Within the European CLIWA-NET project, an integrated profiling technique (IPT) has been developed to determine continuous vertical profiles of temperature, humidity and cloud liquid water (LWC) during non-precipitating cases. This technique consists of a Bayesian retrieval algorithm which combines ground-based 95 GHz cloud radar reflectivity profiles (Z), lidar-ceilometer measurements, 19 microwave brightness temperatures, the closest operational radiosonde profile measurements of temperature, humidity and pressure, standard ground-level meteorological measurements, and a priori data from a LWC climatology derived from a single column cloud model using explicit microphysics and initiated with local radiosonde profiles. In a first step cloud base and cloud top are estimated from lidar-ceilometer and cloud radar. Then a Newtonian iteration applying a microwave radiative transfer simulation and a Z-LWC relationship is performed until convergence is reached, whereby the solution is forced to be consistent with the a priori information obtained from the radiosonde measurements and the cloud model data. Other reasonable physical constraints such as saturation within the cloud boundaries are also met. It is shown, that crucial importance lies within the correct estimate of the error covariance matrices of radiative transfer and a priori information. For each combined measurement, optimized and physically consistent profiles of temperature, humidity and LWC together with their retrieval errors can been derived.

During the third extensive observation period of CLIWA-NET (BBC campaign) in the months of August and September 2001 at Cabauw, the Netherlands, the IPT was successfully applied to simultaneous measurements described above. The derived LWC profiles were vertically integrated and compared to independent retrievals of liquid water path showing a good agreement. Theoretically calculated LWC accuracies bring forth strong improvements compared to single instrument techniques. However, it is shown that forward calculations in the infrared of the optical region may be successfully included to increase accuracies even more.

One main objective of CLIWA-NET was to evaluate and improve existing cloud parameterization schemes within state-of-the-art numerical weather prediction models (NWP), which compute LWC as a prognostic variable. During the BBC campaign, the IPT retrievals of LWC were compared to one global NWP and two regional climate models that were laterally forced from global analyses. It is shown that the models only poorly reproduce the retrieved LWC profiles if exact matches in time and space are compared and that the models produce immense differences among themselves. However, if the average of all non-precipitating model cases over the BBC campaign is compared with the average IPT retrieval, fairly good results are obtained.