Intercomparing atmospheric reanalysis products for hydrodynamic and wave modeling of extreme events during the open-water Arctic season

ABSTRACT

The significant increase in the Arctic open-water extent along with the earlier sea-ice summer melt and later autumn freeze-up seasons observed in the last decades allow the formation of less fetch-limited waves and the further propagation of storm surges to new ice-free shores. Coupled hydrodynamic and wave models can be used to simulate these complex atmospheric–ocean interactions that often result in coastal flood hazards and extreme waves. However, the reliability of such simulations is intrinsically dependent on the quality of their main inputs, including wind and mean sea-level pressure products, which are usually extracted from reanalysis. This study evaluates the storm surge and significant wave height hindcasts from the coupled ADCIRC+SWAN numerical model forced by seven different reanalysis products during contrasting major storms. Model results show that the highest spatial resolution product CFSv2 led to the overall most accurate model simulations, performing particularly well at locations exposed to extreme surge and waves. Average root mean square error increases of up to 100 percent for storm surge and 157.55 percent for significant wave height were observed when using products other than CFSv2, thus highlighting the importance of selecting the proper wind and pressure reanalysis to be implemented as forcing in the hydrodynamic and wave numerical model.

KEYWORDS:

• Numerical modeling
• wind and pressure forcing
• Bering Sea
• ADCIRC+SWAN
• ice-free Alaskan coast

Disclosure statement

No potential conflict of interest was reported by the authors.

Author contributions

Felício Cassalho designed the study, developed the modeling framework, carried out the formal analysis, and drafted the article. Tyler W. Miesse developed the modeling framework and contributed to the model output postprocessing, visualization, and validation. Arslaan Khalid contributed to the processing of model inputs and data visualization and reviewed the article. André de S. de Lima contributed to the model output postprocessing and visualization and reviewed the article. Celso M. Ferreira designed the study and supervised the computational analyses and article writing process. Martin Henke contributed to the formal analysis and research investigation and reviewed the article. Thomas M. Ravens supervised the computational analyses and the article writing process.

Supplementary material

Supplemental material for this article can be accessed on the publisher’s website.

Additional information

Funding

This material is based on work supported by the National Science Foundation under Grant No. 1927785. This funding is gratefully acknowledged but implies no endorsement of the findings. This research used the computational resources from Cheyenne (doi:10.5065/D6RX99HX) provided by NCAR’s Computational and Information Systems Laboratory, sponsored by the National Science Foundation and the Extreme Science and Engineering Discovery Environment (XSEDE) STAMPEDE2 resources through allocation ID TG-BCS130009, which is supported by the National Science Foundation (Grant No. ACI-1548562). Financial support was provided to the fourth author, ASL, by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior–Brasil (CAPES), Finance Code 001.