Thorwald Stein

Department of Meteorology, University of Reading

Radar observations of clouds and precipitation provide detailed information on the microphysical and macrophysical structure of convective systems. Now that operational weather services run their models at convection-permitting resolutions, the challenge is to find appropriate measures for radar-based model evaluation. Simulators calculate radar reflectivities from model hydrometeor fields using the model microphysical parameters, providing a like-with-like comparison that avoids the need for microphysical assumptions on the radar measurements.

Results are presented from the Cascade project, in which the UK Met Office model was set up in a limited-area mode for West Africa and the Indo-Pacific to study the impact of model resolution on the representation of tropical convection. The CloudSat simulator was run on the West Africa simulations and results are presented comparing the vertical cloud structure of the West African monsoon with a CloudSat observed climatological cloud structure. Several model errors are highlighted, particularly the height and extent of anvil cloud and the abundance of shallow precipitation.

In the DYMECS project (the Dynamical and Microphysical Evolution of Convective Storms), storms were tracked in the southern UK using the Chilbolton precipitation radar, providing a large data set of storm structures to evaluate Met Office forecasts. At the standard 1.5 km resolution, the modelled storms tend to have too large precipitation areas, but at higher resolution (500 m and 200 m) the storm structures are comparable to those observed by the radar. However, errors in storm size distribution and storm-averaged rainfall distribution are evident at all resolutions.

These results indicate what blind spots in cloud observations remain to be resolved for further evaluation studies. A second point for discussion will be the appropriateness of simulating radar reflectivities from single-moment microphysics schemes.