Statistical Fluctuations in Convective Forcing Computed from Big-Domain Cloud-Resolving Model Simulations

Abstract
Deep convection takes the form of populations of clouds embedded within mesoscale and synoptic scale disturbances or forced in topographically-favoured locations. Deterministic convective parametrization is based on an assumed relationship between the ensemble mean effect of the clouds and the vertical profile of temperature,moisture and wind (averaged in the horizontal). For the instantaneous cloud forcing effect to be well-approximated by the ensemble mean, the spatial density of clouds must be large compared with the spatial density of model gridpoints (i.e. the number of clouds per gridbox must be much greater than one).

The Met Office cloud-resolving model has been used to quantify the statistics of tropical convection and its forcing effect by coarse-graining the model tendencies. The probability density functions (pdfs) associated with temperature are computed for all points, and separately for points whose parametrized convective tendency (obtained diagnostically from the coarse grained fields) lies within pre-defined ranges. In this way, the pdfs of effective convective temperature tendency can be grouped according to a measure of mean convective forcing strength. Craig and Cohen developed a statistical mechanics approach to modelling convective mass flux fluctuations and this turns out to be helpful in understanding the model-derived statistics.

Some results from aquaplanet simulations are used to quantify the statistical fluctations of convective temperature forcing in the Unified Model and are compared with those derived from the cloud-resolving model simulation.

Back to agenda