The parameterization schemes used to represent clouds in general circulation models have significantly evolved in their complexity over the last 10 yr. This increases the demand for a thorough evaluation of their performance. Several techniques ranging from the evaluation of the model climate to single column modeling have been proposed for that purpose. This paper aims to provide a strategy for an improved, more coherent use of these techniques. An overview of the different techniques is given using examples from the evaluation of the global model of the European Centre for Medium-Range Weather Forecasts. Advantages and disadvantages of the individual methods are highlighted. The paper closes by proposing a strategy to join the different techniques into a coherent procedure of cloud parameterization evaluation.
ECMWF, Shinfield Park, Reading, United Kingdom