Climate Change Impact Assessment on Freshwater Inflow into the Small Aral Sea

Round 1

Reviewer 1 Report

Dear Editor,

I found the paper "Climate Change Impact Assessment on Freshwater
Inflow into the Small Aral Sea" interesting.

The authors develop a hydrological machine learning model to mimic the freshwater inflow into the Small Aral Sea, which is, in my opinion, the most important aspect of this paper.

Then, the authors use climatic simulations for a number of GCMs for attempting possible forecast according to different scenarios.

In this regard, it is unclear why the authors claim to use the RCP4.5 (line 58), while in Figures 6, 8, 9 and 10 they use the RCP6.0. Please, correct or explain.

In any case, the authors discover that the GCMs significantly contradict each other and, therefore, it is not possible to use them for actual forecasts not even if the real RCP for the future was known. I think that also this result is important because it suggests that climatic science is still not mature enough.

I suggest the authors try some explanation for the contradictory results among the models. For example, does the different equilibrium climate sensitivity to radiative forcing of the models matter?  

  Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper presents an interesting experience on the machine learning application to a specific hydrological problem, i.e. monthly runoff prediction in the Aral Sea basin. While the subject is relevant and interesting, I have multiple concerns in respect to this manuscript, which is the reason it should be substantially revised and reconsidered for publication.

My main concern is that the authors are too cryptic in much details all across the article, and I am never sure whether it is by negligence or by intention. The manuscript leaves too many 'why?' questions open.

Why formation and transformation zones? What hydrological basis is put behind this? While this may be true, and I am convinced that there are reasons for this subdivision, they are left unexplained and this is why I pull attention to this matter.

Why gridded runoff dataset while you could just model RCP hydrology with the same HBV model you have used for a gridded dataset production?

Observed monthly data (or rather daily data averaged by months, I would assume) are cited as personal communication, this means they are of unknown and potentially dubious quality; at least the methods used to produce this dataset are to be explicitly explained.

Results should be separated from Discussion; give runoff changes in Results and reflect on potential outcomes for the Aral Sea basin and future hydrology. Better description on the potential changes in the climatic variables should be given. Why are these changes expected - is this owing to more/less precipitation? changes in PET? Besides, reanalysis results at least for the historic should be provided (like PET distribution in the region etc).

More comments could be found in the attached file. This manuscript should be revised, resubmitted and re-evaluated.

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf