Crowdsourcing photograph locations for debris flow hot spot mapping

Hone Jay Chu, Yi Chin Chen

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

This study shows the feasibility of obtaining hazardous hot spot information on landslide and debris flow from crowdsourced data. Historical hazard or disaster photographs were voluntarily uploaded by the public to a Web photograph album. A total of 2245 hazard photographs from 1973 to 2015 were crowdsourced, and each photograph was tagged with geographical coordinates. After the removal of outliers, 96% of the photograph points were found within the 4 km potential debris flow buffer of existing databases, and none was found along the steep slopes with a mean of 14°. The photograph hot spot analysis using local Moran’s I or Gi* was identified statistically significant without subjective judgment. The DBSCAN model was also used to detect hot spot clusters effectively. The model parameters were nearly automatically generated on the basis of the count plot and the nearest neighbor distance graph. The results of these approaches were generally consistent with the hazardous hot spot maps and strongly related to central and southern Taiwan from the crowdsourced photograph data. Results reveal that the hot spot areas are found in areas with faults and near the potentially weak and fractured rocky regions. The majority of the landslides occur near the fault line because the strong ground motions triggered by an earthquake propagated along the fault rupture plane. Hot spot mapping using crowdsourced data can be used to estimate where debris flow will frequently occur and show how large the debris flow will be. Potentially hazardous areas can be effectively determined by the hot spot analysis of crowdsourced data.

Original languageEnglish
Pages (from-to)1259-1276
Number of pages18
JournalNatural Hazards
Volume90
Issue number3
DOIs
Publication statusPublished - 2018 Feb 1

Fingerprint

debris flow
photograph
hot spot
landslide
hazard
potential flow
outlier
ground motion
rupture
disaster
earthquake

All Science Journal Classification (ASJC) codes

  • Water Science and Technology
  • Atmospheric Science
  • Earth and Planetary Sciences (miscellaneous)

Cite this

@article{3fd5c1fe373a4e42a178f5dfac210a62,
title = "Crowdsourcing photograph locations for debris flow hot spot mapping",
abstract = "This study shows the feasibility of obtaining hazardous hot spot information on landslide and debris flow from crowdsourced data. Historical hazard or disaster photographs were voluntarily uploaded by the public to a Web photograph album. A total of 2245 hazard photographs from 1973 to 2015 were crowdsourced, and each photograph was tagged with geographical coordinates. After the removal of outliers, 96{\%} of the photograph points were found within the 4 km potential debris flow buffer of existing databases, and none was found along the steep slopes with a mean of 14°. The photograph hot spot analysis using local Moran’s I or Gi* was identified statistically significant without subjective judgment. The DBSCAN model was also used to detect hot spot clusters effectively. The model parameters were nearly automatically generated on the basis of the count plot and the nearest neighbor distance graph. The results of these approaches were generally consistent with the hazardous hot spot maps and strongly related to central and southern Taiwan from the crowdsourced photograph data. Results reveal that the hot spot areas are found in areas with faults and near the potentially weak and fractured rocky regions. The majority of the landslides occur near the fault line because the strong ground motions triggered by an earthquake propagated along the fault rupture plane. Hot spot mapping using crowdsourced data can be used to estimate where debris flow will frequently occur and show how large the debris flow will be. Potentially hazardous areas can be effectively determined by the hot spot analysis of crowdsourced data.",
author = "Chu, {Hone Jay} and Chen, {Yi Chin}",
year = "2018",
month = "2",
day = "1",
doi = "10.1007/s11069-017-3098-6",
language = "English",
volume = "90",
pages = "1259--1276",
journal = "Natural Hazards",
issn = "0921-030X",
publisher = "Springer Netherlands",
number = "3",

}

Crowdsourcing photograph locations for debris flow hot spot mapping. / Chu, Hone Jay; Chen, Yi Chin.

In: Natural Hazards, Vol. 90, No. 3, 01.02.2018, p. 1259-1276.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Crowdsourcing photograph locations for debris flow hot spot mapping

AU - Chu, Hone Jay

AU - Chen, Yi Chin

PY - 2018/2/1

Y1 - 2018/2/1

N2 - This study shows the feasibility of obtaining hazardous hot spot information on landslide and debris flow from crowdsourced data. Historical hazard or disaster photographs were voluntarily uploaded by the public to a Web photograph album. A total of 2245 hazard photographs from 1973 to 2015 were crowdsourced, and each photograph was tagged with geographical coordinates. After the removal of outliers, 96% of the photograph points were found within the 4 km potential debris flow buffer of existing databases, and none was found along the steep slopes with a mean of 14°. The photograph hot spot analysis using local Moran’s I or Gi* was identified statistically significant without subjective judgment. The DBSCAN model was also used to detect hot spot clusters effectively. The model parameters were nearly automatically generated on the basis of the count plot and the nearest neighbor distance graph. The results of these approaches were generally consistent with the hazardous hot spot maps and strongly related to central and southern Taiwan from the crowdsourced photograph data. Results reveal that the hot spot areas are found in areas with faults and near the potentially weak and fractured rocky regions. The majority of the landslides occur near the fault line because the strong ground motions triggered by an earthquake propagated along the fault rupture plane. Hot spot mapping using crowdsourced data can be used to estimate where debris flow will frequently occur and show how large the debris flow will be. Potentially hazardous areas can be effectively determined by the hot spot analysis of crowdsourced data.

AB - This study shows the feasibility of obtaining hazardous hot spot information on landslide and debris flow from crowdsourced data. Historical hazard or disaster photographs were voluntarily uploaded by the public to a Web photograph album. A total of 2245 hazard photographs from 1973 to 2015 were crowdsourced, and each photograph was tagged with geographical coordinates. After the removal of outliers, 96% of the photograph points were found within the 4 km potential debris flow buffer of existing databases, and none was found along the steep slopes with a mean of 14°. The photograph hot spot analysis using local Moran’s I or Gi* was identified statistically significant without subjective judgment. The DBSCAN model was also used to detect hot spot clusters effectively. The model parameters were nearly automatically generated on the basis of the count plot and the nearest neighbor distance graph. The results of these approaches were generally consistent with the hazardous hot spot maps and strongly related to central and southern Taiwan from the crowdsourced photograph data. Results reveal that the hot spot areas are found in areas with faults and near the potentially weak and fractured rocky regions. The majority of the landslides occur near the fault line because the strong ground motions triggered by an earthquake propagated along the fault rupture plane. Hot spot mapping using crowdsourced data can be used to estimate where debris flow will frequently occur and show how large the debris flow will be. Potentially hazardous areas can be effectively determined by the hot spot analysis of crowdsourced data.

UR - http://www.scopus.com/inward/record.url?scp=85033475793&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85033475793&partnerID=8YFLogxK

U2 - 10.1007/s11069-017-3098-6

DO - 10.1007/s11069-017-3098-6

M3 - Article

VL - 90

SP - 1259

EP - 1276

JO - Natural Hazards

JF - Natural Hazards

SN - 0921-030X

IS - 3

ER -