|Title||Evaluating cloud fraction and ice water content in the ECMWF integrated forecast system and three other operational forecasts models using long-term ground-based radar and lidar measurements.|
|Publication Type||Technical memorandum|
|Secondary Title||ECMWF Technical Memorandum|
|Authors||Bouniol, D, Protat, A, Delanoë, J, Pelon, J, Donovan, DP, Piriou, J-M, Bouyssel, F, Tompkins, AM, Wilson, DR, Morille, Y, Haeffelin, M, O'Connor, EJ, Hogan, R, Illingworth, AJ|
The present paper evaluates the parameters involved in the prognostic cloud schemes of some of these models: the cloud fraction and the ice water content. To avoid mixing of different effects these parameters are evaluated only when model and observations agree on a cloud occurrence. Several comparisons and diagnostics are proposed. As a first step the complete two years of observations are considered in order to evaluate the "climatological" representation of these variables in each model. Overall, models do not generate the same cloud fractions, and they all under-represent the width of the ice water content statistical distribution. For the high-level clouds all models fail to produce the observed low cloud fraction values at these levels. Ice water content is also generally overestimated, except for RACMO. These clouds are considered as radiatively important because of their feedback on weather and climate and therefore their relatively inaccurate representation in models may be significant when computing fluxes with the radiation scheme. Mid-level clouds seem to have too many occurrences of low cloud fraction, while the ice water contents are generally well reproduced. Finally the accuracy of low-level clouds occurrence is very different from one model to another. Only the arpege2 scheme is able to reproduce the observed strongly bimodal distribution of cloud fraction. All the other models tend to generate only broken clouds. The data set is then split on a seasonal basis, showing for a given model the same biases as those observed using the whole data set. However, the seasonal variations in cloud fraction are generally well captured by the models for low and mid-level clouds. Finally a general conclusion is that the use of continuous ground-based radar and lidar observations is a powerful tool for evaluating model cloud schemes and for monitoring in a responsive manner the impact of changing and tuning a model cloud parametrisation.