Identification of flash flood susceptible zones in a highly complex topography and altitude dependent climatically sensitive Himalayan River Basin

Abstract

<p>Flash floods are among the most destructive natural disasters causing extremely adverse impacts on the lives and livelihoods of people across the world. These events occur due to weather-dependent phenomena like cloudbursts or extreme rainfall characterized by a very short lead time for warning. In recent years, the Indian Himalayan state of Uttarakhand has been experiencing frequent flash flood disasters resulting in massive damage and losses in terms of life and property. To mitigate the damaging effects of these phenomena, there is a need to identify and spatially represent the surfaces/areas prone to excessive runoff due to flash floods. However, the dynamic nature of flash flooding, the complexity of the terrain, and altitude-dependent climatic sensitivity make predicting flooding sites in the region very difficult. Geospatial technology, advanced statistical techniques in conjunction with remotely sensed datasets can be potentially employed to identify the possible areas, which are susceptible to flash flooding. Mandakini River Basin (MRB) is among one of the most flash floods prone basins in Uttarakhand. In this study, Frequency Ratio (FR) and Index of Entropy (IOE) methods have been integrated to make a hybrid statistical model to calculate flash flood potential index (FFPI). Subsequently, assessment and identification of the flash flood susceptible zones were carried out for MRB. In this study, an inventory of locations where flash flood events had occurred in the past was prepared. 70% of these locations were utilized in the training sample and the remaining 30% in the testing (validation) sample. Furthermore, 15 flash flood conditioning factors were utilized for training and testing the model. The results of the model revealed that the areas with high and very high susceptibility account for approximately 9.7% and 17.4%, respectively of the entire study area. The performance assessment of the model was examined by Receiver Operating Characteristic (ROC) curve method for both training and validation event locations. The area under the curve (AUC) values obtained for the success and prediction rates were 0.871 and 0.847, respectively. The final output susceptibility map generated after the analysis depicts the study area in five (very low, low, medium, high, and very high) flash flood susceptibility zones.  As a contribution to devise appropriate basin management plans and mitigate the damage in the highly susceptible areas to flash floods, the present research results may be an important input to disaster governance.</p><p><strong>Keywords: </strong>Flash flood susceptibility; Flash flood potential index; Frequency Ratio; Index of Entropy; Indian Himalayas</p>