A fully-open approach to modeling TC storm surge on a global scale

The existing research about hindcasting flood surges and flood plains caused by tropical cyclones (TCs) is largely specific to single storms, regions or countries. Often enough, the tools and data used are not publicly available, making it challenging to extend those analyses to other world regions. For instance, there is not yet a global data set of flood surges and flood extent maps from tropical cyclones. 

We present a modeling framework for hindcasting TC surges, based on the open-source software GeoClaw. We use open-source software and publicly available data sources only, making our approach fully transparent, reproducible and reusable for a global community of impact and risk modelers. The selected input data products include all world regions. Thus, our setup can be applied to all ocean basins that experience TCs. We evaluate our framework by comparing i) the flood surges of selected events to tide gauge records and modeled (GTSM) sea level, and ii) the flood extent maps to a global data set of satellite-based (MODIS) flood maps for the period 2000-2019.

Our analysis of tide gauge records shows that the setup captures the storm surge component of ocean dynamics at gauge locations very well, even though the model's capability to incorporate astronomical tides is limited. From the satellite-based flood maps, it becomes clear that beneath storm surge also rainfall and fluvial floods are important drivers of TC-related flooding, making it difficult to evaluate our model's performance based on its agreement with observed flood extents. Further, a comparison with high water marks in the field indicates that satellite-based products are often not sufficient to capture the full extent of short-term coastal flooding due to limited satellite overpass times. In this sense, storm surge model output can complement satellite observations.

The modeling framework allows us to generate globally consistent TC storm surge hazard data for all world regions, including the global south and least developed countries. It is suitable for historical analyses as well as for attribution studies or future climate projections, based on synthetic events or ensembles.