Extended Abstract
Background: Floods are one of the most important natural hazards that annually inflict irreparable devastation across the country. Spatial analysis of flood hazard sensitivity and the preparation of hazard maps are important approaches in flood management. The Aji Chai Basin's large area and special geographical conditions make it one of the basins with high flooding hazards. As a result, the current study mainly aims to use a statistical model and multi-criteria decision-making techniques to create a flood hazard map for this basin. For this reason, flood hazard maps were created using 18 parameters affecting the occurrence of floods. The investigated parameters were Elevation, Slope, Aspect, Topographic wetness index, Sediment transport index, Stream power index, Earth curvature, Drainage texture, Rainfall, Distance to the river, River density, Lithology, Hydrological soil groups, Geomorphology, Distance to bridge, Distance to dam, Normalized Difference Vegetation Index, and Land use.
Methods: The study area of this research is the Aji Chai Basin, which is located in East Azerbaijan Province in terms of political divisions. The Aji Chai Basin is located in an almost rectangular shape between the Sabalan Mountain and the Ghoshe Dagh mountain range in the north, the Boz Gosh mountain range in the south, and the Sahand Mountain in the southwest. The area of this basin is about 10985.9 km2. The elevation changes of the basin are from 1255 m at the outlet of the basin to 3816 m on the slopes of Sabalan Mountain. In general, the Sabalan and Sahand mountains, with a height above 3600 m, are considered the most important topographic features in the region's roughness. The average annual rainfall of the Aji Chai Basin is about 315 mm based on the information from synoptic stations (Four stations, viz. Tabriz, Bostan Abad, Sarab, and Heris) and rain gauges (24 stations) available in the region.
To achieve the research aim, the Analytical Network Process (ANP) model was employed as a multi-criteria decision analysis method, and the Statistical Index (SI) model was utilized as a two-variable statistical method. The Analytical Network Process is one of the multi-criteria decision-making methods developed by Saaty in 1996. The statistical index (SI) method was introduced by Van Westen in 1997. The research models were implemented using the location of 274 flood points that happened in the past. The map of the location of flood points in the area was prepared through the information of the regional water company of East Azerbaijan Province, field survey, and the Landsat 8 satellite image of the OLI-TIRS sensor. The accuracy of the results was evaluated using three statistical indices, namely Sensitivity, Specificity, and Accuracy, along with the ROC curve and the area under the curve (AUC).
Results: The results of parameter weighting using the ANP model showed that the rainfall, geomorphology, and slope are the three parameters with the highest weight with coefficients of 0.137, 0.104, and 0.101, respectively, indicating the great influence of these factors on the occurrence of floods in the region. On the other hand, the sediment transport index and stream power index were the two parameters of the lowest weight. The evaluation of the importance of the parameters using the SI model also showed that areas near rivers and bridges and low-altitude and low-slope areas were susceptible to flood hazards. The final maps were prepared from the product of the weights of each of the parameters in their information layers and in five classes from very low to very high potential. Examining the final maps showed that the distribution pattern of hazard zones was similar in both models, and flat and plain surfaces were identified as areas with a high flooding hazard. The important cities of the basin, such as Tabriz, Sarab, and Bostanabad, are also in high and very high hazard classes, which shows the vulnerability of these cities when destructive floods occur. Since these cities are formed along the rivers, it shows the need for authorities to pay serious attention to urban flood management. On the other hand, the heights and steep slopes have the lowest potential for flooding.
Conclusion: Examining the area of each flood hazard class in the research models showed that about 34% and 46% of the areas of the region in the ANP and SI models, respectively, were in high and very high areas in terms of flooding. Examining the maps shows that the metropolis of Tabriz, which is considered the most important population center in the basin, is located in high and very high classes in terms of flooding due to its development along the Mehranroud and Aji Chai rivers. This shows the need for the serious attention of the regional authorities to manage the flood hazard in the basin as best as possible. The results of evaluating the accuracy of the models showed that the performance of both models was good in preparing the maps of the flood hazard potential in the region. Nevertheless, the SI model with a coefficient of 0.945 has the highest value of the AUC, which indicates the greater accuracy of this model compared to the ANP model.