Reconstruction of a Flash Flood in a Semi‐Arid Ungauged Catchment Using Integrated Surface–Subsurface Modelling and Non‐Stationary Extreme Value Analysis
Abstract
A high‐intensity precipitation event triggered a deadly flash flood in a semi‐arid ungauged catchment upstream of a major urban area, resulting in severe loss of life and extensive infrastructure damage. This study reconstructs the hydrological response of the catchment and identifies the key physical and anthropogenic factors that contributed to the flood magnitude and timing. Although focused on a single event in Iran, the findings are broadly applicable to Mediterranean‐type catchments experiencing intensified short‐duration precipitation extremes and expanding urban development under climate change. A physics‐based, fully integrated surface–subsurface hydrological model (HydroGeoSphere) is applied to simulate runoff generation and flood propagation. Model results indicate that antecedent precipitation and recent residential development each increased peak discharge by approximately 10%, while infiltration into the subsurface reduced peak discharge by nearly 40%. Sensitivity analysis identifies saturated hydraulic conductivity and surface roughness as the dominant parameters controlling flood magnitude. Stationary and non‐stationary generalised extreme value analyses show substantial uncertainty at long return periods and strong sensitivity to inclusion of the 2019 extreme event. The simulated peak discharge of approximately 55 m 3 s −1 significantly exceeded the reported design discharge of about 12 m 3 s −1 , indicating likely exceedance of the intended design conditions. Results highlight the critical role of infiltration, transient subsurface storage and channel conveyance in attenuating flash flood peaks in small, steep catchments, while demonstrating that conventional stationary design assumptions may substantially underestimate flood risk under non‐stationary climatic conditions. The findings emphasise the importance of integrated surface–subsurface modelling for ungauged basins and support adaptive flood risk management strategies that combine enhanced conveyance, upstream mitigation and real‐time early warning systems.