Time series

cloud cover
cloud liquid water
cloud base height
cloud top height
integrated turbulent kinetic energy
qt at 300 m
qt at 1500 m
Kh at 300 m
Kh at 1500 m

Fields at 21.30 UTC

cloud fraction
liquid water
total water
potential temperature
u-velocity
relative humidity
Kh

Some short summary I made. Please comment on it !
 

The results of the SCMs confirm the results of previous inter-comparisons: most models produce too high values of cloud liquid water and cloud cover. In addition, most models have difficulties with the diurnal cycle, in particular with "dissolving" the cloud after sunset. Some models have a fairly good diurnal cycle, both in cloud liquid water as in cloud cover (e.g., RACMO, MESO-NH); other models are characterized by the complete absence of a diurnal cycle (ECHAM, HIRLAM). The other models are in between these extremes; for example, MESO-INM has a rather good diurnal cycle in cloud liquid water but not in cloud cover, and in ECMWF the clouds appear at the right time but, after sunset thick clouds remain.

In one important aspect, the SCMs can be split in two different groups: one group with mass flux type closures for mixing in the cloud layer  (HIRLAM, ECHAM, ECMWF, RACMO, MESO-NH), and those with moist turbulence schemes performing mixing in the cloud layer (MESO-INM, CT01). Of themodels with mass flux closures, most models (except ECMWF) models had too little activity, resulting in too shallow cloud layers, and too moist profiles in the cloud layer. The most outspoken of these models is the HIRLAM model, with a very shallow boundary layer typical of Stratocumulus clouds. RACMO and MESO-NH have rather good profiles in the cloud layer, though they have somewhat (RACMO) or considerably (MESO-NH) too shallow cloud layers. In some of these models, (over) activity of the turbulence scheme also contributed to the well mixed profiles in the cloud layer. In particular, the bad behavior of the ECHAM model seems to be caused by the strong activity and extremely noisy behavior of the turbulence scheme. The ECMWF model, on the other hand, had a rather (over) active mass flux scheme with too a strong moistening in the inversion and drying near cloud base, in particular around midday. Both models with a diffusion scheme for mixing in the cloud do a fairly reasonable job: both produce a conditionally unstable cloud layer, though too well mixed (in the same degree as the better mass flux models). On the downside, both models produce rather noisy profiles in the cloud layer, in particular CT01. MESO-INM also produces a strong inversion at the cloud top. CT01 is characterized by rather aggressive mixing, with a too diffuse inversion (there are some indications that, for the cloud layer, the turbulent kinetic energy budget is not completely consistent in this model).

Except ECMWF and ECHAM, all models are considerably too moist in the cloud layer. At 21.30 UTC, relative humidity ranges from 92 % (RACMO) to near 100 % (HIRLAM, MESO-INM, MESO-NH), whereas LES predict about 86 %. Suprisingly, despite the high relative humidity some model have rather realistic values of cloud cover and cloud liquid water. On the other hand, despite the too low relative humidity, ECHAM and ECMWF have highest values of cloud liquid water (and cloud fraction).