From a software engineering perspective applied to climate models, it would be interesting to gain a deeper understanding of the architectural decisions behind the development of the model's code. Therefore, we pose the following questions:

1. For the new configuration, were specific design patterns such as Facade or Dependency Injection used to structure the model’s code? This would ensure the separation between physical parameterisations, dynamic processes, and the model logic.
2. To coordinate the interactions between the different physical and dynamic parameterisations, do you use patterns such as Observer or Mediator? If so, what benefits have you observed in terms of performance, maintainability, or scalability?
3. To manage complex parameterisations (such as orographic gravity waves), have specific techniques based on patterns like Strategy been implemented?
4. To ensure the reliability of both the simulations and the implemented code, have automated testing frameworks or static code analysis tools, such as pFUnit and FortranAnalyser, been employed in the development process? If so, how have they contributed to identifying and addressing potential issues in the codebase?  The use of tools such as FortranAnalyser would be interesting to mention in order to be able to verify that the quality of the developed code is maintained or improved with the development of new versions of the software. 

This is a modeling study with a ground to lower thermosphere general circulation model that implements a new non-hydrostatic dynamical core. The study explores the behavior of the UA-ICON model in different configurations with respect to the gravity wave drag implementation or tuning. The main conclusion of this voluminous and exhaustive manuscript is that the implementation of the NWP physics package improves considerably the previous ECHAM implementation. In addition, specific gravity wave tuning parameters are explored and an optimal combination assessed. While the climatological behavior is improved, the authors find that the individual representation of SSW, especially those associated with zonal wavenumber 2, remains deficient.

The results of this study are in line with experiments carried by other groups, now and in years prior. As a reviewer, and a modeler myself, I appreciate these types of studies that are difficult to publish. There is really no science in them, they are an engineering exercise, but one that is necessary to document. With this preamble in mind, I find the manuscript too long and delving into details that are not interesting to the general reader. I do think the study needs to be published, but it also needs some thinning, which I am suggesting in the detailed comments below.

From a general point of view, one difficulty in tuning the gravity wave drag parameterization is the coupling with interactive chemistry. When comprehensive models like WACCM were developed, it was often found that the while a set of tuning parameters reproduced a desirable climatology, the timing of the SH warming (i.e., reversal of the winds) was unrealistic and ozone chemistry was adversely affected. In this sense, what I find missing is a discussion of the timing of the springtime warming in the lower stratosphere, like Figure 10 in doi: 10.1175/2009JAS3112.1. Moreover, the Richter et al. paper is sorely missing from the present discussion; consider also discussing Sassi et al. (doi: 10.1029/2003JD004434).

Detailed comments:

Abstract, line 13: This concept is repeated many time throughout the paper and I might have missed some important details. My reservation about this statement is that the forcing (SST, etc.) is climatological: how do you expect to represent realistically the occurrence of phenomena that are so intimately associated with tropospheric weather, such as wave-2 forcing?

Figure 1. Are these the results of one model simulation or of averaging (climatology) of several instances? If the latter, then you should be able to say something about the range of uncertainty; if the former, I would caution overinterpreting these results. In the following, when difference from observed climatologies are reported, the reader needs to know how significant they are.

Line 171. This is odd sequencing. You cannot show a result and say we will explain later the methodology: the reader needs to see first the methodology and then we can make an informed decision on the validity of the results. Otherwise, the study is reduced to a beauty contest. I suggest moving section 4 ahead of section 3 and streamline the discussion. This will also help the flow of the paper, which instead appears is interrupted. In the same spirit, it may help to reduce the number of cases that are compared: the following figures (through Figure 5) are more appropriate for a PhD thesis, as opposed to a journal paper. Honestly, there is way too much information which could be moved to a digital supplement, if it is supported by the Journal.

Figure 5. Are these hemispheric averages? It would be more informative to exclude the tropics and average only poleward of 30 degrees in each hemisphere. More generally, it is hard to follow each single line in Figure 5. The vertical line plots are useful in the process of tuning a parameterization, but as they are, they become confusing, hard to read and not very useful in a scientific paper. Moreover, since the arguments are clearly developed referring to a global circulation, the reader's mind goes to the zonal mean plots. It would be much more useful to describe the un-tuned simulations in the previous section and introduce the tuned GWD schemes in this section.

Figure 7. I suggest marking the ERA-5 and SABER PDF with thick bold lines.  This is a very busy figure; it will be hard to read for many people: I cannot tell the light green from the magenta. I recommend reducing the number of cases reported in these panels (same for Figure 8) to make reading easier. An additional figure with all the cases can be added as a supplemental image.

Figure 8. As noted above, these figures are too busy. Also, I recommend discussing the lower stratospheric temperature, important for ozone depletion.

Figure 10. Very difficult to tell colors apart; use different type of line style.

Line 320. Aren’t these biases the result of excessive dynamical heating?

Fig. 11e,f. Note the persistent divergence at higher altitudes. Why?

Fig. 11, the V* and W* panels. Are these necessary? Again, this looks like more a dump from a PhD thesis; these panels do not add anything to the paper, except space and length.

Figure 12. What are the empty bars? I know it is described in the text, but you don't want the reader to go hunting for that information. Repeat in the caption.

Table 5. I feel these statistics are an overkill. If the goal is to evaluate the climatological behavior, one doesn't need much beyond Figure 12. If the goal is instead to accurately represent individual SSW, then the whole premise is wrong, since, as I stated above, my understanding is that the boundary forcing is climatological.

Line 495. I understand the methodology is borrowed from the YS24 paper but a little more information is necessary here. Is the “migrating” component representative of a 60-day composite, or longer?

Line 505. Need to cite Pedatella et al. 2014 (doi:10.1002/2013JA019421)