The effect of varying the source spectrum of a gravity wave parameterization in a middle atmosphere general circulation model

Climate simulations of the middle atmosphere circulation with general circulation models are now starting to include parameterizations of the momentum flux deposition due to unresolved gravity wave spectra. A current uncertainty in the application of such parameterizations is the specification of the imposed gravity wave spectrum. The aim of this work is to quantify the effect of varying within a realistic range the source spectrum of a gravity wave parameterization in a general circulation model. Results from two simulations with the gravity wave spectrum launched at two different heights, the surface and the 110-hPa pressure level, respectively, are compared. Noteworthy differences found in the simulated middle atmosphere response include the following (1) The average temperature in the southern winter upper stratosphere is about 40 K warmer in the experiment with the surface as the launching height, virtually eliminating the typical cold polar bias that affects many general circulation models. (2) Stronger easterlies in the subtropical summer mesosphere, again in the experiment with the surface as the launching height. Diagnostics of the parameterized gravity waves indicate that in the experiment with the surface as the launching height, the net zonal momentum flux transported by the gravity waves is negative just above the troposphere at middle latitudes. This negative net momentum flux facilitates the deceleration of the mesospheric winter westerlies. The meridional circulation induced by such deceleration is thereafter responsible for the substantial polar winter warming. In contrast, in summer the negative net momentum flux limits the upper mesospheric deceleration of the easterlies. In the experiment with launching height at 110-hPa, the gravity wave net momentum flux is instead zero by construction at the launching height.