Summary
We use a set of metrics for the polar atmosphere to characterize and assess the performance of the ACCESS-1.0, ACCESS-1.3 and CSIRO-Mk3.6 coupled ocean-atmosphere general circulation models over the recent historical period. The metrics are based on diagnostics used by Butchardt et al. [2011] to assess models of the 2nd Chemistry-Climate Model Validation (CCMVal-2) activity of SPARC. Cold temperature biases occur in upper levels of all models. The CSIRO-Mk3.6 model has a warm winter lower stratosphere. Similar winter and spring mean biases are generally shown by CCMVal-2 models. The strength of the Polar Night Jet (PNJ) is anomalously weak at 60°S in spring in all models (weakest for CSIRO-Mk3.6). In general, the CSIRO-Mk3.6 model has difficulties reproducing the observed location of the maximum in the geopotential height stationary wave pattern. For ACCESS-1.0, the Southern Hemisphere (SH) winter jet and spring stationary wave pattern are 5-10° equatorward of the climatology. The amplitudes of the stationary wave patter in geopotential height is generally well matched to observations, except in SH spring where it is weak compared with observations (approximately 50% smaller in amplitude). The spring SH mean zonal mean winds are weak in both the CSIRO-Mk3.6 and ACCESS-1.0 models during spring in the SH. At other times of the year, the winds are stronger than observed in the upper stratosphere. Overall, reconciling the cold biases in the upper levels of the models, particularly in winter for both hemispheres, and spring, deserves further attention. In both hemispheres, the PNJ is displaced equatorward relative to the observed climatology, and this may lead to stronger radiative cooling over the poles. Further work is required to examine how well dynamical heating (e.g. by downwelling from the upper levels during winter) is represented in the models.
Related projects
