Do We Need Parameterised Convection to Get the Model Climatology Right?

Convection is one of the most difficult processes to parameterise in global circulation models and its impact on the large-scale flow is not well understood. If parameterised convection is turned off, such that the only precipitation is resolved at the grid scale, what impact will this have on model climatology and the large-scale flow? Using model data from the Selected Process On/Off Klima Intercomparison Experiment (SPOOKIE) we find that the ensemble model climatology, and in particular the large-scale circulation, is similar for runs with and without parameterised convection. The differences in large-scale fields are smaller than perhaps might have been expected, with similar total precipitation and precipitation patterns in the two cases.

Although the broad model climatology is similar with and without parameterised convection we do find subtle changes in gross moist stability and vertical velocity profiles. Specifically, there is more entropy export when the parameterised convection is turned off, especially in the western Pacific. Furthermore, the east and west Pacific vertical velocity profiles capture the general shape of the profiles from reanalysis but the magnitude changes when parameterised convection is turned off. In order to interpret these results, we investigate the behaviour of a moist aquaplanet model with and without parameterised convection at different resolutions. Our results suggest although the large-scale, time-averaged fields of comprehensive climate models are not particularly sensitive to parameterised convection schemes, subtler changes in the distribution of tropical moisture and temperature fields may impact a model’s ability to respond to climate events and will influence local and time dependent phenomena.