Selection of the best statistical model for hazard zonation and assessment of landslide risk (Case study: Doab Samsami Watershed)

Document Type : Research

Authors

1 Assistant professor and faculty member of Chaharmahal and Bakhtiari Agricultural and Natural Resources Research Center

2 PhD in watershed management, Gorgan University of Agricultural Sciences and Natural Resources

3 Faculty member of the Research Institute of Soil Protection and Watershed Management

4 Expert of the General Department of Natural Resources of Chaharmahal and Bakhtiari Province

5 Expert of the General Department of Natural Resources of Chaharmahal and Bakhtiari province

10.22092/wmej.2016.109745

Abstract

The Doab Samsami watershed is located in Chaharmahal Va Bakhtiari province and is one of the main tributaries of the Karoon River basin. Landslide distribution map of the study area was prepared using air photos interoperation and field studies. 37 cases of happened landslides were observed. Nine parameters including elevation, slope, aspect, lithology, distances from fault, stream and road, land use, and annual precipitation were chosen as landslide effective factors. Potential landslide hazard maps were also provided using weighted density area bivariate statistical model (semi-quantitative model), multivariate stepwise regression model and logistic multivariate regression model and were compared by field data. Chi-square test was used to analyze difference between hazard classes of models. Kappa index was used to compare the conformity rate of the hazard maps. Qs and ROC Indexes was also used for evaluation of models accuracy. Hazard map, elements at risk map and vulnerability map were used in the preparation of risk map. Results show that measured Chi-square statistics is statistically significant at 95% of confidence interval, and there is an appropriate differentiation among landslide hazard classes. Kappa index between hazard maps of models vary between -1 to 0.2, indicating that there is a slight conformity between them. Weighted bivariate statistical model has been identified as the superior model for Doab Samsami watershed with Qs and ROC equal to 0.87 and 0.914.

Keywords

Watershed Management Research
Volume 28, Issue 4
January 2016
Pages 62-73

Files

History

  • Receive Date: 13 January 2015
  • Revise Date: 19 July 2015
  • Accept Date: 22 August 2015

Share

How to cite

Statistics