Uncoupled high-latitude wave models in COAMPS

TL;DR

This paper evaluates uncoupled high-latitude wave models SWAN and WW3 within COAMPS, analyzing their performance in sea ice regions through multiple case studies and various model configurations.

Contribution

It demonstrates the application and assessment of two wave models in high-latitude regions with sea ice, including new empirical sea ice dissipation formulas and sensitivity experiments.

Findings

Model skill varies with sea ice dissipation settings.

Higher resolution forcing improves model accuracy.

Sea ice dissipation significantly impacts wave predictions.

Abstract

This report describes six demonstration cases of two numerical ocean wave models in high latitude regions where waves interact with sea ice. The two wave models are SWAN (Simulating WAves Nearshore) and WW3 (WAVEWATCH III), run in uncoupled mode within the Navy's coupled regional modeling system, COAMPS(R) (Coupled Ocean Atmosphere Mesoscale Prediction System). The COAMPS software handles a large majority of the tasks associated with the setup, running, and post-processing of the wave models. All six cases are cycling runs with 12-hour increments, each providing a continuous hindcast of four to 26 days duration. SWAN is applied in a Bering Strait case and two Gulf of Bothnia cases. WW3 is applied in a Sea of Okhotsk case and two Barents Sea cases. Verification is performed by visual inspection of model output fields, and by comparing model runs with alternative settings. In the standard configuration, forcing comes from archived global model output, including information on surface wind vectors, sea ice concentration and thickness, and surface current vectors, and the wave model uses a new empirical formula for dissipation of wave energy by sea ice that is dependent on ice thickness. The Barents Sea cases are compared to spectral wave data from satellite (SWIM instrument on CFOSAT) and from motion sensors deployed on the ice by the Norwegian Meteorological Institute. Experiments are performed with non-standard settings, 1) disabling the dissipation by sea ice, 2) using an older formula for dissipation by sea ice which does not depend on ice thickness, 3) using higher resolution wind and sea ice concentration forcing fields, and 4) omitting surface currents. The impact of these settings on model skill is quantified by comparison to the observations. The skill is also compared to that from a global (thus, lower resolution) ocean wave model.