Flash floods pose a distinct challenge compared to traditional fluvial flooding, with infrastructure-based solutions proving less effective. Effective responses hinge on advanced early warning systems providing actionable information, emphasising the necessity for computational flood forecasting models. However, hydrodynamic models, renowned for accuracy and completeness, face limitations due to computational intensity. This study explores two 2D flood forecasting models, RIM2D and SERGHEI, both with GPU implementations which allow us to maximise the forecast lead time. While RIM2D is less computationally intensive, suitable for operational use, SERGHEI, with higher computational costs, targets large-scale high-performance computing (HPC) systems. The assessment of applicability and trade-offs is carried out on the 2021 Eifel flood event, particularly in the lower Ahr valley. A set of simulations were performed at various resolutions from 1 to 10 m, which reveal similar accuracy among both models at coarser resolutions, yet discrepancies arise at finer resolutions due to the distinct formulations. Both models exhibit a rapid computational cost escalation, but at resolutions equal to or coarser than 5 m, forecasts are remarkably faster than the real-time ideal for operational use, paving the way for their use in early warning systems. However, higher resolutions necessitate multi-GPU and HPC capabilities, underlining the importance of embracing such technology in addressing broader flood domains.
Rainfall is a spatiotemporally varied process and key to accurately capturing catchment runoff and determining flood response. Flash flood response of a catchment can be strongly governed by a rainfall’s spatiotemporal variability and is influenced by storm movement which drives a continuous spatiotemporal change throughout a rainfall event. In this work, the sensitivity of runoff and flooded areas to rainfall movement are assessed in the Kan catchment (Iran). The allochthonous nature of floods in the catchment and how they interact with the effects of rainfall movement are investigated. Fifty synthetic rain hyetographs are generated and traversed over the catchment under different velocities and directions and used to force a 1D/2D hydrodynamic model. The results suggest rainfall movement affects the runoff response in different degrees. Peak discharge, hydrograph shapes and flooded areas are affected. Storms with higher velocities result in higher peaks and faster onsets of runoff and consequently higher flooded areas in comparison to slower storms. The direction of the movement also plays a role. Storms moving along the average direction of the stream result in higher peaks and flooded areas. The relevance of storm direction is greater for slow moving storms. Additionally, the influence of rainfall movement is modulated by hyetograph structure, and the allochthonous behavior is greatly dependent on the location within the drainage network at which it is assessed.
14 May 2025 in Natural Hazards and Earth System Sciences
In recent years, urban areas have been increasingly affected by more frequent and severe pluvial floods, attributed to climate change and urbanization. This trend is expected to continue in the future, underscoring the critical importance of flood warning and disaster management. However, pluvial flood forecasts on a communal level do not exist in practice, mainly due to the high computational runtimes of high-resolution flood simulation models. Here, we showcase the capability of the hydrodynamic model RIM2D (Rapid Inundation Model 2D) to deliver highly detailed and localized insights into expected flood extent and impacts in very short computational processing times, enabling operational flood warnings. We demonstrate these capabilities using the case of the June 2023 torrential rain and resulting flood event in the city of Braunschweig, located in Lower Saxony, Germany. During this event, the city experienced intense rainfall of 60 L m−2 within a short time frame, resulting in widespread inundation, significant disruption to the residents' daily life, and substantial economic costs to the city. This study serves as a clear indication that different dimensions of flood consequences can be simulated at very high resolutions in extremely short computational times and that models such as RIM2D, along with the necessary hardware for their operation, have reached a level of maturity suitable for integration into operational early warning systems and impact-based forecasting systems for such floods.
4 March 2025 in Natural Hazards and Earth System Sciences
Hydrodynamic models are crucial for flood forecasts and early warnings, particularly in response to events such as the devastating floods in Germany's Ahr region in July 2021. However, several uncertainties can be present in these models stemming from various sources, such as model structure, parameters, and boundary conditions. In this study, we aim to address these uncertainties and enhance the existing hydrodynamic model set up using RIM2D (Rapid Inundation Model 2D), introduced by Apel et al. (2022) for the Ahr region. The goal is to fortify its robustness and reliability for inundation simulations in the area. For this, we employ a large number of Monte Carlo simulations, assessing the effects of various model elements such as floodplain and channel roughness coefficients, as well as terrain resolution, on river dynamics and inundation. Our findings emphasize the critical role of proper parameter assignment in attaining optimal simulation results. The results show that sensitivity to input factors varies depending on the performance metrics used and the output predicted. We demonstrate that to simulate flow formation and water level in the river channel, the roughness parameter of the river channel and the model's resolution are paramount. On the other hand, to simulate flood extent and the distribution of water depth across the domain, even coarser resolutions prove adequate and, due to their reduced computation time, might be better suited for early-warning systems. Furthermore, our findings suggest that the differences observed between finer- and coarser-resolution models may stem from the varying representations of the river channel and buildings within the model. Ultimately, this work provides a guideline for the parameterization of RIM2D and similar physically based fluvial models tailored to the Ahr region, offering valuable insights for future hydrodynamic modeling endeavors in the area.
Rainfall is a spatiotemporally variated process and one of the key elements to accurately capture both catchment runoff response and floodplain extents. Flash floods are the result of intense rainfall, typically associated to highly variable rain in both space and time, such as convective storms. In this work, the extent within-storm variability affects runoff and flooding is explored. The Kan catchment (Tehran, Iran) is used as base topography for the simulations. The allochthonous nature of floods in the catchment and how they interact with the effects of storm variability are further investigated. For this, 300 synthetic rainfall signals with different hyetograph variabilities are generated and imposed on a 1D/2D hydrodynamic model. Additionally, a set of simulations with different levels of spatial variability are performed. The results suggest that temporal and spatial variability affect the runoff response in different degrees. Peak discharge and hydrograph shapes, as well as flooded areas, are affected. The effect of storm temporal variability is shown to be significantly higher than storm spatial variability and storm properties such as return period, duration, and volume. Further on the influence of storm spatiotemporal variability on stream discharge and flood response is seen to be strongly dependent on the location within the drainage network at which it is assessed.
Cascading drought-flood events (CDFEs), also referred to as ‘drought-to-flood transitions’ or ‘drought–flood abrupt alternations,’ in which a flood follows a period of drought, may have different flood generation mechanisms than floods occurring independently from drought, as the drought could affect soil infiltration rates and, consequently, runoff dynamics. With the increasing frequency of extreme weather events driven by climate change, understanding the cascading nature of drought and flood events has become crucial for effective disaster risk management. However, there is a lack of empirical evidence on how these drought-flood interactions work and translate to economic losses. This study addresses this gap by identifying CDFEs and flood-only events (FEs) across Central Europe and linking them to their flood impacts from the modelled Historical Analysis of Natural Hazards in Europe database. CDFEs are associated with significantly higher maximum daily mean streamflow (58.51 m3 s−1 vs 38.20 m3 s−1), deeper mean water depths (1.90 m vs 1.88 m), and greater economic losses (€33.09 million km−2 vs €29.75 million km−2) compared to FEs. These findings underscore the special features of CDFEs and the need to take them into account in flood risk management.