High-resolution interactive global flood forecast for past, present, and future

Bastian Van den Bout, David Meijvogel, Katherine van Roon, Faheed Kolaparambil

Globally operating modelling approaches for flood dynamics have shown significant increases in usage, quality and availability in recent years. Flood modelling at global scale stretches the computational demand, due to combinations of spatial scale and resolution, limiting to static pre-computed maps and prohibiting interactivity. Particular high-resolution global models are only available as static information layers, whereas dynamic models operate at lower resolution to address operational shortcomings. In this article, a novel system is described that links a state-of-the-art efficient flood modelling core with an automated global parameterization scheme based on global, bias-corrected and space-based datasets. This combined approach delivers a first-generation, zero configuration flood modelling system capable of producing high-resolution, interactive flood maps. It supports a highly efficient global early warning system for both flash and fluvial floods and enables the generation of historic flood map time series for any location worldwide. The utilized event-based simulation approach has some limitations in applicability to continuous modelling, the system shows major speed increases at near-identical accuracy with global availability.