Sensitivity testing of a new global model with special field observations

A compressible, nonhydrostatic global model has been developed that allows variable spatial resolution for the case of spherical coordinates. The model produces numerically stable short-term forecasts while retaining relatively little explicit horizontal diffusion. Perhaps because of the small diffusivity, the model exhibits more sensitivity to regional initial state changes than do some other models. This is demonstrated with a series of real-data forecast experiments based upon a special field program performed over sub-tropical South America during the southern summer of 2002-2003. These are compared with predictions performed by a global primitive equation model and by the regional WRF model for the same cases. The cases examined include active mesoscale convective events. Preliminary results suggest that: 1) the new model displays at least twice as much 48 hour forecast sensitivity to the special field observations as do the regional WRF model or the global PE model, 2) model forecast sensitivity to initial state perturbations is strongly influenced by model diffusion 3) forecast sensitivity is strongly affected by moist processes