The multivariate logistic regression based cross application
approach was successfully used for the three studyregions: Penang Island, Cameron Highland, and Selangor.
The multivariate logistic regression model permitted to
determine the coefficients for the input layers and produce
nine sets of landslide hazard maps after the cross application
of the coefficients to the three study areas. This allows
to draw the following conclusions from the experience
gained in these study areas with similar geological and geomorphological
environment.
For the landslide hazard analysis and the establishment
of a landslide-related GIS database of all three study areas
landslide locations were mapped using aerial photographs,
field surveys and technical reports. For the landslide hazard
analysis, the multivariate logistic regression model,
was applied, validated, and cross-validated for the three
study areas using the landslide database. Then, the results
were validated by calculating the correlation between
actual landslide locations and probable occurrences.
The calculated coefficients based on the multivariate
logistic regression showed a similar trend for each study
area. Out of nine cases, among the three cases of the
application of logistic regression coefficients in the same
study area, the case of Selangor based on the Selangor
logistic regression coefficients showed the highest accuracy
(94%), where as Penang based on the Penang coefficients
showed the lowest accuracy (86%). Simialrly, among the
six cases from the cross application of logistic regression
coefficients in other two areas, the case of Selangor based
on logistic coefficients of Cameron showed highest (90%)
prediction accuracy where as the case of Penang based on
the Selangor logistic regression coefficients showed the
lowest accuracy (79%). Qualitatively, the cross application
model yields reasonable results which can be used for preliminary
landslide hazard mapping. Generally, however,
the validation results showed a satisfying agreement
between the hazard maps and the landslide locations verified
in the field.
In the present study, only the hazard analysis was performed,
because all the three study area had a good record
of rainfall data from the past. If, however, data on other
landslide-causing parameters such as earthquake shaking,
or slope cutting exist, then a probability analysis including
these values could also be made. Similarly, if factors relevant
to the vulnerability of buildings and other property
were available, a risk analysis could also be performed.
These landslide hazard maps are of great help for planners
and engineers to identify suitable locations for development.
These results can be used as basic data to assist slope
management and land-use planning.
Landslides are among the most hazardous natural disasters
in Malaysia. The Government and research institutionsare trying to analyze the landslide hazard and risk and to
show its spatial distribution over the regions. The use of
multi-temporal radar data such as TerraSAR for observing
the landslides and residues in the research phase could be
one of the prominent future directions. In the same line,
there is a lot of work to be done to investigate the landslide
causative parameters and their direct relationship between
the triggering of future landslides.
The multivariate logistic regression based cross applicationapproach was successfully used for the three studyregions: Penang Island, Cameron Highland, and Selangor.The multivariate logistic regression model permitted todetermine the coefficients for the input layers and producenine sets of landslide hazard maps after the cross applicationof the coefficients to the three study areas. This allowsto draw the following conclusions from the experiencegained in these study areas with similar geological and geomorphologicalenvironment.For the landslide hazard analysis and the establishmentof a landslide-related GIS database of all three study areaslandslide locations were mapped using aerial photographs,field surveys and technical reports. For the landslide hazardanalysis, the multivariate logistic regression model,was applied, validated, and cross-validated for the threestudy areas using the landslide database. Then, the resultswere validated by calculating the correlation betweenactual landslide locations and probable occurrences.The calculated coefficients based on the multivariatelogistic regression showed a similar trend for each studyarea. Out of nine cases, among the three cases of theapplication of logistic regression coefficients in the samestudy area, the case of Selangor based on the Selangorlogistic regression coefficients showed the highest accuracy(94%), where as Penang based on the Penang coefficientsshowed the lowest accuracy (86%). Simialrly, among the
six cases from the cross application of logistic regression
coefficients in other two areas, the case of Selangor based
on logistic coefficients of Cameron showed highest (90%)
prediction accuracy where as the case of Penang based on
the Selangor logistic regression coefficients showed the
lowest accuracy (79%). Qualitatively, the cross application
model yields reasonable results which can be used for preliminary
landslide hazard mapping. Generally, however,
the validation results showed a satisfying agreement
between the hazard maps and the landslide locations verified
in the field.
In the present study, only the hazard analysis was performed,
because all the three study area had a good record
of rainfall data from the past. If, however, data on other
landslide-causing parameters such as earthquake shaking,
or slope cutting exist, then a probability analysis including
these values could also be made. Similarly, if factors relevant
to the vulnerability of buildings and other property
were available, a risk analysis could also be performed.
These landslide hazard maps are of great help for planners
and engineers to identify suitable locations for development.
These results can be used as basic data to assist slope
management and land-use planning.
Landslides are among the most hazardous natural disasters
in Malaysia. The Government and research institutionsare trying to analyze the landslide hazard and risk and to
show its spatial distribution over the regions. The use of
multi-temporal radar data such as TerraSAR for observing
the landslides and residues in the research phase could be
one of the prominent future directions. In the same line,
there is a lot of work to be done to investigate the landslide
causative parameters and their direct relationship between
the triggering of future landslides.
การแปล กรุณารอสักครู่..
