Developing a regional model to forecast human fatalities during flash floods in urban areas and its uncertainty

While it is crucial to evaluate the risk of human casualties related to floods, many developing countries are facing to challenge of finding simple mathematical models that can estimate this risk with minimal data. This study introduces a regional model designed to predict the annual number of fatalities by considering the hydraulic parameters of the flood and utilizing the morphological characteristics of watersheds. This approach indirectly considers the time available to evacuate individuals at risk. To validate the developed model, historical data were utilized for calibration and validation. Its performance was then compared with the results of the well-known Mortality Function (MF) method (Jonkamn et al. in J Nat Hazard 46(3):353–389, 2008). Both models effectively simulated historical conditions, although the Mortality Regional (MR) model demonstrated superior results for both high-risk and low-risk areas. By evaluating the MF method, which is considered a parametric model based on its structure and statistical characteristics, it was observed that adjusting the mean of the distribution by twice the standard deviation resulted in acceptable predictions for casualty numbers during a flash flood. Furthermore, this research assessed the uncertainties associated with both the MR and MF methods in estimating expected annual casualties from flooding.