Cloud albedo changes in response to anthropogenic sulfate and non-sulfate aerosol forcings in CMIP5 models

Frey, Lena; Bender, Frida A.-M.; Svensson, Gunilla

doi:https://doi.org/10.5194/acp-17-9145-2017

Articles | Volume 17, issue 14

https://doi.org/10.5194/acp-17-9145-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

https://doi.org/10.5194/acp-17-9145-2017

© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 17, issue 14

Research article

|

31 Jul 2017

Research article |

| 31 Jul 2017

Cloud albedo changes in response to anthropogenic sulfate and non-sulfate aerosol forcings in CMIP5 models

Lena Frey, Frida A.-M. Bender, and Gunilla Svensson

Download

Final revised paper (published on 31 Jul 2017)
Preprint (discussion started on 10 Jan 2017)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review of Frey et al., "Cloud albedo changes in response to anthropogenic sulfate and non-sulfate aerosol forcings in CMIP5 models"', Anonymous Referee #1, 08 Feb 2017
- AC1: 'Response to reviewer #1', Lena Frey, 01 May 2017
RC2: 'Review of ‘Cloud albedo changes in response to anthropogenic sulfate and non-sulfate aerosol forcings in CMIP5 models’ by Lena Frey et al.', Anonymous Referee #2, 20 Mar 2017
- AC2: 'Response to reviewer #2', Lena Frey, 01 May 2017

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Lena Frey on behalf of the Authors (01 May 2017) Author's response Manuscript

ED: Referee Nomination & Report Request started (05 May 2017) by Philip Stier

RR by Anonymous Referee #1 (30 May 2017)

Suggestions for revision or reasons for rejection

The main parts of the revised paper, in my understanding, are:

\begin{enumerate}
\item Analysis of the temporal variability of preindustrial and present-day AOD
and cloud droplet number on monthly scales in CMIP5 models in stratocumulus
regions; natural variability is found to be large compared to the
anthropogenic perturbation, and the anthropogenic change in CDN is found to
depend most strongly on anthropogenic sulfate AOD.
\item Use of the cloud-fraction--albedo framework to investigate the effect of
AOD and vertically integrated cloud droplet number on cloud albedo in CMIP5
models; the relationship is stronger than in satellite observations, but the
authors do not come to any systematic conclusions about the representation of
aerosol--cloud interactions in the models investigated in this study.
\end{enumerate}

Both in content and in clarity of the presentation, this version is a major
improvement over the first submission. An application of the
cloud-fraction--albedo framework to the CMIP5 models has not been published
before, and the authors' findings in both main parts of the paper are worthy of
publication. There are, however, a few points that should still be addressed.

Methodological comments (page and line numbers refer to the track-changes
manuscript supplied as part of the author response):

\begin{itemize}
\item I like the addition of cloud droplet number to the analysis. However, I
expect the \textit{vertically integrated} cloud droplet number to be strongly
correlated with cloud thickness (which in turn strongly affects the cloud
albedo). Thus, ``cloud brightening'' due to CDN is probably a mixture of CDNC
and LWP brightening, but the paper treats it as purely CDNC brightening. For
most of the models used in this study (except for CSIRO and NorESM), the CMIP5
archive also contains the cloud-top CDNC (variable name cldncl); I think the
authors need to test whether their conclusions based on CDN also apply when
they use CDNC, or otherwise reword the manuscript significantly.
\item Abstract, last sentence: in the usual taxonomy, aerosol--cloud
interactions are divided into a CDNC effect, an LWP effect, and a cloud
fraction effect. Since the cloud albedo is evaluated at fixed $f_c = 1$, I
don't see what other variables apart from LWP and CDNC the albedo could depend
on.
\item In Figs.~3 and 4, the three fit lines always appear to have the same $y$
intercept. Are the fits constrained to the same $y$ intercept? If so, does
it make a difference to the results whether this constraint is enforced? If
so, is there a conceptual reason to prefer one over the other?
\item Conclusions, p.~14, l.~32: since the gradients are not normalized to the
AOD variability, the dust-dominated regions can have large gradients even if
the model representation of ACI is insensitive to dust. This paragraph might
present a stronger conclusion if you looked at the gradient normalized to the
spread in AOD.
\item It would be interesting to see the equivalent of Fig.~3 or 4 for a region
or model where the AOD or CDN gradient is not consistently positive.
\end{itemize}

Comments on the presentation (page and line numbers refer to the track-changes
manuscript supplied as part of the author response):

\begin{itemize}
\item p.~3, l.~12: the phrase ``second order'' is interesting enough that I
would like to see a few words of definition
\item l.~21 ff.: Is it possible to link this result to the underestimate of
aerosol ERF from present-day variability?
\item p.~9, l.~21: ``all regions and all experiments'' -- from the caption, I
thought Fig.~3 only showed one region
\item p.~10, l.~21: where does the number 8\% come from?
\item Fig.~2: A logarithmic vertical scale would improve the legibility of the
non-dusty panels and make it easier to follow the discussion in Sec.~3.2
visually.
\item Figs.~3 and 4: these figures are swapped; also, if it's not too much
trouble, it would be useful for the reader to see a present-day satellite plot
for comparison.
\item Is there a reason for switching the horizontal and vertical axes between
Figs.~2 and 5?
\item I am not a huge fan of the use of $\nabla$ as an abbreviation for gradient
(it should be reserved for actual spatial gradients), and having $\nabla$ and
$\Delta$ appear on the same plot is a bit of a cognitive challenge, but I
understand that some shorthand is necessary. It may help to reorder the
variables so that the $\nabla$s all appear before the $\Delta$s.
\end{itemize}

The newly added text in particular could use a thorough proofreading; the
bibliography has the standard problems (inconsistent capitalization,
inconsistent abbreviation of journal names).

Hide

RR by Anonymous Referee #2 (04 Jun 2017)

Suggestions for revision or reasons for rejection

Second review of Frey et al., ‘Cloud albedo changes in response to anthropogenic sulphate and non-sulfate aerosol forcings in CMIP5 models

The authors have made improvements to the manuscript since the last review. In particular, the methodology is clearer, and they have moved away from using only AOD as an aerosol metric. The article should be published, following a few more minor revisions.

Page 1, line 10: ‘The magnitude of cloud albedo changes in response to the aerosol changes are most closely related to changes in cloud droplet number and liquid water path in the models.’ This seems obvious to me – my understanding is that these are the only related variables in models. In most climate models reff is a function of CDNC and LWP, optical depth is a function of reff and LWP, and albedo is a function of optical depth. It would be much more interesting to know if one of reff or LWP is more closely related to the change in albedo.

Page 4, L30: This repeats the sentence at line 27, but is phrased much better. I recommend moving this to line 27, in place of the original L27 sentence.

Page 5, L19: I suspect there is a degree of model tuning behind this statement. Model AODs are much more uniform than the aerosol loadings, where even sulfate can have a factor of 4 spread in the global mean (e.g. Wilcox et al., 2015). Since you focus on the indirect effect, I think it is worth highlighting this difference, as you are likely to see effects of it in your results. Of course, AOD is easier to observe than aerosol loads, so it is harder to know which model is closer to the truth in this case, but an awareness of model diversity in aerosol load is useful for interpretation of the modeled cloud changes.

Page 7, L27: The CDN variability is primarily [typo in manuscript] related to anthropogenic sulfate, but this is not to be expected from the model parameterization of CDNC. CDNC in the models is a function of the mass/number concentration of hydrophilic aerosols, which is, in all the cases where it is specified in literature for the models used in this study, a linear sum of the relevant species, with no preferential weighting given to sulfate. CDN variability is primarily related to sulfate because sulfate accounts for most of the mass (as shown earlier in the manuscript, and in previous studies). The only model I can find where this finding might be a true result of the parameterization is HadGEM2-ES, where the number concentration of hydrophilic aerosols is a function only of sulfate and sea salt (Bellouin et al., 2007; Wilcox et al., 2015).

Figure 3: Should the color bar be CDNC, not albedo?

Figure 3/Page 8, L2: I think it would be interesting to see the results for LWP. We know from the parameterization in the models that increased CDN and LWP will increase albedo. It would be interesting to see which has the greatest effect in each model.

Figure 4: Should the color bar be AOD, not CDNC? Are Figures 3 and 4 paired with the wrong captions?

Figure 5: Is there a reason for using green twice? More distinct colors may be helpful.

Page 8, L16: Why do you think this is?

Page 9, L14: Why do you think this is? LWP?

Page 13, L21: CDNC typically depends on other species, as well as sulfate. I think you’re seeing the effect of having a larger mass concentration of sulfate aerosols.

Hide

ED: Reconsider after minor revisions (Editor review) (05 Jun 2017) by Philip Stier

AR by Lena Frey on behalf of the Authors (21 Jun 2017) Author's response Manuscript

ED: Publish as is (21 Jun 2017) by Philip Stier

AR by Lena Frey on behalf of the Authors (28 Jun 2017) Author's response Manuscript

Short summary

In this study, the cloud albedo effect in climate models is investigated, separating the influence of anthropogenic sulfate and non-sulfate aerosols. Cloud albedo changes induced by added anthropogenic aerosols are found to be determined by changes in the cloud water content rather than model sensitivity to monthly aerosol variations. The results also indicate that the background aerosol is the main driver for a cloud brightening effect on the month-to-month scale.