|Title: Implementation of aerosol-cloud interaction in the regional |
atmosphere-aerosol model COSMO-MUSCAT and evaluation
using satellite data
Authors: S. Dipu, J. Quaas, R. Wolke, J. Stoll, A. Muhlbauer,
M. Salzmann, B. Heinold, I. Tegen
The manuscript has substantially improved with respect to the first
version. That not all changes (additions) to the text are marked in
bold face makes the evaluation of the revised manuscript unnecessarily
difficult. Some replies were given only as 'reply to reviewer' but
should have entered the manuscript as well. Just one example to
illustrate the point: while the 'reply to reviewer' now states that
Figures 7 and 8 show a 24h average for February 17, this information
is not given anywhere in the manuscript.
Evidence for the claimed superiority of COSMO-MUSCAT over COSMO2M is,
however, still more on the qualitative than on the quantitative side.
With comparatively little effort this could be further improved, I
think, and I would encourage the authors to do so. What must be
improved in any case is the language. In a number of places in the
current manuscript it is not even clear what the authors want to say.
Also, there is still an overall lack of precision (see minor points
The manuscript still requires major revisions to meet GMD standards.
1) The language remains a major issue that must be improved. In
several places, it is not even clear what the authors want to say. I
give only two examples.
p.9, l.8: "Further, the satellite retrieval (mainly thin clouds) are
affected by snow cover, which could be rather ignored."
p.9, l.20: "It is also noted that the cloud microphysics radiation
coupling results in reduction in cloud optical properties, which would
results more downward shortwave and upward longwave especially at the
There are many more sentences that are equally unclear, but I do not
consider it my task as a reviewer to list them all. If none of the
authors has sufficient command of the English language they should
seek assistance from a native speaker.
2) While improvements were made, the manuscript still lacks precision
and relies more on qualitative than quantitative statements. For
example, in the abstract it is said that the cloud effective radius
shows an increase of 1 to 4 micro meter and the cloud droplet number
concentration is reduced by 100 to 200 cm-3. Where do we see that in
the manuscript? I greatly appreciate the newly added pdfs (Figure
5). But if you want to make the point that COSMO-MUSCAT is closer to
MODIS than COSMO2M, why not show also COSMO2M in that figure?
Another example concerns Figures 7 and 8, comparison of shortwave and
longwave fluxes among the different models and with CERES. From
looking at the figures you conclude that differences are neither large
nor systematic. Equally based on just looking at the figures I would
argue that panel 8e (CERES) is most similar to panel 7A (COSMO2M), and
differs more strongly from panels 7e (COSMO2Mrad) and 8a (COSMO-MUSCAT).
Why not remap all the data on the same grid and provide quantitative
estimates (means, pattern correlations, etc) to decide the issue?
p.4, l.11: What is the numerical value of mu?
p.5, l.9: What is TNO?
p.6, l.3: What do you mean by a positional error due to mismatch
between meteorological regimes?
p.6, l.25: You may want to add / state explicitly that COSMO-MUSCAT
treats radiation in the same way as COSMO-2MR.
p.7, l.9: Rainfall of 100 kg m^3? Please clarify.
p.7, l.22: You write that the cloud optical depth of the satellite
data varies between 5 and 54. However, the pdf in Figure 5 shows
values larger than 100. Please clarify.
p.7, l.24: You say that the satellite derived cloud optical depth and
liquid water path are overestimated. Do you mean that the satellite
overestimates these quantities systematically, for example with
respect to surface based observations or reanalysis? If so, please
give a reference.
p.7, l.26: 'cloud droplet radius between 2 and 20 micro meter'. But
the pdf in Figure 5 shows values up to 30 micro meter. Please clarify.
p.8, l.8: 'For cloud optical depth, the model overestimate low clouds
(optical depth below 10)...' Do you mean optically thin clouds?
p.8, l.12: Why not also show COSMO2M in Figure 5?
p.8, l.29: '...satellite derive Nd values are overestimated...' Can
you give a reference?
p.9, l.5: 'However, model derived cloud optical properties strongly
correlate.' With what do they correlate?
Section 3.3: Looking at Figures 7 and 8 the agreement between CERES and
the different models for the surface shortwave radiation seems better
in the absence of the revised radiation scheme. Please comment.
p.10, l.14: "The satellite retrievals suggest the revised model version
is more realistic in both quantities." I find it difficult to draw
this conclusion based on Figure 4 (showing cloud effective radius, but
only for MODIS and COSMO-MUSCAT) and Figure 6 (showing cloud droplet
number, COSMO2M looking more similar (more red) to MODIS than
COSMO-MUSCAT). Could you further corroborate your conclusion?
Conclusions: I think it would be worthwhile to state again that
you consider only warm clouds.
Figures 2 and 6: Does the 100% cloud cover from Figure 2 together with
the close to zero cloud droplet number concentration in Figure 6 imply
that the majority of clouds is ice (i.e. not warm / liquid clouds)?