|I find the revised manuscript improved, easier to read and follow. The authors have considered my comments and I have now only two minor comments below for the authors and some corrections to the text.|
1) The authors answered to my comment n.3 writing that “…we are convinced that the global effects of the discretization cannot be neglected in some cases as the step-wise representation of the surface causes e.g. significant change of the surface area, roughness length, normal angles, mutual visibility, etc..”
Their rotation test shows indeed that there is an effect on heat fluxes in the most extreme case of a rotation of 45 degree for a flat surface. However, in my opinion the rotation test included by authors show a very partial view and I remark what I expressed in my original comment:
a) The error in the simulation is mainly induced by the radiative model not the LES flow model, i.e. not by the mask method in LES but by the stepwise representation in the radiative transfer model. The correction of geometry for the radiative transfer model can be done irrespective of the representation of the geometry in the LES. Therefore, the change in the representation by using IMB in the LES flow solver mentioned in 6.2, line 1006, is somewhat unrelated to the present issue.
b) The effects discussed by the authors will become quickly negligible when the resolution of the model approaches the uncertainty in surface representation, i.e. what in the model appears as a flat surface is in reality characterized by protruding objects (e.g. balconies, windows frames etc.., see also e.g. figure 22 top-left) that have sizes of the order of a meter or so. The uncertainty in the representation of the surfaces is likely comparable to the error induced by the discretization as soon as the resolution in the model is of the order of about 1 meter.
I mention that the new detailed discussion in sect 5.1.5 explains that the unrealistic spikes in the modelled temperatures (e.g. Fig. 11) can be mainly attributed to the DEM used for the buildings not being sufficiently accurate, this seems to confirm the view that I expressed above in point (b).
2) Figure 9, please add the Kolmogorov inertial range scaling in the figure for reference.
Line 78, remove “complex” does not seem appropriate.
Line 129, 140. Rewrite the sentences removing “Important”, it seems unnecessary.
Line 597,”5.1.5” not “5.1.3”.
Line 860, I do not think that “expedient” is the correct word.