This study investigates the differences between simulations using high-resolution GCM with T254 truncation and regional spectral model developed at NCEP. The regional model is configured in such a way that the spatial resolution is close to the GCM and starts from the same initial condition. The newly implemented NCEP regional spectral model allows us to study the spectral growth of the perturbation and thus understand factors that determine the quality of simulation. These two models employ the same model physics. The same model configurations allow us to pinpoint the necessity of regional model in downscaling application.

With the advent of increasing computational power, whether running high resolution GCM will improve the predictability or not still remains an unanswered question. Systematic comparison between model large-scale fields and precipitation will be presented here to understand the advantages and disadvantages of these two approaches.

We believe such a study would also provide some useful ideas for improving the forecast as well.