Diagnostic vs dynamic representation of the inverse barometer effect in a global ocean model and its potential for probabilistic storm surge forecasting

TL;DR

This study evaluates how to represent the inverse barometer (IB) effect in a global ocean model to improve storm surge forecasting. By comparing a no-IB run, a diagnostically added IB (diaIB) run, and a dynamically forced IB (dynIB) run in NEMO forced by ERA5, and validating against the GESLA3 dataset, the authors quantify global and regional performance, including case studies of Didrik, Elsa, Michael, Mangkhut, and an ENS-based Xaver forecast. diaIB generally yields the best overall RMSE and MAE, with dynIB occasionally outperforming in semi-enclosed basins like the Baltic Sea; both IB approaches outperform noIB, especially for larger surges. The results demonstrate the practical value of incorporating the IB effect—either diagnostically or dynamically—in global models to improve storm surge forecasts and probabilistic warnings within the ECMWF ensemble system, enabling earlier, more reliable extreme-surge signals for risk management. The work suggests that coarse global models can support regional surge forecasting when the IB effect is accounted for, and that diagnostic IB remains a robust option when dynamic IB is not feasible.

Abstract

The global ocean model NEMO is run in a series of stand-alone configurations (2015-2022) to investigate the potential for improving global medium-range storm surge forecasts by including the inverse barometer effect. Here, we compare a control experiment, where the inverse barometer effect was not included, against a run dynamically forced with mean sea level pressure. In the control experiment, the inverse barometer effect was then calculated diagnostically and added to the ocean model sea surface elevation, resulting in a total of three experiments to investigate. We compare against the global GESLA3 water level data set and find that the inclusion of the inverse barometer effect reduces the root-mean-square error by $\sim 1~cm$ on average. When we mask out all data where the observed storm surge is less than $\pm1$ or $\pm2$ standard deviations, including the inverse barometer effect reduces the RMS error by $4-5$ cm. While both methods reduce water level errors, there are regional differences in their performance. The run with dynamical pressure forcing is seen to perform slightly better than diagnostically adding the inverse barometer effect in enclosed basins such as the Baltic Sea. Finally, an ensemble forecast experiment with the Integrated Forecast System of the European Centre for Medium-range Weather Forecasts demonstrates that when the diagnostic inverse barometer effect is included for a severe storm surge event in the North Sea (Storm Xaver, December 2013), the ensemble spread of water level provides a stronger and earlier indication of the observed maximum surge level than the when the effect is excluded.

Figures (19)

• Figure 1: Positions of all available observation records in the GESLA3 dataset. The blue dots indicate the positions of those records that was used in this work, while the red dots were omitted either due to no overlapping time coverage between model and observation, or due to the record being located too far away from a model ocean point. The horizontal and vertical histograms indicate the geographical distribution in longitude and latitude directions of the records.
• Figure 2: The maximum water level in the stand-alone storm surge hindcast with the IB effect included diagnostically as a post-process is displayed in panel (a). Panel (b) shows the difference between the diagnostic and dynamical inclusion of the IB effect. The red areas indicate where the diagnostic inclusion has the largest maximum water level, and the blue areas where the dynamical inclusion give the largest maximum water level.
• Figure 3: Distribution of the Root Mean Square Error (RMSE) for all records for the three different storm surge representations. The top panels (a-c) present the RMSE in meters, with a bin size of 2 cm. The bottom panels (d-f) show the RMSE scaled by the standard deviation ($\sigma$) of each observation record, using a bin size of $0.1\sigma$. Panels (a) and (d) present the RMSE distribution for the complete dataset. Panels (b) and (e) illustrate the RMSE for observed storm surge absolute values more than one standard deviation ($|\mathrm{obs}| > \sigma$). Panels (c) and (f) show the RMSE for observed storm surge absolute values exceeding two standard deviations ($|\mathrm{obs}| > 2\sigma$). The $y$-axis indicates the number of occurrences (records).
• Figure 4: Map of difference in RMSE for all records showing geographical differences. The circles indicate the position of the record, and the color indicate the difference in relative RMSE in number of standard deviations between the diagnostic and dynamical IB. Blue colors indicate that the diagnostic IB performs better than the dynamical IB, and red vice versa.
• Figure 5: Same as Figure \ref{['fig:map_rmse_diff_rel']}, but zoomed to the European region to highlight the difference between the European North West Shelf and Baltic Sea regions. Blue colors indicate that the diagnostic IB performs better than the dynamical IB, and red vice versa.