While increased horizontal resolution generally translates into stronger hurricanes, it is not granted that a more accurate representation of the storm scale, structure and processes can be achieved. For example, it would be desirable that the eye diameter and the radius of maximum wind become smaller with increased resolution. Increased resolution should also produce more realistic rain bands, which in turn could enhance the representation of the low-level moisture flux convergence, and more realistic warm core structure, which enhances the overall thermodynamics of the storm.
However, these improvements are dependent upon a number of factors, some intrinsic to the model, some related to the analysis. Analysis resolution or rejection of critical data which may cause analysis increments to oppose an intensification process are some of the problems encountered while representing strong TCs, particularly during rapid intensification (RI) phases. Even a model that in a free-running mode can generate very realistic tropical cyclones, could be penalized in forecast mode by an inadequate TC representation in the analysis.
In this talk several global simulations of Katrina will be presented, with different resolutions, forecast times and initialization strategies. Impact and limitations of different resolutions will be discussed, with emphasis on resolved physical processes which can, or cannot, be expected at any given resolutions.