Sensitivity of General Circulation Model Performance to Convective Parametrization

Abstract

This chapter describes the sensitivity of aspects of global circulation model (GCM) performance to convective parametrization. This is a difficult task for a number of reasons. Firstly we have to define what is meant by “performance”. From a numerical weather forecasting perspective a model must produces a good estimate of the three dimensional structure and evolution of synoptic scale weather systems. On climate timescales the model should capture a realistic mean distribution of thermodynamic variables (including cloud cover), wind and surface precipitation. For coupled ocean-atmosphere modelling the surface fluxes of heat and water vapour need to be correctly simulated. Also the simulated variability of the atmosphere (from diurnal through to the intraseasonal and interannual timescale) needs to be well captured. Convective parametrization has a large influence upon all these areas and it is difficult to make definitive statements covering each area here. Secondly it is difficult to make definitive statements concerning the impact of convection schemes alone as other parametrization (for example the boundary-layer scheme), together with the methods used to simulate resolved motions, also have a impact upon the ability of atmospheric models to simulate the features of the general circulation referred to above. Even if identical convection schemes were used in two GCMs in which other processes were represented in a substantially different manner, it would not be certain that the distribution of precipitation simulated, together with other features of the mean climate, would be similar. However as will be illustrated below, some features of GCM simulations appear well correlated with certain aspects of convective parametrization.