Data assimilation of seaice motion vectors: Sensitivity to the parameterization of sea-ice strength
ABSTRACT. Data assimilation techniques are one method by which to improve the quality of model simulations of sea ice. The availability of daily gridded fields of sea-ice motion makes this field one that can be readily assimilated. These fields are generally of higher resolution than forcing values...
| Main Authors: | , , |
|---|---|
| Other Authors: | |
| Format: | Text |
| Language: | English |
| Subjects: | |
| Online Access: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.429.9949 http://www.igsoc.org/annals/44/a44a051.pdf |
| Summary: | ABSTRACT. Data assimilation techniques are one method by which to improve the quality of model simulations of sea ice. The availability of daily gridded fields of sea-ice motion makes this field one that can be readily assimilated. These fields are generally of higher resolution than forcing values such as atmospheric wind which are used to drive the model, and on any given day may depict ice circulation that is dramatically different than what the model solution represents. Typically, a blending method such as optimal interpolation (OI) is used and corrections are applied to the initial modeled velocity field such that the new solution corresponds better with actual observations. However, care must be taken in such a technique, as the corrections are not applied directly to the model physics, and the underlying physical assumptions in the ice dynamics may be violated. Previous studies have shown that improvements in the ice-motion solution come at the cost of the quality of other modeled fields. The strength parameterization in sea-ice models controls the ice velocity in the model, and is obtained in part by comparison with observed motions. Here we investigate the sensitivity of the sea-ice model to variations in the strength parameterization, and determine the effect of using data assimilation to impose observed velocities. We find that the alternation of the frictional loss parameter has limited effect on model performance. Rather, it is the assimilated data that overwhelm and degrade the solution, bringing into question whether underlying physical assumptions in the model may be compromised. |
|---|