

Saro Lee, Joong Sun Won, Seong Woo Jeon, Inhye Park, Moung Jin Lee

Research output: Contribution to journalArticle

17 Citations (Scopus)

Abstract

The aims of this study were to apply and validate a logistic regression model for landslide hazard, considering rainfall probability and using a geographic information system. The study focused on the Deokeokri and Karisanri areas of Inje, South Korea. We chose logistic regression for its mathematical rigor and its use of implementation in GIS software. Rainfall probability is analyzed for a quantitative prediction of rainfall changes in the study area. The rainfall probability was calculated using the Gumbel distribution. Then, the probabilities of landslides in the study area in target years (1, 3, 10, 50, and 100 years in the future) were calculated assuming that landslides are triggered by daily rainfall of 202 mm or 3-day cumulative rainfall of 449 mm. Landslide hazard maps were developed for the two study areas, and the logistic regression coefficients for one area were applied to the other area to validate the method. In Karisanri, all recorded landslides were used for validation. Validation results for the 202-mm daily precipitation threshold in Karisanri showed an average accuracy of 79.14 %, whereas those for the 449-mm 3-day cumulative precipitation threshold showed an average accuracy of 81.31 %. A combination of rainfall probability and logistic regression with a GIS is an effective method for analyzing the possibility of future landslides.

Original languageEnglish
Pages (from-to)565-589
Number of pages25
JournalMathematical Geosciences
Volume47
Issue number5
DOIs
Publication statusPublished - 2015 Jul 20

Fingerprint

Landslide
Logistic Regression Model
Rainfall
Hazard
landslide
logistics
hazard
rainfall
Prediction
prediction
Logistic Regression
GIS
Gumbel Distribution
Geographic Information Systems
Regression Coefficient
Choose
software
Target
Software

Keywords

  • GIS
  • Landslide hazard map
  • Logistic regression
  • Probability rainfall
  • Validation

ASJC Scopus subject areas

  • Mathematics (miscellaneous)
  • Earth and Planetary Sciences(all)

Cite this

Spatial Landslide Hazard Prediction Using Rainfall Probability and a Logistic Regression Model. / Lee, Saro; Won, Joong Sun; Jeon, Seong Woo; Park, Inhye; Lee, Moung Jin.

In: Mathematical Geosciences, Vol. 47, No. 5, 20.07.2015, p. 565-589.

Research output: Contribution to journalArticle

Lee, Saro ; Won, Joong Sun ; Jeon, Seong Woo ; Park, Inhye ; Lee, Moung Jin. / Spatial Landslide Hazard Prediction Using Rainfall Probability and a Logistic Regression Model. In: Mathematical Geosciences. 2015 ; Vol. 47, No. 5. pp. 565-589.
@article{cd5e4626f400425baf7273bc1b8cb003,
title = "Spatial Landslide Hazard Prediction Using Rainfall Probability and a Logistic Regression Model",
abstract = "The aims of this study were to apply and validate a logistic regression model for landslide hazard, considering rainfall probability and using a geographic information system. The study focused on the Deokeokri and Karisanri areas of Inje, South Korea. We chose logistic regression for its mathematical rigor and its use of implementation in GIS software. Rainfall probability is analyzed for a quantitative prediction of rainfall changes in the study area. The rainfall probability was calculated using the Gumbel distribution. Then, the probabilities of landslides in the study area in target years (1, 3, 10, 50, and 100 years in the future) were calculated assuming that landslides are triggered by daily rainfall of 202 mm or 3-day cumulative rainfall of 449 mm. Landslide hazard maps were developed for the two study areas, and the logistic regression coefficients for one area were applied to the other area to validate the method. In Karisanri, all recorded landslides were used for validation. Validation results for the 202-mm daily precipitation threshold in Karisanri showed an average accuracy of 79.14 {\%}, whereas those for the 449-mm 3-day cumulative precipitation threshold showed an average accuracy of 81.31 {\%}. A combination of rainfall probability and logistic regression with a GIS is an effective method for analyzing the possibility of future landslides.",
keywords = "GIS, Landslide hazard map, Logistic regression, Probability rainfall, Validation",
author = "Saro Lee and Won, {Joong Sun} and Jeon, {Seong Woo} and Inhye Park and Lee, {Moung Jin}",
year = "2015",
month = "7",
day = "20",
doi = "10.1007/s11004-014-9560-z",
language = "English",
volume = "47",
pages = "565--589",
journal = "Mathematical Geosciences",
issn = "1874-8961",
publisher = "Springer Netherlands",
number = "5",

}

TY - JOUR

T1 - Spatial Landslide Hazard Prediction Using Rainfall Probability and a Logistic Regression Model

AU - Lee, Saro

AU - Won, Joong Sun

AU - Jeon, Seong Woo

AU - Park, Inhye

AU - Lee, Moung Jin

PY - 2015/7/20

Y1 - 2015/7/20

N2 - The aims of this study were to apply and validate a logistic regression model for landslide hazard, considering rainfall probability and using a geographic information system. The study focused on the Deokeokri and Karisanri areas of Inje, South Korea. We chose logistic regression for its mathematical rigor and its use of implementation in GIS software. Rainfall probability is analyzed for a quantitative prediction of rainfall changes in the study area. The rainfall probability was calculated using the Gumbel distribution. Then, the probabilities of landslides in the study area in target years (1, 3, 10, 50, and 100 years in the future) were calculated assuming that landslides are triggered by daily rainfall of 202 mm or 3-day cumulative rainfall of 449 mm. Landslide hazard maps were developed for the two study areas, and the logistic regression coefficients for one area were applied to the other area to validate the method. In Karisanri, all recorded landslides were used for validation. Validation results for the 202-mm daily precipitation threshold in Karisanri showed an average accuracy of 79.14 %, whereas those for the 449-mm 3-day cumulative precipitation threshold showed an average accuracy of 81.31 %. A combination of rainfall probability and logistic regression with a GIS is an effective method for analyzing the possibility of future landslides.

AB - The aims of this study were to apply and validate a logistic regression model for landslide hazard, considering rainfall probability and using a geographic information system. The study focused on the Deokeokri and Karisanri areas of Inje, South Korea. We chose logistic regression for its mathematical rigor and its use of implementation in GIS software. Rainfall probability is analyzed for a quantitative prediction of rainfall changes in the study area. The rainfall probability was calculated using the Gumbel distribution. Then, the probabilities of landslides in the study area in target years (1, 3, 10, 50, and 100 years in the future) were calculated assuming that landslides are triggered by daily rainfall of 202 mm or 3-day cumulative rainfall of 449 mm. Landslide hazard maps were developed for the two study areas, and the logistic regression coefficients for one area were applied to the other area to validate the method. In Karisanri, all recorded landslides were used for validation. Validation results for the 202-mm daily precipitation threshold in Karisanri showed an average accuracy of 79.14 %, whereas those for the 449-mm 3-day cumulative precipitation threshold showed an average accuracy of 81.31 %. A combination of rainfall probability and logistic regression with a GIS is an effective method for analyzing the possibility of future landslides.

KW - GIS

KW - Landslide hazard map

KW - Logistic regression

KW - Probability rainfall

KW - Validation

UR - http://www.scopus.com/inward/record.url?scp=84935023569&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84935023569&partnerID=8YFLogxK

U2 - 10.1007/s11004-014-9560-z

DO - 10.1007/s11004-014-9560-z

M3 - Article

AN - SCOPUS:84935023569

VL - 47

SP - 565

EP - 589

JO - Mathematical Geosciences

JF - Mathematical Geosciences

SN - 1874-8961

IS - 5

ER -