Evaluating of Flood hazard potential using bivariate statistical analysis method (Case study: Aji Chai basin)
Document Type : Original Article
Authors
Dept of Geomorphology University of Tabriz
10.22034/gmpj.2024.429929.1473
Abstract
Introduction
Floods are the most important and abundant environmental hazards that cause yearly human and financial losses. Aji Chai basin, located in East Azerbaijan province, is prone to destructive floods due to special topographical conditions. The primary purpose of this study is to prepare a flood hazard potential map using the weight of evidence (WOE) statistical method. To achieve this aim, 18 parameters effective in flood occurrence were investigated. The investigated parameters were Elevation, Slope, Aspect, Topographic wetness index, Sediment transport index, Stream power index, earth curvature, Rainfall, Normalized Difference Vegetation Index, land use, Distance to dam, Distance to bridge, Distance to the river, River density, hydrological soil groups, Drainage texture, Geomorphology and lithology.
Methodology
Aji Chai basin is located at latitudes between 37° 41΄ and 38° 29΄ North, and at longitudes between 45° 48΄ and 47° 53΄ East. The study area, with an area of about 10985.9 Km2, is one of the largest sub-basins of the Urmia Lake basin. The most important river that drains the surface water of this basin is Aji Chai.
This study used the weight of evidence (WOE) statistical method to prepare a flood hazard map in the Aji Chai basin. The weight of evidence model is one of the bivariate statistical methods. In this method, the weight of each parameter class is calculated based on the presence or absence of flood in the desired class. The weighting of the classes was done using the location of the floods that occurred in the area. From 274 flood points, 70% were selected as training data and 30% as validation data.
Results and Discussion
The weighting results of different classes for each layer show that the 0-10% class has the highest weight in relation to the slope layer. Due to having more humidity, the northern slopes have the highest weight in terms of flood potential. About the parameter of distance to the river, as expected, the areas around the rivers have experienced many times more floods than other parts. Therefore, the 0-250 meter class had the most weight. In the investigation of the river density parameter, it was found that the areas with the highest drainage density are the most susceptible to floods. The weight calculation results for the classes of land use layer showed that the classes of agriculture, garden, and built areas have the most weight in the occurrence of floods in the area. Concerning the parameter of distance to the bridge, the areas around the bridges have a high potential for flooding due to human manipulations. Investigating the topographic wetness index layer showed that the areas with higher values of this index have a high potential for flooding.
Conclusion
The final map was obtained using the Raster Calculator tool and the product of the weight of the parameter classes in its information layers. The analysis of the final map showed that the downstream areas of the basin have a high potential for flooding. These parts mainly include low, flat, and low-sloping surfaces, which are the places where all the runoff formed in the region's highlands is accumulated. Therefore, these areas are constantly flooded with heavy rains and sudden melting of snow in the region's highlands. The important cities of the basin, such as Tabriz, Sarab, and Bostanabad, which are formed along the main rivers, are also located in high-risk areas, which shows the vulnerability of these cities when flooding occurs. The calculation of the area of each hazard class showed that 32% of the area is in high and very high classes in terms of flooding. The evaluation of the model's accuracy based on the ROC curve and the area under the curve (AUC) showed that the model's accuracy had an excellent performance in training data with a coefficient of 0.898.
- - - - - -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- ---- ---- - - - - - - - -- - - - -- -- --- --- -- -- - -- -- - - -- -- -- - -- --- -- - -- -- -- - - -- -- - --- - - - - -- -- - - -- - -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --- -- -- -- -- -- -- -- - -- -- -- -- -- --- - -- -- -- --- -- -- -- - -- -
Floods are the most important and abundant environmental hazards that cause yearly human and financial losses. Aji Chai basin, located in East Azerbaijan province, is prone to destructive floods due to special topographical conditions. The primary purpose of this study is to prepare a flood hazard potential map using the weight of evidence (WOE) statistical method. To achieve this aim, 18 parameters effective in flood occurrence were investigated. The investigated parameters were Elevation, Slope, Aspect, Topographic wetness index, Sediment transport index, Stream power index, earth curvature, Rainfall, Normalized Difference Vegetation Index, land use, Distance to dam, Distance to bridge, Distance to the river, River density, hydrological soil groups, Drainage texture, Geomorphology and lithology.
Methodology
Aji Chai basin is located at latitudes between 37° 41΄ and 38° 29΄ North, and at longitudes between 45° 48΄ and 47° 53΄ East. The study area, with an area of about 10985.9 Km2, is one of the largest sub-basins of the Urmia Lake basin. The most important river that drains the surface water of this basin is Aji Chai.
This study used the weight of evidence (WOE) statistical method to prepare a flood hazard map in the Aji Chai basin. The weight of evidence model is one of the bivariate statistical methods. In this method, the weight of each parameter class is calculated based on the presence or absence of flood in the desired class. The weighting of the classes was done using the location of the floods that occurred in the area. From 274 flood points, 70% were selected as training data and 30% as validation data.
Results and Discussion
The weighting results of different classes for each layer show that the 0-10% class has the highest weight in relation to the slope layer. Due to having more humidity, the northern slopes have the highest weight in terms of flood potential. About the parameter of distance to the river, as expected, the areas around the rivers have experienced many times more floods than other parts. Therefore, the 0-250 meter class had the most weight. In the investigation of the river density parameter, it was found that the areas with the highest drainage density are the most susceptible to floods. The weight calculation results for the classes of land use layer showed that the classes of agriculture, garden, and built areas have the most weight in the occurrence of floods in the area. Concerning the parameter of distance to the bridge, the areas around the bridges have a high potential for flooding due to human manipulations. Investigating the topographic wetness index layer showed that the areas with higher values of this index have a high potential for flooding.
Conclusion
The final map was obtained using the Raster Calculator tool and the product of the weight of the parameter classes in its information layers. The analysis of the final map showed that the downstream areas of the basin have a high potential for flooding. These parts mainly include low, flat, and low-sloping surfaces, which are the places where all the runoff formed in the region's highlands is accumulated. Therefore, these areas are constantly flooded with heavy rains and sudden melting of snow in the region's highlands. The important cities of the basin, such as Tabriz, Sarab, and Bostanabad, which are formed along the main rivers, are also located in high-risk areas, which shows the vulnerability of these cities when flooding occurs. The calculation of the area of each hazard class showed that 32% of the area is in high and very high classes in terms of flooding. The evaluation of the model's accuracy based on the ROC curve and the area under the curve (AUC) showed that the model's accuracy had an excellent performance in training data with a coefficient of 0.898.
- - - - - -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- ---- ---- - - - - - - - -- - - - -- -- --- --- -- -- - -- -- - - -- -- -- - -- --- -- - -- -- -- - - -- -- - --- - - - - -- -- - - -- - -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --- -- -- -- -- -- -- -- - -- -- -- -- -- --- - -- -- -- --- -- -- -- - -- -
Keywords
)