Articles | Volume 10, issue 5
https://doi.org/10.5194/gmd-10-1961-2017
https://doi.org/10.5194/gmd-10-1961-2017
Model description paper
 | 
22 May 2017
Model description paper |  | 22 May 2017

Vorticity-divergence semi-Lagrangian global atmospheric model SL-AV20: dynamical core

Mikhail Tolstykh, Vladimir Shashkin, Rostislav Fadeev, and Gordey Goyman

Abstract. SL-AV (semi-Lagrangian, based on the absolute vorticity equation) is a global hydrostatic atmospheric model. Its latest version, SL-AV20, provides global operational medium-range weather forecast with 20 km resolution over Russia. The lower-resolution configurations of SL-AV20 are being tested for seasonal prediction and climate modeling.

The article presents the model dynamical core. Its main features are a vorticity-divergence formulation at the unstaggered grid, high-order finite-difference approximations, semi-Lagrangian semi-implicit discretization and the reduced latitude–longitude grid with variable resolution in latitude.

The accuracy of SL-AV20 numerical solutions using a reduced lat–lon grid and the variable resolution in latitude is tested with two idealized test cases. Accuracy and stability of SL-AV20 in the presence of the orography forcing are tested using the mountain-induced Rossby wave test case. The results of all three tests are in good agreement with other published model solutions. It is shown that the use of the reduced grid does not significantly affect the accuracy up to the 25 % reduction in the number of grid points with respect to the regular grid. Variable resolution in latitude allows us to improve the accuracy of a solution in the region of interest.

Download
Short summary
We present the hydrostatic dynamical core of the SL-AV20 global atmosphere model used for operational numerical weather prediction in Russia. The article describes model design, particularly vorticity-divergence formulation combined with the use of an unstaggerred grid. The model has an option to use a reduced latitude-longitude grid. The results for standard tests agree well with reference solutions.