Napathida Hunhaboon. Application of potential surface analysis and logistic regrssion model for landslide risk assessment in Uttaradit province. Master's Degree(Technology of Environmental Management ). Mahidol University. Mahidol University Library and Knowledge Center. : Mahidol University, 2009.
Application of potential surface analysis and logistic regrssion model for landslide risk assessment in Uttaradit province
Abstract:
The objective of this study was to evaluate landslide risk area in Uttaradit
province using the Geographic Information System integrated with Potential Surface
Analysis and logistic regression model. In additional, the models comparing with
actual landslide occurrence were also identified.
It was found that landslide risk area, manipulated by Potential Surface
Analysis using maximum daily rainfall 30 years return period has shown landslide risk
area and non-landslide risk area of 1,731.36 km2 (23.06%) and 5,778.77 km2
(76.94%). Meanwhile the landslide risk area manipulated by logistic regression model
using actual rainfall intensity on 22 May 2006 has shown landslide risk area and nonlandslide
risk area of 3,207.34 km2 (42.7%) and 4,302.8 km2 (57.29%). The overall
accuracy of the prediction by Potential Surface Analysis comparing with the
prediction by logistic regression model was 68.11% and Kappa index was 0.36. In
addition, the landslide risk area manipulated by Potential Surface Analysis comparing
with actual landslide area on 23 May 2006 has shown the overall accuracy of 80.09%
and Kappa index was 0.60. Meanwhile landslide risk area manipulated by logistic
regression model comparing with actual landslide area on 23 May 2006 has shown the
overall accuracy of 66.76% and Kappa index was 0.34.
Thus, landslide risk area identification using Potential Surface Analysis is
more appropriate than landslide risk area identification by logistic regression model.