Global-scale modelling of sea level extremes to estimate coastal flood risk

Sanne Muis, Martin Verlaan, Hessel Winsemius, Philip Ward, Jeroen Aerts

Tuesday 30 june 2015

11:00 - 11:15h at North America (level 0)

Themes: (T) Extreme events, natural variability and climate change, (ST) Hydrological extremes: floods and droughts

Parallel session: 5I. Extreme events - FloodDrought

Increased coastal flood losses are likely to be one of the most costly impacts of climate change. Despite this, there are only a few studies on global coastal flood risk. These studies are all based on the DIVA database, which contains the only globally available data on extreme water levels. However, the water levels included in this database are based on simple assumptions and contain information for only four return periods. Moreover, the validation of these water levels is poorly documented. Hence, before we can accurately assess the impact of changing climate on global coastal flood losses to prioritise global adaptation actions, there is a need for an improved estimation of coastal water level extremes on a global-scale. On a regional scale, hydrodynamic models with meteorological forcing are frequently applied to calculate high water levels resulting from storm surges. However, as the modelling of surges in shallow coastal areas requires a high-resolution model grid, this approach is computationally too costly for most global models. The recent application of unstructured grids (or flexible mesh) in hydrodynamic models, allowing local refinement of the grid, has enabled the development of a global tide and surge model with a sufficient resolution in coastal areas, while maintaining computational efficiency. The global model is developed using Delft3D Flexible Mesh software from Deltares. By forcing the model with wind speed and atmospheric pressure derived from the ERA-Interim global reanalysis we can generate a 34 year hindcast of coastal water levels. Subsequently, extreme value distributions are fitted to the annual maxima to estimate exceedance probabilities for each output location. Here, we present first insights in the performance of the model by a comparison of the simulated water levels with tide gauge observations and the levels in the DIVA database.