Comparing Model-produced Convective Cloudiness with Observations

Abstract

Convective cloudiness generated by a cumulus parameterization scheme of a large-scale numerical weather prediction model was compared with analyses of clouds observed by geosynchronous satellites. The comparisons were performed over an equatorial Pacific Ocean region and the Amazon basin region for the period January-February 1979. Fractional coverage of predefined areas by clouds with tops in several altitude-temperature categories was averaged in each area for the entire period of record and also for individual times of the day. The overall mean provided a basis for comparison of the general spatial distributions of observed and model-produced cloud, while hourly departures from the overall mean allowed a comparison of the diurnal variations. Geosynchronous satellite cloud analysis serves as a useful reference for assessing the performance of a moist convective parameterization scheme. In this case, it was found that the subject convective scheme produces an accurate relative horizontal distribution of convective cloudiness, but produces too large a proportion of deep convection compared to overall convection. This was believed to be due to a lack of an entertainment mechanism in the scheme. The diurnal variation of model-produced convection was 9?12 h out of phase and somewhat smaller in magnitude than the observed variation over the ocean region. Good agreement was noted between the diurnal characteristics of observed cloud and model-produced convection over the land region.