Quantitative comparison between simulated cloudiness and clouds objectively derived from satellite data*

The cloud cover and cloud water content obtained from integration of a limited area mesoscale prediction model are compared with the corresponding quantities derived from AVHRR and SSM/I satellite data. The model has an elaborate condensation-cloud parameterization scheme, in which cloud water is a prognostic variable. From AVHRR data, both cloud cover and liquid water content are retrieved by utilization of radiances from the visible, near infrared and infrared channels. From the 37GHz polarization difference in the microwave data obtained from SSM/I measurements, liquid water content is interpreted; this is only possible over the sea. An important feature of this comparison exercise is that the retrieved data are distributed on the same grid as the model. The difference between the simulation and the analysed cloud fields can thus easily be quantified and exhibited on the model domain. It is demonstrated that this type of retrieved meteorological parameters constitutes potentially useful data samples for quantitative verification of model simulations. There are some weaknesses in the retrieval methods that have to be removed before we can have sufficient confidence in the data to make quantitative verification meaningful. Furthermore, calibration against cloud parameters obtained through other measurements is necessary, although the three data sets show a remarkably good agreement in pattern and range of magnitude in the liquid water content. Even if improvements are needed, the present study has made it possible to make inferences about some deficiencies in the parameterization scheme of the model.