Coupled Simulations by the UCLA AGCM with a new PBL Parameterization and the MIT OGCM: Sensitivity to the AGCM Resolution

The performance of coupled atmosphere-ocean GCMs crucially depends on the successful simulation of processes that control the surface fluxes of radiation, latent and sensible heat and momentum. The planetary boundary layer parameterization (PBL) is key for the computation of these fluxes, as well as the PBL clouds and their effects. In this work we present two multi-decadal simulations by the UCLA AGCM with an upgraded PBL parameterization, coupled with the MIT global OGCM. The simulations differ in the spatial resolution of the AGCM, which is 4 lat x 5 lon x15 layers in one case and 2 lat x 2.5 lon x 29 layers in the other. As in the previous versions of the UCLA AGCM, the new PBL parameterization predicts the PBL depth using a vertical coordinate that shares a coordinate surface with the free atmosphere at the PBL top. This framework facilitates an explicit representation of processes concentrated near the PBL top and PBL clouds. In the new PBL parameterization, the bulk turbulence kinetic energy (TKE) of the PBL is predicted. The surface fluxes are determined from an aerodynamic formula, in which a combination of the square root of the TKE and the grid-scale surface wind are used to represent the velocity scale. With this formulation, we expect better estimates of surface fluxes than with the traditional method at locations of weak grid-scale wind and strong convective mixing. The PBL-top mass entrainment is explicitly computed with a formulation that also uses the bulk TKE. We have analyzed the behavior of the simulated PBL. The simulated seasonal cycle of marine stratocumulus over the eastern oceans is realistic; so are the diurnal cycles of the PBL depth and stratocumulus over land. The simulated surface fluxes of latent heat, and short wave radiation heating at ocean surface show significant improvements over previous model versions. The simulated surface stress with low resolution, however, is too weak over the middle and eastern tropical Pacific. Consistently, the zonal and meridional components of simulated trades in those regions are too weak. The simulation with high resolution, on the other hand, shows much more realistic wind stresses, SLP pressure and precipitation fields. The AGCM with the revised PBL parameterization coupled to the MIT oceanic GCM has a superior performance in the stratocumulus regions over oceans and continents. The simulated equatorial gradient of SST in the Pacific is very realistic, and that in the Atlantic has at least the correct sign. In the eastern equatorial Pacific, both simulations show a cold tongue that is asymmetric about the equator. In the equatorial Pacific the interannual variability resembles ENSO. The simulation with the low resolution shows wind stresses underestimated by a factor of approximately 2, and a rather prominent double ITCZ in both the Pacific and Atlantic oceans. The simulation with high resolution, on the other hand, shows much more realistic wind stresses, and the double ITCZ bias in the Pacific is not eliminated but at least considerably alleviated. These results suggest that increased AGCM resolution can alleviate systematic errors of coupled atmosphere-ocean GCMs.