Comparison between the gravity wave stress parameterized in a climate model and simulated by the high-resolution non-hydrostatic global model ICON

We compare the parameterization schemes that represent gravity waves in the Atmospheric Component of the IPSL Climate Model (LMDZ6A) and the high-resolution ICOsahedral Nonhydrostatic Weather and Climate Model (ICON). Our focus lies in assessing the capabilities of the gravity wave drag schemes to predict zonal momentum fluxes derived from ICON. The parameterization is run offline using ICON meteorological fields coarse grained to a healpix grid with size representative of an ESM grid (around 100km x 100km). We then examine the temporal mean, horizontal mean, and zonal mean gravity wave stresses predicted by the parameterizations and compare them to the zonal momentum fluxes associated with the ICON subgrid scale fields (e.g. the motions that are filtered out during the coarse-graining). The investigation reveals that in the stratosphere, the parameterizations have some skill at predicting zonal momentum fluxes of ICON, and this without prior tuning. More specifically, the parameterized gravity wave stresses due to mountains, convection and fronts align reasonably well with the zonal momentum fluxes from ICON in the stratosphere, each scheme consistently playing a dominant role where it should (frontal waves dominating in the midlatitude storm tracks, convective waves in the tropics, and mountain waves over orography). This permits physical interpretations of the origin of the gravity waves predicted by ICON, but raises challenges when extending this comparison to the troposphere. There, the agreement between the parameterized stress and the ICON subgrid scale stress is much weaker, which is likely attributable to the fact that in the troposphere subgrid scale forced motions like convective cells produce stresses much larger than the gravity wave stresses.