Combination of Multi-criteria Decision-making Models and Regional Flood Analysis Technique to Prioritize Sub-watersheds for Flood Control (Case study: Dehbar Watershed of Khorasan)

1

The flood is one of the most important natural disasters, which is causing significant damage to affected areas. In flood management process, the factors which effectively responsible in flood formation are identified, and then areas with high potential for flood occurrence are identified. Because of the vast extent of catchment areas and the limited economic and administrative resources, the implementation of flood control projects in all flood-producing areas is not feasible. Therefore, prioritizing sub-watersheds is one of the chief measures for sustainable management of watersheds, with the purpose of controlling the flood. Multi-criteria decision-making (MCDM) methods such as the analytic hierarchy process (AHP) method and the analytic network process (ANP) are the techniques for identifying the areas with high flood-producing potential.
Since soil properties, infiltration rate, and quantitative geomorphological characteristics determine the amount of excess rainfall and runoff production, thus, the simultaneous application of morphometric analysis method and decision making models is very useful in the areas with data scarcity. In morphometric analysis, the physiographic and morphological characteristics of the watershed are analyzed based on the digital elevation model and finally the sub-watersheds are prioritized. Dividing large areas into multiple sub-watersheds and prioritizing these sub-watersheds reduce the time and cost of running watershed operations as well as making watershed projects more efficient.
The purpose of this study was to determine the sub-watersheds with critical conditions in terms of flooding risk in Dehbar watershed, Khorasan-e Razavi Province, Iran to reduce the costs of carrying out the watershed management projects focusing on flood control. It is worth noting that due to the lack of required data, morphometric and hydrologic analysis methods were used. In order to prioritize the sub-watersheds of Dehbar watershed, multi-criteria decision-making methods including AHP, VIKOR, and Permutation were employed. Afterwards, the results of these models were compared and verified by regional flood analysis method.
2

The study area was Dehbar watershed located in Torqabeh and Shandiz County, Khorasan-e Razavi Province, Iran. The area of the Dehbar watershed was estimated to be 115.73 km2. In order to better identify and evaluate runoff production capabilities, the watershed is divided into smaller hydrological units that have been separately investigated. This classification was made based on the location of the water resources, the location of the villages, the hydrographic network, the topographic contour lines, the satellite imagery, the field visit, and the integrative view in the GIS system, so that through the use of the ArcHydro extension in ArcMap, the Dehbar watershed was divided into 6 hydrologic and 4 non-hydrologic sub-watersheds.
The Dehbar watershed was divided into 10 sub-watersheds. In the current study, 13 evaluation criteria including area, compactness coefficient, drainage density, circularity factor, form factor, curve number, bifurcation ratio, main channel length, average slope, average height, time of concentration, rainfall and runoff coefficient were selected, and the amount of each for each sub-watershed was calculated. The weight of parameters was derived by the AHP technique. After determining the weights of the evaluation criteria and the preparation of decision matrix, VIKOR and Permutation models were employed for prioritization. After prioritizing, the regional flood analysis method (based on existing stations in the watershed) was used to compute maximum flood discharge in different return periods in order to evaluate and validate the considered models.
To this end, a homogeneous area with hydrometric stations was first identified based on geographical and climatic conditions within the region of the study area. The outliers were then eliminated and the frequency analysis was performed for each station individually and the best-fit statistical distribution was identified and selected. Finally, for regional flood analysis, a regression relation was acquired between peak discharges and contributing area in adjacent catchments; thus, it is possible to estimate peak discharge in the study area.
Finally, the outcomes of three employed multi- criteria decision making methods were combined using the average rating method.
3

Pairwise comparisons between considered criteria were performed based on AHP method and the relative weight of each criterion was obtained. Runoff coefficient with the relative weight of 0.221 had the highest importance among the considered criteria. Subsequently, rainfall criterion, time of concentration criterion, and curve number criterion were in the following ranks with the relative weights of 0.148, 0.116 and 0.109, respectively. The criteria of average elevation and form factor also had the lowest relative weights. Meanwhile, the inconsistency rate in AHP method was 0.04, indicating that the decision making process is consistent.
After determining the relative weights of each criterion for each sub-watershed, the VIKOR model was applied. According to this method, sub-watershed No.1 ranked first with Q index of 0.9715, sub-watershed No.3 ranked second with Q index of 0.8739, and sub-watershed No. 2 ranked third with Q index of 0.6030. Therefore, these sub-watersheds should gain high priority in watershed and flood control operations. Meanwhile, sub-watershed No. 7 with Q index of 0.0312, sub-watershed No. 10 with Q index of 0.0950 and sub-watershed No. 9 with Q index of 0.3132 were in tenth, ninth and eighth priority, respectively. Thus, these sub-watersheds had the lowest priority for implementing watershed management activities.
In the next step, the Permutation model was employed. According to the results of this model, sub-watersheds No. 1, No. 3 and No. 2 were ranked first to third, respectively. This is due to the high amount of rainfall, runoff coefficient and curve number in these sub-watersheds. Meanwhile, sub-watersheds 5, 8 and 10 were in the lowest ranks.
In addition, the results of regional flood analysis showed that sub-watersheds 1, 3, 8 and 2 had higher flood peak discharge, respectively. Meanwhile, sub-watersheds 5, 10, 4 and 9 had lower flood peak discharge and were in the last priority in terms of potential for flood generation.
In the final step, in order to provide a proper ranking for sub-watersheds, we used the average rating method to combine the obtained priorities by three different applied techniques. The outcomes showed that sub-watersheds No. 1, No. 3, and No. 2 were in the first rank, and therefore, in terms of the need for watershed management measures were in the top priority.
4

The results of this study showed that the derived priority of the sub-watersheds using morphometric and hydrological parameters as evaluation criteria to identify flood-producing areas is a suitable and appropriate method. Therefore, it is recommended that, in order to reduce costs and gain optimal outcomes, watershed management projects focusing on flood control should be implemented in the identified sub-watersheds which are in top priority in terms of generating flood discharge. The outcomes also showed that one or two factors alone could not determine the priority of flood-producing capability of sub-watersheds and a sub-watershed with a larger area does not necessarily have the higher potential for generating flood, but the interaction of different factors ultimately determines the priority of the sub-watershed in terms of flood-producing potential.