TY - JOUR
T1 - Numerically Converged Solutions of the Global Primitive Equations for Testing the Dynamical Core of Atmospheric GCMs
AU - Polvani, L M
AU - Scott, Richard Kirkness
AU - Thomas, S J
PY - 2004/11
Y1 - 2004/11
N2 - Solutions of the dry, adiabatic, primitive equations are computed, for the first time, to numerical convergence. These solutions consist of the short-time evolution of a slightly perturbed, baroclinically unstable, midlatitude jet, initially similar to the archetypal LC1 case of Thorncroft et al. The solutions are computed with two distinct numerical schemes to demonstrate that they are not dependent on the method used to obtain them.
These solutions are used to propose a new test case for dynamical cores of atmospheric general circulation models. Instantaneous horizontal and vertical cross sections of vorticity and vertical velocity after 12 days, together with tables of key diagnostic quantities derived from the new solutions, are offered as reproducible benchmarks. Unlike the Held and Suarez benchmark, the partial differential equations and the initial conditions are here completely specified, and the new test case requires only 12 days of integration, involves no spatial or temporal averaging, and does not call for physical parameterizations to be added to the dynamical core itself.
AB - Solutions of the dry, adiabatic, primitive equations are computed, for the first time, to numerical convergence. These solutions consist of the short-time evolution of a slightly perturbed, baroclinically unstable, midlatitude jet, initially similar to the archetypal LC1 case of Thorncroft et al. The solutions are computed with two distinct numerical schemes to demonstrate that they are not dependent on the method used to obtain them.
These solutions are used to propose a new test case for dynamical cores of atmospheric general circulation models. Instantaneous horizontal and vertical cross sections of vorticity and vertical velocity after 12 days, together with tables of key diagnostic quantities derived from the new solutions, are offered as reproducible benchmarks. Unlike the Held and Suarez benchmark, the partial differential equations and the initial conditions are here completely specified, and the new test case requires only 12 days of integration, involves no spatial or temporal averaging, and does not call for physical parameterizations to be added to the dynamical core itself.
KW - TEST SET
UR - http://www.scopus.com/inward/record.url?scp=8744240583&partnerID=8YFLogxK
U2 - 10.1175/MWR2788.1
DO - 10.1175/MWR2788.1
M3 - Article
SN - 0027-0644
VL - 132
SP - 2539
EP - 2552
JO - Monthly Weather Review
JF - Monthly Weather Review
IS - 11
ER -