Production of Landslide Susceptibility Map Using Self Organizing Map (Som) (Case Study: Northwest Iran)

Author: Mehran Shaygan* and Marzieh Mokarram**

Journal Name:

Download PDF

Abstract

This study is aimed at performing landslide classification using Kohonen Self Organizing Map (SOM) which is implemented on Shuttle Radar Topography Mission (SRTM) digital elevation models (DEMs) with spatial resolution of 30 m in the parts of northwest Iran. Effective parameters for identification of areas susceptible to landslides consist of elevation, profile, plan, curvature, slope angle and slope aspect. After preparing maps for each of parameters in ArcGIS software, standardization was performed on each of the six layers. Then using SOM susceptible zones to landslide was determined. The results of SOM show that there are seven classes for landslide classification in the study area. Also the results showed that the data had high density and had correlation with each other so that it should be seen that the plan, slope and curvature are closely related to each other.

Keywords

Landslides, geographical information systems (GIS); landslide classification, Kohonen Self Organizing Map (SOM); Northwest Iran.

Conclusion

The aim of the study was to determine the effectiveness of SOM as a clustering tool for landslide classification. In SOM, according to qualitative data, the clustering tendencies of the landslides were investigated using six morphometric parameters (elevation, profile, plan, curvature, slope angle and slope aspect). The U- matrix showed that some of the data are closely related to each other, such as elevation and slope. In addition, considering that PC projection represents the amount of data relationship with each other, PC projection was used to determine the study's data had high density. The results showed that the data had high density and had correlation with each other so that it should be seen that the plan, slope and curvature are closely related to each other. Finally, using the labels in the SOM method, seven classes for the landslides were detected.

References

Using information about landslide occurrence can get accurate information about landslide hazard assessment and risk reduction (Dai et al., 2002). Thus, an accurate susceptibility mapping with different risk levels can be key information for a large variety of users (Fell et al., 2008). There are different methods for landslide susceptibility mapping such as probability and bivariate statistical modeling (Yalcin and Bulut 2008, Althuwaynee et al. 2012; Lee and Pradhan 2006; Youssef et al. 2009), multivariate statistics (Yilmaz, 2009; Yilmaz, 2010a and b). One of the method for preparing landslide mapping with different risk levels is self-organizing map (SOM). A type of artificial neural network (ANN) is SOM that is trained using unsupervised learning to produce a low-dimensional (typically two-dimensional), discretized representation of the input space of the training samples, called a map, and is so a method to do dimensionality reduction. Self-organizing maps differ from other arti

How to cite this article

Mehran Shaygan* and Marzieh Mokarram (2017). Production of Landslide Susceptibility Map Using Self Organizing Map (Som) (Case Study: Northwest Iran). Biological Forum – An International Journal 9(1): 111-117.