2005 Volume 1 Pages 73-76
In order to improve the forecast of convection, the MM5 3D-Var (3-Dimensional Variational Data Assimilation) system is applied to initialize the mesoscale nonhydrostatic model, MM5, for a convective case on 12 to 13 June 2002.
The BES (Background Error Statistics) is the key component used in a 3D-Var system, which helps to extract the useful information from the observations and to make the final analysis balanced. In this study, an approach is proposed to construct a case-dependent BES file for a convective case with a relatively inexpensive computing cost.
By using MM5 3D-Var system and MM5 forecast model, a series of the 3D-Var experiments with the default BES and the newly constructed BES are carried out. The results showed that the assimilation of the conventional and GPS (Global Positioning System) PW (Precipitable Water) data with the new BES improved the forecast of convection. The real-time GPS PW itself also has a positive impact on the forecast of convection.