A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Landslide susceptibility mapping using backpropagation neural networks and logistic regression: The Sephidargole case study, Semnan, Iran
This paper documents a low cost qualitative evaluation scheme using artificial neural networks (ANN) and logistic regression (LR) in a GIS environment for landslide susceptibility mapping. Three categories of major factors responsible for landslide, i.e., geomorphometric (slope, slope aspect, altitude), geological (lithology, distance to faults) and environmental (landuse and vegetation cover, precipitation) were considered. The method was illustrated with a case study on Sephidargole area in Semnan province, Iran. The study area was divided in 86.4 × 86.4-m units and thematic layers for each factor were prepared using field data, available geological, topographical, aerial photos and landuse maps of the area. Feed forward backpropagation neural networks (BPNN) and feed forward LR were used to prepare landslide susceptibility maps. All seven factors were considered in BPNN but precipitation and distance to faults were excluded from the final analysis of LR model because these factors did not significantly add to the predictive power of LR. The overall success rate of BPNN and LR for landslide susceptibility mapping of the study area were 91.25 and 92.75%, respectively. Although, the overall success rate of LR is slightly higher than BPNN, results from landslide inventory maps show better agreement with results using the BPNN technique.
Landslide susceptibility mapping using backpropagation neural networks and logistic regression: The Sephidargole case study, Semnan, Iran
This paper documents a low cost qualitative evaluation scheme using artificial neural networks (ANN) and logistic regression (LR) in a GIS environment for landslide susceptibility mapping. Three categories of major factors responsible for landslide, i.e., geomorphometric (slope, slope aspect, altitude), geological (lithology, distance to faults) and environmental (landuse and vegetation cover, precipitation) were considered. The method was illustrated with a case study on Sephidargole area in Semnan province, Iran. The study area was divided in 86.4 × 86.4-m units and thematic layers for each factor were prepared using field data, available geological, topographical, aerial photos and landuse maps of the area. Feed forward backpropagation neural networks (BPNN) and feed forward LR were used to prepare landslide susceptibility maps. All seven factors were considered in BPNN but precipitation and distance to faults were excluded from the final analysis of LR model because these factors did not significantly add to the predictive power of LR. The overall success rate of BPNN and LR for landslide susceptibility mapping of the study area were 91.25 and 92.75%, respectively. Although, the overall success rate of LR is slightly higher than BPNN, results from landslide inventory maps show better agreement with results using the BPNN technique.
Landslide susceptibility mapping using backpropagation neural networks and logistic regression: The Sephidargole case study, Semnan, Iran
Khamehchiyan, M. (author) / Abdolmaleki, P. (author) / Rakei, B. (author)
Geomechanics and Geoengineering ; 6 ; 237-250
2011-09-01
14 pages
Article (Journal)
Electronic Resource
English
Presenting logistic regression-based landslide susceptibility results
| British Library Online Contents | 2018
| British Library Online Contents | 2005