Sea ice effects in a spectral wave model: principles and practical implementation

Numerical wave models have their largest errors around sea ice, and their accuracy is generally unknown in the ice as very few data are available. This is largely because they do not, or in a coarse way, take into account the interactions of waves and sea ice, and because the necessary information about sea ice properties is not readily available. Recent progress have expanded our knowledge of wave scattering by sea ice as well as several dissipation processes, highlighting the need to include ice thickness and information on the ice floes size. Starting from a consistent representation of energy and dispersion in the presence of sea ice, we have redefined a set of self-consistent dissipation and scattering parameterizations for the WAVEWATCH III model which is expected to apply to a variety of ice conditions with the exception of forming ice. In our model the ice is treated as a single layer that can be fractured in many floes expected to be equivalent to circular floes with a power law distribution of diameters that is defined from the maximum diameter Dmax and a fragility parameter. This layer of ice induces a dissipation of the wave energy through basal friction (Stopa et al. The Cryosphere, 2016) and secondary creep associated with ice flexure (Cole et al. 1998), in addition to an energy-conserving scattering modeled following Kohout and Meylan (2006). The ice thickness and concentration are taken uniform over a model grid cell, and are typically provided by model products or satellite data, and are not affected by the waves. The wave model results are used to update Dmax by a probabilistic evaluation of ice break-up by the waves. This process introduces an interesting feedback on the wave scattering and dissipation. The combination of dissipation and scattering leads to spatial patterns in the wave height and directional spreading of the wave field that can be easily tested with in situ or remote sensing data (Sutherland and Gascard GRL 2016, Ardhuin et al. GRL 2016). In particular our model assumes that wave dissipation in unbroken ice is dominated by creep, which then vanishes once the ice is broken, reproducing the wave attenuation observed by Collins et al. (GRL 2015). Consistent with the low directional spread in most SAR imagery, scattering does not appear as the main factor for the attenuation of dominant swells with periods 12 to 25 s.