Tropical Cyclone Storm Surge Hindcasting Using ECMWF Integrated Forecasting System

We analyze the suitability of ECMWF Integrated Forecasting System (IFS) for storm surge hindcasting. This is done by extracting meteorological forcing (U10, MSLP) from IFS at 0.225° horizontal resolution for eight historical tropical cyclones (TCs), and simulating storm surges using the Global Tide and Surge Model. We validate U10 and MSLP of ECMWF IFS against IB-TrACS data and maximum surge heights are validated using tide gauge data and/or other literature. In general, there is good agreement between modeled and observed U10 and MSLP. However, maximum intensities in small TCs (such as Haiyan and Patricia) are underestimated due to the horizontal resolution of IFS. Modeled maximum surge heights seem to be well in line with observations and literature, with R²=0.90 for Sandy and R²=0.91 for Irma.

We also discuss the effect of horizontal resolution of meteorological forcing on maximum surge heights. This is carried out by upscaling the meteorological forcing from 0.225° to coarser resolutions up to 1.0°. For all TCs, TC intensity is decreased in coarser resolution datasets, resulting in lower maximum surge heights. In addition, reductions are larger for larger maximum surge heights. Moreover, results suggest that coastal slope and coastline complexity may affect surge heights at coarser resolutions more than TC size. For relatively simple coastlines (fractal dimension < 1.10), the reduction in maximum surge height is 0.02 – 0.17 m (10-34%) when using 1.0° resolution forcing compared to 0.225° resolution forcing. For more complex coastlines, reductions exceed 0.5 m, up to 1.24 m for Irma. This suggests that particularly along complex coastlines, high resolution meteorological forcing results in the more adequately modeling of storm surges.