The second referee deems that there are not enough new findings in this manuscript to warrant publication. Your answers point out that your main finding is that CMIP5 models that include only the direct-aerosol effect are unable to reproduce the observed trend of the Asian monsoon in recent decades.
Although, your findings are summarized in your abstract, you should have more explicitly stated them in the introduction. Please show how your study stands apart from previsous published ones and brings new insights into this Asian monsoon trend .
In addition, you should strenghten this point has in your conclusions.

I would like you to work out these parts of the manuscript and point me to the changes you make in highlighting the original findings of this paper. This should alleviate the concerns of the referee.

Thank you

Yves Balkanski

Your answers to the reviewers comments are often deflecting the improvement that the reviewers asks you to include in the paper. You cannot just answer with general comments without making the requested changes or add-on to the paper.
Let me give you 3 instances where I found your replies not precise or not pointing to the changes requested. Please point to me to the exact changes made in the text.

First instance:
3) Recent study showed that the CMIP5 models fail to simulate the post-°©‐1950 decreasing trend of monsoon rainfall (Saha et al., 2014). The failure of representing large-°©‐scale changes caused this issue in most of the CMIP5 models. Only 8 out off 48 CMIP5 models used is able to capture the monsoon rainfall features (Saha et al., 2014). This indicates that skill of simulating Asian summer monsoon is very poor in most of the CMIP5 models. Authors need to comment if models are poor in simulating essential features, how much reliable the model-°©‐ simulated impacts of 20th century aerosol emissions on rainfall trends.

It is by no means clear that reproducing the observed trend in a phenomenon is a suitable first-°©‐order metric of whether a model can successfully simulate that phenomenon. Contemporary studies judging the skill of models in simulating the Asian monsoon tend to focus on features more directly connected to the mechanisms involved, such as the mean rainfall pattern (its strength and distribution) and the seasonal cycle, as well as other aspects of variability (e.g., Annamalai et al., 2007; Sperber et al., 2013). This is counter to the reviewer’s suggestion that reproducing the observed trend is the first-°©‐order metric of a model’s success. . Indeed, in order to interpret any trend, we must consider its drivers, and if those drivers are being successfully simulated. There is little doubt that when we consider idealised future scenarios out to the end of the 21st century (such as doubled CO2 concentrations), or even the IPCC-°©‐style economics scenarios (such as SRES or now the RCP), then monsoon rainfall is shown to increase (e.g. see the review in Turner and Annamalai, 2012). Inherent in these scenarios is the dominance of greenhouse gas forcing and relatively low aerosol, and we also have good theoretical arguments to link warming with increased moisture supply to the monsoon. Thus, the lack of increasing monsoon rainfall towards the end of the 20th century under such global warming conditions, and indeed the presence of a weakening of the monsoon, was regarded as puzzling (Goswami et al., 2006).

Our study, and several others that we cite, in fact offers an explanation for differences in the ability of global models to simulate the trend alluded to by the reviewer. The suggestion is that anthropogenic aerosols, particularly scattering aerosols such as sulphate, are acting to counter the effects of warming over South Asia, and even override them. Their emissions and burdens over South Asia and the northern hemisphere have increased strongly since the mid-°©‐20th century. There are no other known mechanisms by which the monsoon can decrease.

The point of our paper is thus that it is only the CMIP5 models that most faithfully represent aerosol processes (including both direct and indirect effects) that can capture this declining rainfall trend. Without the indirect effect, a model can thus not truly represent the full magnitude of negative radiative forcing. Our study is thus consistent with the reviewer’s suggestion that only some models are able to capture this trend. In fact, our own analysis of other studies such as Ramesh and Goswami (2014) and Saberali et al. (2014) suggests that the few models that do reproduce the observed trend do contain indirect effects of aerosol.

Instance 2:
5) In the Conclusion section, authors discussed about the increases in monsoon rainfall in future CMIP5 projections. Sabeerali et al., 2013 found that most models produce too much (little) convective (stratiform) precipitation compared to observations. Is a too strong aerosol effect causing this issue? Better to add remarks on this issue in current context of the work. Whether aerosol indirect effects are enhancing the convective rainfall in the model?

In fact, studies have shown that once aerosol‐cloud interactions are included in GCMs, then both marine stratiform clouds and stratiform precipitation have been improved (Wood, 2012). This again suggests that inclusion of aerosol indirect effects leads to a more faithful representation of the physics involved. Our search of previous studies of the interaction between aerosol indirect effects and convection suggests that they are generally performed using radiative-°©‐ convective equilibrium models or large‐eddy models, and there have been no consistent conclusions to support the viewpoint that aerosol indirect effects are enhancing convective rainfall in models.

Looking in more detail, the representation of aerosol-°©‐cloud interactions vary from model to model. For many models, the aerosol‐cloud interactions apply only to stratiform clouds in any case and not convective clouds. In theory, with more aerosols to act as cloud condensation nuclei (CCN), which reduces drizzle, stratiform precipitation can be reduced.

We also note that caution must be used in interpreting the comparison between modelled convective and “large-°©‐scale” rainfall with the convective and stratiform rainfall measured by products such as TRMM.


Instance 3:
When you answer this specific comment I need you to be specific about which sentence was added. I cannot make a wild guess, I need the extact sentence(s) to judge if you address the reviewer's comment:
(1) Page 30641, Lines 4‐7: Whether the simulated aerosol distributions (aerosol optical depth, column burden) are too high over China? All the CMIP5 models are very poor in the simulating the aerosol‐distributions over South Asia, especially Indian region. If the emissions are large enough, then aerosol distribution or deposition should be wrong. Rewrite the sentence.

There is certainly large uncertainty in measurements of natural and anthropogenic aerosols over South and East Asia (especially complicated during the monsoon season when remote sensing measurements may be unreliable due to cloud cover). Thus it is very difficult at present to assess the CMIP5 model skill at simulating these distributions. However, observations do indicate that the strongest aerosol burdens in South Asia are found in the northern plains region, closest to the Himalayas. This is consistent with the common aerosol emissions used to drive the CMIP5 models, and our resultant multi-°©‐model mean of sulphate and black carbon aerosol loading shown in Fig. 5a,b.

However, regardless of the agreement with observations, the multi-°©‐model mean aerosol loadings over China are much stronger for sulphate (approximately double) as shown in Fig. S1 below (and added as Fig. 6 to the new paper). This helps inform our closing hypothesis that aerosol loadings may be large enough over China for the direct aerosol effect to act alone in weakening East Asian monsoon rainfall, and warrants further investigation

New text has been added in the conclusions and discussions, please refer to reply to technical comment (5).


Last, but not least, you answers to my comments are too brief. I would like you to do more than add a single sentence to answer them:

In the last paragraph of Introduction, added:
“… Various authors have questioned the ability of CMIP5 models to
capture observed monsoon rainfall trends over India (e.g. Saha et
al., 2014) and therefore the reliability of CMIP5 at making
projections of future monsoon rainfall (Ramesh and Goswami,
2014; Sabeerali et al., 2014). In our study we shall demonstrate
that it is only when aerosol indirect effects are included in models
that they are able to capture the observed trends. …”
In the fifth paragraph of Conclusion, modified:
“… While we generally have more confidence in our models if the
present-day simulations perform well at simulating the mean
monsoon, its seasonal cycle, and variability (Turner and Annamalai,
2012; Ramesh and Goswami, 2014; Sabeerali et al., 2014), a key
novelty of our study is that it is only the models containing aerosol
indirect effects that can reasonably be expected to represent the
observed trend.”

After you address these remarks I will re-consider the paper towards being accepted in ACP.

Sincerely,

Yves Balkanski

The authors have answered all the concerns I raised with very sound scientific arguments. This manuscript is now ready to go into publication in ACP.

Yves Balkanski