A Cloud-Resolvable and Deep-Atmospheric Model Dynamics for the NCEP Global Forecast System

We are in the stage on developing cloud-resolvable and deep-atmospheric capabilities into NCEP GFS (National Centers for Environmental Prediction Global Forecast System). To have a very fine resolution global spectral model, there are two major steps have to conquer, one is the accuracy of spectral transform with thousands waves, another one is the speed of fast Legendre transform. There is underflow problem in computing Legendre base function for transform, which results error transformation while wave larger than 1900, with double precision. A method called x-number has been implemented into NCEP GFS to avoid underflow problem, thus the transformation can be applied to as high as any given waves. A fast Legendre transform will be implemented to speed up Legendre transform. The higher the wave number truncation, the more the effective transform.

The NCEP GFS has been supported Space Weather Prediction Center (SWPC) in research and operation, thus the deep atmospheric dynamic system is considered. Current NCEP global model has an option to consider any continents of gases in the atmosphere by enthalpy, thus thermodynamic variable is ready to be used with deep atmosphere. However, the dynamics have to be further improved to consider the variation of elevation and to include nonhydrostatic system. The concept of hydrostatic coordinates similar to mass coordinates, as used in NCEP MSM, will be used and extended to a generalized vertical coordinates. The conservations of nature will be considered in the discrete equation to code the model. The discretization of the system will be illustrated, and the preliminary of the model results will be presented.