Using machine learning for pluvial flood prediction tasks has gained growing attention in the past years. In particular, data-driven models using artificial neuronal networks show promising results, shortening the computation times of physically based simulations. However, recent approaches have used mainly conventional fully connected neural networks which were (a) restricted to spatially uniform precipitation events and (b) limited to a small amount of input data. In this work, a deep convolutional generative adversarial network has been developed to predict pluvial flooding caused by nonlinear spatial heterogeny rainfall events. The model developed, floodGAN, is based on an image-to-image translation approach whereby the model learns to generate 2D inundation predictions conditioned by heterogenous rainfall distributions—through the minimax game of two adversarial networks. The training data for the floodGAN model was generated using a physically based hydrodynamic model. To evaluate the performance and accuracy of the floodGAN, model multiple tests were conducted using both synthetic events and a historic rainfall event. The results demonstrate that the proposed floodGAN model is up to 10⁶ times faster than the hydrodynamic model and promising in terms of accuracy and generalizability. Therefore, it bridges the gap between detailed flood modelling and real-time applications such as end-to-end early warning systems.

In times of increasing weather extremes and expanding vulnerable cities, a significant risk to civilian security is posed by heavy rainfall induced flash floods. In contrast to river floods, pluvial flash floods can occur anytime, anywhere and vary enormously due to both terrain and climate factors. Current early warning systems (EWS) are based largely on measuring rainfall intensity or monitoring water levels, whereby the real danger due to urban torrential floods is just as insufficiently considered as the vulnerability of the physical infrastructure. For this reason, this article presents a concept for a risk-based EWS as one integral component of a multi-functional pluvial flood information system (MPFIS). Taking both the pluvial flood hazard as well as the damage potential into account, the EWS identifies the urban areas particularly affected by a forecasted heavy rainfall event and issues object-precise warnings in real-time. Further, the MPFIS performs a georeferenced documentation of occurred events as well as a systematic risk analysis, which at the same time forms the foundation of the proposed EWS. Based on a case study in the German city of Aachen and the event of 29 May 2018, the operation principle of the integrated information system is illustrated.

This study presents a deep-learning-based forecast model for spatial and temporal prediction of pluvial flooding. The developed model can produce the flooding situation for the upcoming time steps as a sequence of flooding maps. Thus, a dynamic overview of the forthcoming flooding situation is generated to support the decision of crisis management actors. The influence of different input data, data formats, and model setups on the prediction results was investigated. Data from multiple sources were considered as follows: precipitation information, spatial information, and an overflow forecast. In addition, models with different layers and network architectures such as convolutional layers, graph convolutional layers, or generative adversarial networks (GANs) were considered and evaluated. The data required to train and test the models were generated using a coupled hydrodynamic 1D/2D model. The model setup with the inclusion of all available input variables and an architecture with graph convolutional layers presented, in general, the best results in terms of root mean square error (RMSE) and critical success index (CSI). The prediction results of the final model showed a high agreement with the simulation results of the hydrodynamic model, with drastic reductions in computation time, making this model suitable for integration into an early warning system for pluvial flooding.

The effective forecast and warning of pluvial flooding in real time is one of the key elements and remaining challenges of an integrated urban flood risk management. This paper presents a new methodology for integrating risk-based solutions and 2D hydrodynamic models into the early warning process. Whereas existing hydrodynamic forecasting methods are based on rigid systems with extremely high computational demands, the proposed framework builds on a multi-model concept allowing the use of standard computer systems. As a key component, a pluvial flood alarm operator (PFA-Operator) is developed for selecting and controlling affected urban subcatchment models. By distributed computing of hydrologic independent models, the framework overcomes the issue of high computational times of hydrodynamic simulations. The PFA-Operator issues warnings and flood forecasts based on a two-step process: (1) impact-based rainfall thresholds for flood hotspots and (2) hydrodynamic real-time simulations of affected urban subcatchments models. Based on the open-source development software Qt, the system can be equipped with interchangeable modules and hydrodynamic software while building on the preliminary results of flood risk analysis. The framework was tested using a historic pluvial flood event in the city of Aachen, Germany. Results indicate the high efficiency and adaptability of the proposed system for operational warning systems in terms of both accuracy and computation time.

Increased interest in combining compound flood hazards and social vulnerability has driven recent advances in flood impact mapping. However, current methods to estimate event-specific compound flooding at the household level require high-performance computing resources frequently not available to local stakeholders. Government and non-governmental agencies currently lack the methods to repeatedly and rapidly create flood impact maps that incorporate the local variability in both hazards and social vulnerability. We address this gap by developing a methodology to estimate a flood impact index at the household level in near-real time, utilizing high-resolution elevation data to approximate event-specific inundation from both pluvial and fluvial sources in conjunction with a social vulnerability index. Our analysis uses the 2015 Memorial Day flood in Austin, Texas, as a case study and proof of concept for our methodology. We show that 37 % of the census block groups in the study area experience flooding from only pluvial sources and are not identified in local or national flood hazard maps as being at risk. Furthermore, averaging hazard estimates to cartographic boundaries masks household variability, with 60 % of the census block groups in the study area having a coefficient of variation around the mean flood depth exceeding 50 %. Comparing our pluvial flooding estimates to a 2D physics-based model, we classify household impact accurately for 92 % of households. Our methodology can be used as a tool to create household compound flood impact maps to provide computationally efficient information to local stakeholders.

The domain of spatial flood prediction is dominated by hydrodynamic models, which, while robust and adaptable, are often constrained by computational requirements and slow processing times. To address these limitations, the integration of Deep Learning (DL) models has emerged as a promising solution, offering the potential for rapid prediction capabilities, while maintaining a high output quality. However, a critical challenge with DL models lies in their requirement for retraining for each new domain area, based on the outputs of hydrodynamic simulations generated for that specific region. This need for domain-specific retraining hampers the scalability and quick deployment of DL models in diverse settings. Our research focuses on bridging this gap by developing a fully generalized DL model for flood prediction.FloodWaive's approach pivots on creating a DL model that can predict flood events rapidly and accurately across various regions without requiring retraining for each new domain area. The model is trained on a rich dataset derived from numerous hydrodynamic simulations, encompassing a wide spectrum of topographical conditions. This training is designed to enable the model to generalize its predictive capabilities across different domains and weather patterns, thus overcoming the traditional limitation of DL models in this field.Initial findings from the development phase are promising, showcasing the model's capability to process complex data and provide quick, accurate flood predictions. The success of this fully generalized DL modeling approach could revolutionize applications of flood predictions such as flood forecasting and risk analysis. Regarding the later, real-time evaluation of flood protection measures could become a reality. This would empower urban planners, emergency response teams, and environmental agencies with the ability to make informed decisions quickly, potentially saving lives and reducing economic losses.While this project is still in its developmental stages, the preliminary results point towards a significant leap in flood forecasting technology. The ultimate goal is to offer a universally deployable, real-time flood prediction tool, significantly enhancing our ability to mitigate the impact of floods worldwide.