Assessment of Analysis Network Process and Heuristic Method in the Investigation of Landslide Potential in the Axis Range and Reservoir Dams (Case Study: Ghalea Chai Dam)

Extended Abstract
Introduction:
Natural disasters management requires local information in order to be ready against dangers and reduce their procedure. Hence, evaluation of landslide occurrence in the area which is prone to landslide due to geographical condition and human constructions is high crucial. Ghale Chai dam located at Ajabshir Watershed is one kind of such areas. So the aim of present investigation is to identify hillside instabilities and movements and their influencing factors to prevent their harmful effects on natural resources and other parts of economical and engineering development and recognize points with high prone of danger. Hence, the aim of present investigation is to assess analytical network process and Heuristic method in determining the landslide prone areas in range axis and reservoir of Ghale Chai dam of Ajabshir.
Methods and Materials:
The efficiency of network analysis process and logistic regression method were studied to investigate landslide potential in studied area dam. ANP model building requires the definition of elements and their assignment to clusters and a definition of their relationships (I.e. the connections between them indicating the flow of influence between the elements). Like AHP, ANP is founded on ratio scale measurements and pair wise comparisons of elements to divide priorities of selected alternatives. In addition relations among criteria and sub-criteria are included in evaluations, allowing dependencies both within a cluster (inner dependence) and between clusters (outer dependence) (Saaty: 2001). Pairwise comparison is now done, both for weighting clusters (criteria) and for estimating the direction and importance of influences between elements, numerically pictured as ratio scale in a so-called super matrix. Network analysis process was used for the first time in Iran in order to evaluate landslide, done using super decision and arc GIS software. However, to assess landslide susceptibility using heuristic method there are two common approaches: direct and indirect method. The first method applies direct assessment to interpret susceptibility in the field on the basis of detailed maps (geomorphological maps, for instance). The latter does not assess directly in the field, but via data integration techniques in any particular software. This study uses indirect heuristic method. Heuristic approach is a semi-qualitative method. Besides uses knowledge properties (expert opinions, previous research results or literature recommendations), it also uses index-based procedures such as simple ranking and rating or analytical hierarchy process (AHP) in assigning weight and creating model. Concerning this, scoring and weighting process are crucial to build a model in heuristic approach.
Results and Discussion:
Considering research questions, a three-layer network model composed of target layer, criteria layer and options layer was designed and organized in network analysis process. The priority of danger classes was determined based on their coefficients after doing paired comparisons among elements and clusters. Zoning map was classified in five classes from very high to very low. The weighting judgment process in pair-wise comparison gives a weight for every Influencing factor. From the calculation, the final criteria tree (with weight in 2 digits) was created. Bigger weights indicates that, the pertinent factor gives bigger influence toward the model .Aspect has the biggest contribution (0.2519), followed by distance to road and litology with value 0.1786 and 0.1747, respectively. On the other side, the lowest contribution is given by slope (0.0387), followed by Dem (0.0590). No negative weights in heuristic method. The inconsistency value is 0, 0 62194: smaller than 0, 1. It means, according to SMCE validation, the choosing process is consistent. No improper stage while positioning the factor based on its importance to another After running paired comparisons between elements and clusters the priority of the danger classes based on their significance was determined and the coefficients of the factors showed that the aspect factor has the maximum effect in occurrence of the landslides if the area .the zoning of map were classified in five classes of very high to very low risk class
Results:
Results obtained from this investigation indicated that from eight influential factors of landslide occurrence in the area, land use, height classes and domain direction have the highest influence in landslide occurrence. Moreover comparisons of distributed landslide proportion degrees with zoning maps of above mentioned models indicated that Heuristic model with 86.25 percent of proportion had better performance than network analysis. So from two statistical models obtained from two methods used in this study, statistical model obtained from Heuristic method administration (2nd equation) is selected and introduced as the best model. Moreover, considering results obtained from landslide danger zoning in studied area, using two early mentioned methods, it was conclude that 67.33% of zone total area has very high danger of landslide occurrence.
Keywords: Ghale Chai Dam, Analytic Network Process (ANP), Heuristic method, Landsat Satellite, Landslide.