Prioritization of Effective Factors on Landslide Occurrence and Mapping of its Sensitivity in CherikAbad Watershed, Urmia Using Shannon Entropy Model

Abstract

Landslides are natural and incidentally man-made phenomena. Developing landslide risk reduction plans in order to preserve the natural and human resources are pivotal. This study sets out to prioritize landslide-controlling factors by using the Shannon’s entropy model and the GIS techniques. A total of 90 landslides were identified using extensive filed surveys and interpreting the Google Earth images. A Landslide inventory map was prepared in the GIS environment together with eleven controlling factors as inputs, namely precipitation, elevation, slope percentage, slope aspect, lithology, land use/land cover, the normalized difference vegetation index (NDVI), and the distance to streams/faults/roads/villages. The results revealed that the land use/land cover, the distance to streams, and the distance to faults were the top-ranked factors of the highest importance in landslide occurrence, while the distance to villages, and the distance to roads were of the least importance. Further, the model validation results attested to the great performance of the adopted model where the area under the receiver operating characteristic curve (AUROC) was 0.879. Considering that about 32% of the watershed is located in the high and very high sensitive area, it is recommended to avoid land use change in landslide-prone areas to reduce relative risk of landslide.

Keywords

• cherikabad watershed
• gis
• landslide
• roc curve
• shannon’s entropy