Empirical prediction of coseismic landslide dam formation

Xuanmei Fan, D.G. Rossiter, C.J. Van Westen, Qiang Xu, Tolga Görüm

Research output: Contribution to journalArticleAcademicpeer-review

26 Citations (Scopus)

Abstract

In this study we develop an empirical method to estimate the volume threshold for predicting coseismic landslide dam formation using landscape parameters obtained from digital elevation models (DEMs). We hypothesize that the potential runout and volume of landslides, together with river features, determine the likelihood of the formation of a landslide dam. To develop this method, a database was created by randomly selecting 140 damming and 200 non-damming landslides from 501 landslide dams and>60 000 landslides induced by the M w 7.9 2008 Wenchuan earthquake in China. We used this database to parameterize empirical runout models by stepwise multivariate regression. We find that factors controlling landslide runout are landslide initiation volume, landslide type, internal relief (H) and the H/L ratio (between H and landslide horizontal distance to river, L). In order to obtain a first volume threshold for a landslide to reach a river, the runout regression equations were converted into inverse volume equations by taking the runout to be the distance to river. A second volume threshold above which a landslide is predicted to block a river was determined by the correlation between river width and landslide volume of the known damming landslides. The larger of these two thresholds was taken as the final damming threshold. This method was applied to several landslide types over a fine geographic grid of assumed initiation points in a selected catchment. The overall prediction accuracy was 97.4% and 86.0% for non-damming and damming landslides, respectively. The model was further tested by predicting the damming landslides over the whole region, with promising results. We conclude that our method is robust and reliable for the Wenchuan event. In combination with pre-event landslide susceptibility and frequency-size assessments, it can be used to predict likely damming locations of future coseismic landslides, thereby helping to plan emergency response.
Original languageEnglish
Pages (from-to)1913-1926
Number of pages14
JournalEarth surface processes and landforms
Volume39
Issue number14
Early online date14 Apr 2014
DOIs
Publication statusPublished - Nov 2014

Fingerprint

landslide
dam
river
prediction
regression
empirical method
event
natural disaster
China
Sichuan earthquake 2008
digital elevation model

Keywords

  • Landslide dam
  • Multivariate regression
  • Runout prediction
  • Volume threshold
  • Wenchuan earthquake
  • ITC-ISI-JOURNAL-ARTICLE

Cite this

Fan, Xuanmei ; Rossiter, D.G. ; Van Westen, C.J. ; Xu, Qiang ; Görüm, Tolga. / Empirical prediction of coseismic landslide dam formation. In: Earth surface processes and landforms. 2014 ; Vol. 39, No. 14. pp. 1913-1926.
@article{dda0efc3538b40a7844f47469dbacdea,
title = "Empirical prediction of coseismic landslide dam formation",
abstract = "In this study we develop an empirical method to estimate the volume threshold for predicting coseismic landslide dam formation using landscape parameters obtained from digital elevation models (DEMs). We hypothesize that the potential runout and volume of landslides, together with river features, determine the likelihood of the formation of a landslide dam. To develop this method, a database was created by randomly selecting 140 damming and 200 non-damming landslides from 501 landslide dams and>60 000 landslides induced by the M w 7.9 2008 Wenchuan earthquake in China. We used this database to parameterize empirical runout models by stepwise multivariate regression. We find that factors controlling landslide runout are landslide initiation volume, landslide type, internal relief (H) and the H/L ratio (between H and landslide horizontal distance to river, L). In order to obtain a first volume threshold for a landslide to reach a river, the runout regression equations were converted into inverse volume equations by taking the runout to be the distance to river. A second volume threshold above which a landslide is predicted to block a river was determined by the correlation between river width and landslide volume of the known damming landslides. The larger of these two thresholds was taken as the final damming threshold. This method was applied to several landslide types over a fine geographic grid of assumed initiation points in a selected catchment. The overall prediction accuracy was 97.4{\%} and 86.0{\%} for non-damming and damming landslides, respectively. The model was further tested by predicting the damming landslides over the whole region, with promising results. We conclude that our method is robust and reliable for the Wenchuan event. In combination with pre-event landslide susceptibility and frequency-size assessments, it can be used to predict likely damming locations of future coseismic landslides, thereby helping to plan emergency response.",
keywords = "Landslide dam, Multivariate regression, Runout prediction, Volume threshold, Wenchuan earthquake, ITC-ISI-JOURNAL-ARTICLE",
author = "Xuanmei Fan and D.G. Rossiter and {Van Westen}, C.J. and Qiang Xu and Tolga G{\"o}r{\"u}m",
year = "2014",
month = "11",
doi = "10.1002/esp.3585",
language = "English",
volume = "39",
pages = "1913--1926",
journal = "Earth surface processes and landforms",
issn = "0197-9337",
publisher = "Wiley",
number = "14",

}

Empirical prediction of coseismic landslide dam formation. / Fan, Xuanmei; Rossiter, D.G.; Van Westen, C.J.; Xu, Qiang; Görüm, Tolga.

In: Earth surface processes and landforms, Vol. 39, No. 14, 11.2014, p. 1913-1926.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Empirical prediction of coseismic landslide dam formation

AU - Fan, Xuanmei

AU - Rossiter, D.G.

AU - Van Westen, C.J.

AU - Xu, Qiang

AU - Görüm, Tolga

PY - 2014/11

Y1 - 2014/11

N2 - In this study we develop an empirical method to estimate the volume threshold for predicting coseismic landslide dam formation using landscape parameters obtained from digital elevation models (DEMs). We hypothesize that the potential runout and volume of landslides, together with river features, determine the likelihood of the formation of a landslide dam. To develop this method, a database was created by randomly selecting 140 damming and 200 non-damming landslides from 501 landslide dams and>60 000 landslides induced by the M w 7.9 2008 Wenchuan earthquake in China. We used this database to parameterize empirical runout models by stepwise multivariate regression. We find that factors controlling landslide runout are landslide initiation volume, landslide type, internal relief (H) and the H/L ratio (between H and landslide horizontal distance to river, L). In order to obtain a first volume threshold for a landslide to reach a river, the runout regression equations were converted into inverse volume equations by taking the runout to be the distance to river. A second volume threshold above which a landslide is predicted to block a river was determined by the correlation between river width and landslide volume of the known damming landslides. The larger of these two thresholds was taken as the final damming threshold. This method was applied to several landslide types over a fine geographic grid of assumed initiation points in a selected catchment. The overall prediction accuracy was 97.4% and 86.0% for non-damming and damming landslides, respectively. The model was further tested by predicting the damming landslides over the whole region, with promising results. We conclude that our method is robust and reliable for the Wenchuan event. In combination with pre-event landslide susceptibility and frequency-size assessments, it can be used to predict likely damming locations of future coseismic landslides, thereby helping to plan emergency response.

AB - In this study we develop an empirical method to estimate the volume threshold for predicting coseismic landslide dam formation using landscape parameters obtained from digital elevation models (DEMs). We hypothesize that the potential runout and volume of landslides, together with river features, determine the likelihood of the formation of a landslide dam. To develop this method, a database was created by randomly selecting 140 damming and 200 non-damming landslides from 501 landslide dams and>60 000 landslides induced by the M w 7.9 2008 Wenchuan earthquake in China. We used this database to parameterize empirical runout models by stepwise multivariate regression. We find that factors controlling landslide runout are landslide initiation volume, landslide type, internal relief (H) and the H/L ratio (between H and landslide horizontal distance to river, L). In order to obtain a first volume threshold for a landslide to reach a river, the runout regression equations were converted into inverse volume equations by taking the runout to be the distance to river. A second volume threshold above which a landslide is predicted to block a river was determined by the correlation between river width and landslide volume of the known damming landslides. The larger of these two thresholds was taken as the final damming threshold. This method was applied to several landslide types over a fine geographic grid of assumed initiation points in a selected catchment. The overall prediction accuracy was 97.4% and 86.0% for non-damming and damming landslides, respectively. The model was further tested by predicting the damming landslides over the whole region, with promising results. We conclude that our method is robust and reliable for the Wenchuan event. In combination with pre-event landslide susceptibility and frequency-size assessments, it can be used to predict likely damming locations of future coseismic landslides, thereby helping to plan emergency response.

KW - Landslide dam

KW - Multivariate regression

KW - Runout prediction

KW - Volume threshold

KW - Wenchuan earthquake

KW - ITC-ISI-JOURNAL-ARTICLE

UR - https://ezproxy2.utwente.nl/login?url=https://doi.org/10.1002/esp.3585

UR - https://ezproxy2.utwente.nl/login?url=https://library.itc.utwente.nl/login/2014/isi/rossiter_emp.pdf

U2 - 10.1002/esp.3585

DO - 10.1002/esp.3585

M3 - Article

VL - 39

SP - 1913

EP - 1926

JO - Earth surface processes and landforms

JF - Earth surface processes and landforms

SN - 0197-9337

IS - 14

ER -