Advancing global flood hazard simulations by improving comparability, benchmarking, and integration of global flood models

Abstract

In recent years, a range of global flood models (GFMs) were developed, each utilizing different process descriptions as well as validation data sets and methods. To quantify the magnitude of these differences, studies assessed the performance of GFMs only on the continental and catchment level. Since the default model set-ups resulted in locally marked deviations, there is a clear need for further and especially more standardized research to not only maintain credibility, but also support the application of GFM products by end-users. Consequently, here we conceptually outline the basic requirements and challenges of a Global Flood Model Validation Framework for more standardized model validation and benchmarking. With the proposed framework we hope to encourage the much needed debate, research developments in this direction, and involvement of science with end-users. By means of the framework, it is possible to streamline the data sets used for input and validation as well as the validation approach itself. By subjecting GFMs to more thorough and standardized methods, we think their quality as well as acceptance will increase as a result, especially amongst endusers of their outputs. Otherwise GFMs may only serve a purely scientific purpose of continued 'siloed' model improvement but without practical use. Furthermore, we want to invite GFM developers to make their models more integratable which would allow for representation of more physical processes and even more detailed comparison on a model component basis. We think this is pivotal to not only improve the accuracy of model input data sets, but to focus on the core of each model, the process descriptions. Only if we know more about why GFMs deviate, are we able to improve them accordingly and develop a next generation of models, not only providing first-order estimates of flood extent but supporting the global disaster risk reduction community with more accurate and actionable information.

Flood Risk Assessment of Global Watersheds Based on Multiple Machine Learning Models

“…However, global flood risk assessment obtained by machine learning methods is only a static result rather than a dynamic one, and the physically based model has an advantage. A physically based model can often give more detailed information on flood hazards, such as flow and submerged range, while current machine learning studies focus on qualitative assessment of flood hazards [21][22][23]. These qualitative evaluations can only provide limited reference for watershed management.…”

“…These methods are based on expert knowledge and are susceptible to uncertainty [19]. Physically based models such as VIC and MIKE models at a regional scale, and other hydrological models at the continental and global scale have also been used to study floods, and have shown great advantages in regional or global flood process research [8,[20][21][22][23]. Recently, machine learning methods such as artificial neural networks (ANN), support vector machines (SVM), and decision trees (DT) have been applied to flood hazard assessments, which can identify and evaluate flood-prone areas based on the training and testing of large amounts of data [24][25][26][27].…”

“…This is its advantage over machine learning models. However, in the face of the complexity of global floods, flood models at a global scale often require a large number of model parameters, repeated model debugging, and long computing time [19,[21][22][23][26][27][28]. Machine learning methods may be a good choice for faster access to global flood hazard assessments.…”

Global flood hazard map and exposed GDP comparison: a China case study

“…Contrary to previous studies, 80 no relationship was found between performance and model spatial resolution. In a follow-up study, Hoch and Trigg (2019) proposed a global flood model validation framework. The aim of this framework is to understand the drivers of deviations between GFMs by providing standard forcing data, validating and benchmarking model results, and by sorting and indexing reference output.…”

“…In order to get a better understanding of which of these characteristics has the largest influence on overall model performance, a 510 systematic comparison framework is required, in which each of these modeling components and parameters can be tested individually and in unison. The proposed model comparison framework of Hoch and Trigg (2019) could therefore greatly benefit our current understanding of global flood hazard.…”