Use of Hydrological Models in Global Stochastic Flood Modeling

Typical flood models do not take into consideration the spatial structure of flood events, which can lead to errors in the estimation of flood risk at regional to continental scales. Large-scale stochastic flood models can simulate synthetic flood events with a realistic spatial structure, although this method is limited by the availability of gauge data. Simulated discharge from global hydrological models has been successfully used to drive stochastic modeling in data-rich areas. This research evaluates the use of discharge hindcasts from global hydrological models in building stochastic river flood models globally: synthetic flood events in different regions of the world (Australia, South Africa, South America, Malaysia, Thailand, and Europe) are simulated using both gauged and modeled discharge. By comparing them, we analyze how a model-based approach can simulate spatial dependency in large-scale flood modeling. The results show a promising performance of the model-based approach, with errors comparable to those obtained over data-rich sites: a model-based approach simulates the joint occurrence of relative flow exceedances at two given locations similarly to when a gauge-based statistical model is used. This suggests that a network of synthetic gauge data derived from global hydrological models would allow the development of a stochastic flood model with detailed spatial dependency, generating realistic event sets in data-scarce regions and loss exceedance curves where exposure data are available.

Automated summary

Typical flood models neglect the spatial structure of flood events, but large-scale stochastic flood models can simulate synthetic flood events with realistic spatial structure. Global hydrological models' discharge hindcasts can be used to build stochastic river flood models globally. The model-based approach shows promising performance in simulating spatial dependency in large-scale flood modeling, providing reliable flood risk assessment for data-scarce regions.