- مبلغ: ۸۶,۰۰۰ تومان
- مبلغ: ۹۱,۰۰۰ تومان
Landslides in mountainous areas render major damages to residential areas, roads, and farmlands. Hence, one of the basic measures to reduce the possible damage is by identifying landslide-prone areas through landslide mapping by different models and methods. The purpose of conducting this study is to evaluate the efficacy of a combination of two models of the analytical network process (ANP) and fuzzy logic in landslide risk mapping in the Azarshahr Chay basin in northwest Iran. After field investigations and a review of research literature, factors affecting the occurrence of landslides including slope, slope aspect, altitude, lithology, land use, vegetation density, rainfall, distance to fault, distance to roads, distance to rivers, along with a map of the distribution of occurred landslides were prepared in GIS environment. Then, fuzzy logic was used for weighting sub-criteria, and the ANP was applied to weight the criteria. Next, they were integrated based on GIS spatial analysis methods and the landslide risk map was produced. Evaluating the results of this study by using receiver operating characteristic curves shows that the hybrid model designed by areas under the curve 0.815 has good accuracy. Also, according to the prepared map, a total of 23.22% of the area, amounting to 105.38 km2 , is in the high and very high-risk class. Results of this research are great of importance for regional planning tasks and the landslide prediction map can be used for spatial planning tasks and for the mitigation of future hazards in the study area.
6. Conclusion and future work
Our research aimed to integrate fuzzy set theory with ANP-MCDA for landslide mapping. We introduced an approach that integrates fuzzy set theory and information theory algorithms which could be a useful geospatial tool for integrating multiple features/attributes that affect the landslide mapping process. In conclusion, the work has explored an integrated approach for combining spatial data in a fuzzy-ANP based multi-criteria evaluation of landslide mapping. The approach described could significantly improve the results of GISMCDA based modelling. Based on the results achieved from this research, future research is foreseen, which will include the application of the ANP and spatially explicit reliability models for spatial sensitivity and uncertainty analyses of GIS-MCDA. Our future work will include applying the neural networks and comparing with frequency ratio and bivariate logistic regression modelling for landslide risk mapping. We also aim to study the functionality of these approaches by assessing their results through certainty analyses methods. Finally, we conclude the importance of accuracy in landslide susceptibility maps, for variety of applications especially when they are used as a basis for decision-making plans in light of reducing and mitigating the further hazards. The information provided by these maps shall help citizens, planners, and engineers to reduce losses caused by existing and future landslides by means of prevention, mitigation, and avoidance.