|I re-reviewed this paper and I thin that it improved considerably as compared to the paper that I was origninally submitted. In its current form, the paper describes material that I consider to be a major contribution to geoscientific model development and thus fits well into the journal geoscientific model development. However, I think that the paper is quite lengthy (27 figures!) and for better readibility I suggest that the authors seek to shorten their paper. In my opinion, many parts of the WRF validation can be relegated to the supporting material. Furthermore, I think that part of the figures belonging to the sensitiviy study and the computational aspects could be summarized in a table, while both sections can in my opinion be shortened as well. Futhermore, I have a few remaining issues. My overall advice is to accept the paper after minor revisions.|
1. I agree with the authors that the use of Monin-Obukhoc Similatiry Theory (MOST) is common practice in many surface parameterizations. In contrast to the study under scope in these studies, the lowest atmospheric model layer or the reference layer in case of an offline model, is usually above the surface and located in the surface layer or the upper part of the Roughness Sublayer. The authors apply a vertical grid spacing of only 2 m, which means that exchange between surface elements and the atmosphere is resolved on very small scales. I question whether MOST still applies on these scales, and exchange between surface elements and the adjacent atmospheric layers should not be better simulated using a more detailed model including a 'law of the wall' approach.
2. The authors verify the developed model using measurements taken on 2 days in only one city. In my opinion this time period is very limitied and I suggest that the authors at least provide a discussion on how representive their results are for the performance of their model in other time periods, seasons, and other cities.
3. The authors use an observed temperature profile to initialize their model, whereas they use WRF output for large-scale forcing. I wonder whether this does not lead to inconsistencies as possibly, the large-scale forcing is not equilibrium with the observed profiles. Why did the authors not consider to use the validated WRF profile to initialise their model?
1. Page 2, line 1: 'As more than a half ...' should be 'As more than half'
2. Page2 , line 6/7: Please add that the UHI is manifests itself mainly during the evening and early night,
3. Page 2, line 19: 'approaches to studying' should be 'approaches for studying'
4. Page 4, line 15 to 28: very long sentence
5. Page 5, line 21: The energy budget doens not contain the latent heat flux. Please, explain here why
6.Page 11, line 6/7 and page 13 line 27: latitue and longitude coordinates are given with 8 digits. Is such detail really necessary?
7. Page 13, line 34: 'in 640 m' should be 'at 640 m'
8. Page 19: line 13/14: My idea is that the deeper morning Atmospheric Boundary Layer (ABL)s in the morning over urban areas are not caused by the UHI, buth that both the UHI and the deeper ABLS stem from the same mechanism, namely the observation that urban areas are able to deliver extra heat to the air layers close to the surface and in the ABL.
9. Page 19, line 15: add 'potential' before 'temperature'
10. Page 24, lines 15 to 20: I think these sentences are quite hard to follow. Probably, these sentences should be replaced by clearer sentences.