A small-scale landslide susceptibility assessment for the territory of Western Carpathians


Holec J., Bednarik M., Sabo M., Minar J., YILMAZ I., Marschalko M.

NATURAL HAZARDS, cilt.69, sa.1, ss.1081-1107, 2013 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 69 Sayı: 1
  • Basım Tarihi: 2013
  • Doi Numarası: 10.1007/s11069-013-0751-6
  • Dergi Adı: NATURAL HAZARDS
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Sayfa Sayıları: ss.1081-1107
  • Anahtar Kelimeler: Landslide, Susceptibility map, Bivariate, Multivariate analysis, Validation processes, Western Carpathians, ARTIFICIAL NEURAL-NETWORKS, LOGISTIC-REGRESSION, STATISTICAL-ANALYSIS, CONDITIONAL-PROBABILITY, SAMPLING STRATEGIES, GIS, HAZARD, AREA, FUZZY, MULTIVARIATE
  • Sivas Cumhuriyet Üniversitesi Adresli: Evet

Özet

This study presented herein compares the bivariate and multivariate landslide susceptibility mapping methods and presents the landslide susceptibility map of the territory of Western Carpathians in small scale. This study also describes pioneer work for the territory of Western Carpathians, overreaching state borders, using verified sophisticated statistical methods. In the susceptibility mapping, digital elevation model was first constructed using a GIS software, and parameter maps affecting the slope stability such as geology, seismicity, precipitation, topographical elevation, slope angle, slope aspect and land cover were considered. In the last stage of the analyses, landslide susceptibility maps were produced using bivariate and multivariate analyses, and they were then compared by means of their validations. The validation of the bivariate analysis data was performed using the results of bivariate analysis for landslide areas of Slovakia containing five classes of susceptibility in scale 1:500,000. The validation area is the area of Western Carpathians within Slovakia. Eighty-two per cent of area does not differ in more than one class. The validation of the multivariate analysis data was performed using the results from the Kysuce region in the northern part of Slovakia in scale 1:10,000. The raster calculator was used to express the difference between each pair of pixels within these two layers. Seventy-seven per cent of the pixels do not differ in more than 25 %, 94 % of the pixels do not differ in more than 50 %. The maximal possible difference is 100 % (one pixel with value 0 and other with value 1, or vice versa). Receiver operating characteristic analysis was also performed, the area under curve value for bivariate model was calculated to be 0.735, while it was 0.823 for multivariate. The results of the validation can be considered as satisfactory.