Currently, the ocean model covers the entire Pacific but the atmosphereic model only covers a bean-shaped region in the central Pacific. Is there any reason for choosing such this particular region? Trick question -- of course there is, otherwise we wouldn't have chosen it in the first place. We chose that region because it's where the statistical atmospheric model has skill in reproducing the observed winds. An entirely sensible reason, I'm sure you would agree.
However, there appears to be more to the story. In particular, having the edges of the region slope off relatively close to the equator may have been a mistake. Doing this changes what is called the curl of the wind stress, which is a characteristic of the wind field that the ocean is quite sensitive to. We should't really be artificially changing the curl of the wind stress in this way. So, a new version of the model applies the winds uniformly over the region +/- 20 degrees about the equator.
We are still working on fully testing this change, but preliminary hindcast results show that the model is performing better than it used to. If these improvements in model skill are bourne out by a full model validation, then we will roll these changes into the operational forecast model.
Last modified: 24 July 1997