Hierarchical probabilistic regionalization of volcanism for Sengan region, Japan

Pinnaduwa Kulatilake, Jinyong Park, Pirahas Balasingam, Sean A. Mckenna

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

A 1 km square regular grid system created on the Universal Transverse Mercator zone 54 projected coordinate system is used to work with volcanism related data for Sengan region. The following geologic variables were determined as the most important for identifying volcanism: geothermal gradient, groundwater temperature, heat discharge, groundwater pH value, presence of volcanic rocks and presence of hydrothermal alteration. Data available for each of these important geologic variables were used to perform directional variogram modeling and kriging to estimate geologic variable vectors at each of the 23949 centers of the chosen 1 km cell grid system. Cluster analysis was performed on the 23949 complete variable vectors to classify each center of 1 km cell into one of five different statistically homogeneous groups with respect to potential volcanism spanning from lowest possible volcanism to highest possible volcanism with increasing group number. A discriminant analysis incorporating Bayes' theorem was performed to construct maps showing the probability of group membership for each of the volcanism groups. The said maps showed good comparisons with the recorded locations of volcanism within the Sengan region. No volcanic data were found to exist in the group 1 region. The high probability areas within group 1 have the chance of being the no volcanism region. Entropy of classification is calculated to assess the uncertainty of the allocation process of each 1 km cell center location based on the calculated probabilities. The recorded volcanism data are also plotted on the entropy map to examine the uncertainty level of the estimations at the locations where volcanism exists. The volcanic data cell locations that are in the high volcanism regions (groups 4 and 5) showed relatively low mapping estimation uncertainty. On the other hand, the volcanic data cell locations that are in the low volcanism region (group 2) showed relatively high mapping estimation uncertainty. The volcanic data cell locations that are in the medium volcanism region (group 3) showed relatively moderate mapping estimation uncertainty. Areas of high uncertainty provide locations where additional site characterization resources can be spent most effectively. The new data collected can be added to the existing database to perform future regionalized mapping and reduce the uncertainty level of the existing estimations.

Original languageEnglish (US)
Pages (from-to)79-102
Number of pages24
JournalGeotechnical and Geological Engineering
Volume25
Issue number1
DOIs
StatePublished - Feb 2007

Fingerprint

volcanic activity
regionalization
volcanism
Japan
uncertainty
Groundwater
Entropy
Volcanic rocks
cells
entropy
Cluster analysis
Discriminant analysis
Uncertainty
groundwater
volcanic rocks
site characterization
kriging
geothermal gradient
variogram
Bayesian theory

Keywords

  • Cluster analysis
  • Discriminant analysis
  • Entropy
  • Japan
  • Kriging
  • Probability
  • Regionalized mapping
  • Sengan
  • Variogram modeling
  • Volcanism

ASJC Scopus subject areas

  • Architecture
  • Geology
  • Soil Science

Cite this

Hierarchical probabilistic regionalization of volcanism for Sengan region, Japan. / Kulatilake, Pinnaduwa; Park, Jinyong; Balasingam, Pirahas; Mckenna, Sean A.

In: Geotechnical and Geological Engineering, Vol. 25, No. 1, 02.2007, p. 79-102.

Research output: Contribution to journalArticle

Kulatilake, Pinnaduwa ; Park, Jinyong ; Balasingam, Pirahas ; Mckenna, Sean A. / Hierarchical probabilistic regionalization of volcanism for Sengan region, Japan. In: Geotechnical and Geological Engineering. 2007 ; Vol. 25, No. 1. pp. 79-102.
@article{a48e9e49db354f4886224679d6193ceb,
title = "Hierarchical probabilistic regionalization of volcanism for Sengan region, Japan",
abstract = "A 1 km square regular grid system created on the Universal Transverse Mercator zone 54 projected coordinate system is used to work with volcanism related data for Sengan region. The following geologic variables were determined as the most important for identifying volcanism: geothermal gradient, groundwater temperature, heat discharge, groundwater pH value, presence of volcanic rocks and presence of hydrothermal alteration. Data available for each of these important geologic variables were used to perform directional variogram modeling and kriging to estimate geologic variable vectors at each of the 23949 centers of the chosen 1 km cell grid system. Cluster analysis was performed on the 23949 complete variable vectors to classify each center of 1 km cell into one of five different statistically homogeneous groups with respect to potential volcanism spanning from lowest possible volcanism to highest possible volcanism with increasing group number. A discriminant analysis incorporating Bayes' theorem was performed to construct maps showing the probability of group membership for each of the volcanism groups. The said maps showed good comparisons with the recorded locations of volcanism within the Sengan region. No volcanic data were found to exist in the group 1 region. The high probability areas within group 1 have the chance of being the no volcanism region. Entropy of classification is calculated to assess the uncertainty of the allocation process of each 1 km cell center location based on the calculated probabilities. The recorded volcanism data are also plotted on the entropy map to examine the uncertainty level of the estimations at the locations where volcanism exists. The volcanic data cell locations that are in the high volcanism regions (groups 4 and 5) showed relatively low mapping estimation uncertainty. On the other hand, the volcanic data cell locations that are in the low volcanism region (group 2) showed relatively high mapping estimation uncertainty. The volcanic data cell locations that are in the medium volcanism region (group 3) showed relatively moderate mapping estimation uncertainty. Areas of high uncertainty provide locations where additional site characterization resources can be spent most effectively. The new data collected can be added to the existing database to perform future regionalized mapping and reduce the uncertainty level of the existing estimations.",
keywords = "Cluster analysis, Discriminant analysis, Entropy, Japan, Kriging, Probability, Regionalized mapping, Sengan, Variogram modeling, Volcanism",
author = "Pinnaduwa Kulatilake and Jinyong Park and Pirahas Balasingam and Mckenna, {Sean A.}",
year = "2007",
month = "2",
doi = "10.1007/s10706-006-0008-1",
language = "English (US)",
volume = "25",
pages = "79--102",
journal = "International Journal of Mining Engineering",
issn = "0263-4546",
publisher = "Springer Netherlands",
number = "1",

}

TY - JOUR

T1 - Hierarchical probabilistic regionalization of volcanism for Sengan region, Japan

AU - Kulatilake, Pinnaduwa

AU - Park, Jinyong

AU - Balasingam, Pirahas

AU - Mckenna, Sean A.

PY - 2007/2

Y1 - 2007/2

N2 - A 1 km square regular grid system created on the Universal Transverse Mercator zone 54 projected coordinate system is used to work with volcanism related data for Sengan region. The following geologic variables were determined as the most important for identifying volcanism: geothermal gradient, groundwater temperature, heat discharge, groundwater pH value, presence of volcanic rocks and presence of hydrothermal alteration. Data available for each of these important geologic variables were used to perform directional variogram modeling and kriging to estimate geologic variable vectors at each of the 23949 centers of the chosen 1 km cell grid system. Cluster analysis was performed on the 23949 complete variable vectors to classify each center of 1 km cell into one of five different statistically homogeneous groups with respect to potential volcanism spanning from lowest possible volcanism to highest possible volcanism with increasing group number. A discriminant analysis incorporating Bayes' theorem was performed to construct maps showing the probability of group membership for each of the volcanism groups. The said maps showed good comparisons with the recorded locations of volcanism within the Sengan region. No volcanic data were found to exist in the group 1 region. The high probability areas within group 1 have the chance of being the no volcanism region. Entropy of classification is calculated to assess the uncertainty of the allocation process of each 1 km cell center location based on the calculated probabilities. The recorded volcanism data are also plotted on the entropy map to examine the uncertainty level of the estimations at the locations where volcanism exists. The volcanic data cell locations that are in the high volcanism regions (groups 4 and 5) showed relatively low mapping estimation uncertainty. On the other hand, the volcanic data cell locations that are in the low volcanism region (group 2) showed relatively high mapping estimation uncertainty. The volcanic data cell locations that are in the medium volcanism region (group 3) showed relatively moderate mapping estimation uncertainty. Areas of high uncertainty provide locations where additional site characterization resources can be spent most effectively. The new data collected can be added to the existing database to perform future regionalized mapping and reduce the uncertainty level of the existing estimations.

AB - A 1 km square regular grid system created on the Universal Transverse Mercator zone 54 projected coordinate system is used to work with volcanism related data for Sengan region. The following geologic variables were determined as the most important for identifying volcanism: geothermal gradient, groundwater temperature, heat discharge, groundwater pH value, presence of volcanic rocks and presence of hydrothermal alteration. Data available for each of these important geologic variables were used to perform directional variogram modeling and kriging to estimate geologic variable vectors at each of the 23949 centers of the chosen 1 km cell grid system. Cluster analysis was performed on the 23949 complete variable vectors to classify each center of 1 km cell into one of five different statistically homogeneous groups with respect to potential volcanism spanning from lowest possible volcanism to highest possible volcanism with increasing group number. A discriminant analysis incorporating Bayes' theorem was performed to construct maps showing the probability of group membership for each of the volcanism groups. The said maps showed good comparisons with the recorded locations of volcanism within the Sengan region. No volcanic data were found to exist in the group 1 region. The high probability areas within group 1 have the chance of being the no volcanism region. Entropy of classification is calculated to assess the uncertainty of the allocation process of each 1 km cell center location based on the calculated probabilities. The recorded volcanism data are also plotted on the entropy map to examine the uncertainty level of the estimations at the locations where volcanism exists. The volcanic data cell locations that are in the high volcanism regions (groups 4 and 5) showed relatively low mapping estimation uncertainty. On the other hand, the volcanic data cell locations that are in the low volcanism region (group 2) showed relatively high mapping estimation uncertainty. The volcanic data cell locations that are in the medium volcanism region (group 3) showed relatively moderate mapping estimation uncertainty. Areas of high uncertainty provide locations where additional site characterization resources can be spent most effectively. The new data collected can be added to the existing database to perform future regionalized mapping and reduce the uncertainty level of the existing estimations.

KW - Cluster analysis

KW - Discriminant analysis

KW - Entropy

KW - Japan

KW - Kriging

KW - Probability

KW - Regionalized mapping

KW - Sengan

KW - Variogram modeling

KW - Volcanism

UR - http://www.scopus.com/inward/record.url?scp=33846166067&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33846166067&partnerID=8YFLogxK

U2 - 10.1007/s10706-006-0008-1

DO - 10.1007/s10706-006-0008-1

M3 - Article

VL - 25

SP - 79

EP - 102

JO - International Journal of Mining Engineering

JF - International Journal of Mining Engineering

SN - 0263-4546

IS - 1

ER -