Dear Mr. Alshawaf,

I had close look at your manuscript, before I checked the reviewer comments and your responses. As you can see reviewer II, a leading expert in data assimilation in meteorology, recommended rejection of the manuscript. Although I wouldn’t go that far I agree with quite a few points made in her/his assessment, particularly that the study would greatly benefit from increasing the N i.e. the number of cases before drawing general conclusions. I also share the reviewers concern about the fact you merge information sources with different spatial error characteristics in a geo-statistical context. Reviewer I is less critical with his recommendation (minor) revision but puts the finger on the key deficiencies in the presentation with respect to a better explanation of the MERIS data (which are used for verification here) and the problem of anisotropy due to fronts, which can (depending on being either zonal or geometric) not or easily be accounted for in a geo-statistical context.

Given the partly very critical nature of comments on one hand and the soundness potential of the approach, I think the paper needs to be thoroughly revised, before it might potentially be accepted. Please make sure that you to properly address the critical comments of particularly Reviewer II. On top of that Ialso think the paper needs improvement with respect to the quality of the presentation and transparency of the methods:
- The introduction would greatly benefit from properly stating a science question or even better an hypothesis that is tested.
- I would avoid the term spatially highly variable in the context to interpolation methods. Any interpolation method is a weighted average and smooths out variability (in contrary to simulation methods).
- Please be accurate: a la grange multiplyer is not introduced to assure an unbiased constraint but to assure unbiased estimates...
- There is a little too much self-citing with respect to the use of WRF and data assimilation, please give merits to the achievements of other groups.
- I agree with reviewer II that the explanation of kriging is not very helpful as is stands. I teach geo-statistics and had problems in understanding what it is about (I usually start with Eq. 7).
What I miss in this context, is that the entire quality of any kriging method crucially depends the quality of the experimental variogram or co-variogram (this is where the information is) and density of the data points.
- Maybe I missed it, but I did not see an experimental variogram or co-variogram from the data sources you use and I didn’t find any point on the theoretical variogram or co-variogram function you are using (neither on the points whether data have been de-trended before variogram estimation).
- I also wonder why you did not cross validate your spatial estimates within a Jack Knifing, to test and compare the quality of the spatial models.


Best regards,

Erwin Zehe

Dear Mr. Alshawaf,

I closely checked your revised manuscript and the comments of Reviewer #1 (who also reviewed the manuscript in the first round) and Reviewer #3, an expert in the field of atmospheric modelling and data assimilation, who I asked to get a third alternative perspective on your work. Both reviewers stress that the proposed fusion approach might be of high potential. Both reviewer also stress that the current study does as it stands not allow to come up with general conclusions on the validity of the method and recommend major revisions, partly because the spatial extent of the model domain of your analysis is too small, deficiencies in the driving model and in the WRF simulations with respect to re-assimilation of available data to assure the best model simulation. The last point was already mentioned by Reviewer #2 in the first round (see his point 1) in the context of timing errors which might contaminate your approach.

The usual practice in HESS in such a situation is to reject a manuscript and eventually encourage the author to re-submit their study (without asking for page charges). However, given the potential of the approach, I am willing to go for another final round before it might eventually be accepted. Please make sure advise you to properly and thoroughly addressed the critical comments of particularly Reviewer #3 and Reviewer # 2 of the first round (e.g. the influence of topography and the MERIS retrieval only from above the clouds, which needs to be addressed within more cases) before you submit the revised version.

Best regards,

Erwin Zehe

Dear Mrs. Alshawaf,

I am pleased to let you know that this manuscript version has been ranked as acceptable for publication.

Best regards,

Erwin Zehe