General comments:

The paper presents an interesting study that statistically compares polarimetric radar observations from three storms systems in northwestern Germany to ensemble modeling results.  I enjoyed reading the paper and found the results to be interesting, valuable, and very worthy of publication.  I do, however, have several comments that I feel would improve the manuscript.  I will address these in more detain in the next section.

Individual comments:

1) While I do have a background in polarimetric radar observations and, to some degree, the use of polarimetric radar data in numerical models, I am unfamiliar with most of the models used in this study. Upon my first reading, I must admit that I was a little overwhelmed by the numerous acronyms that were being introduced and found myself continually going back to refresh my memory.  After gett8ng a few pages in, I skipped to the back of the paper to see if there was perhaps an appendix that summarized the list of acronyms.  Is this something the authors might consider?

2) In section 3.2, I really don’t feel like I had a good understanding of how the 20 ensemble members for each case were obtained. That is, it states the 20 ensemble members represent “uncertainties in model physics and lateral boundary conditions by combining five model physics perturbations with four global models are used for the initial and lateral boundary conditions”.  I found this description to ban little vague.  Can you be more specific about what those five model physics perturbations and, more importantly, four global models are?  And the results from those runs are used as the initial and boundary conditions for the COSMO runs?  Also, what does COSMO-DE stand for?  COSMO is introduced earlier but, looking back into the paper, I was unable to find what COSMO-DE referred to. 

3) At the beginning of section 5.3, there is a short discussion of clustering. Clustering, I believe, refers to how combined plots of two polarimetric variables will cluster in multi-dimensional apace.  This seems totally unrelated to computing convective area fractions of a single radar variable, such as reflectivity.  Also, can the authors provide a more complete description of convective area fraction (CAF)?  I know this is a concept that has been used in numerous papers, but without description I am often left wondering if the convective area fraction is with respect to the grid being used, or with respect to all reflectivity points (for example) above a certain dBZ threshold?  It seems to me that a CAF can be defined in many different ways.  Also, how is CAF impacted if, for example, a portion of the system that is being studied is moving off of the grid over which the CAF is being sampled?

4) In sections 5.3.1, 5.3.2, and 5.3.3, I am confused why the elevation angle 8.2 is used for a PPI for cases 1 and 3 (Figs. 5 and 9) and an elevation angle of 1.0 is used for case 2 (Fig. 7). Using an elevation angle of 8.2 for a PPI seems very unusual.  Please explain why such high elevation angles are being used for these plots.

5) Overall comment on the figures, my philosophy has always been that figure captions should contain enough information that they could be “stand alone”, i.e., that the reader should be able to fully interpret the figure without having to refer back to the text. That being said, I feel that much can be done in this manuscript to improve figure captions and, in a few cases, the figures as well.  As an example, I felt that the caption describing the right most panel of Fig. 2 could have been much better, particularly the description of the rightmost panel.

6) The text states that there were 104 GRDC stations, 36 were considered useful for this study. These figures show 22 or 23 stations, but the figure is very crowded with all of the stations grouped together in the rightmost 2/3rds of the figure with lots of unneeded and wasted white space on the left, etc.  If several of the stations are not going to be used (presumably those that were to be lotted on the left side of each of the figures), I would suggest eliminating the “which space” and making the figure easier to read.  Also, are there 20 asterisks representing the 20 ensembles plotted for each station with the some of them just overlayed on each other?

7) Figure 4: Time labels need to be improved.

Specific comments:

1) Overall, the paper is well written. There are some very minor grammatical issues throughout the text.  I’ll make just a few suggestions here.

2) There is also some inconsistency throughout the paper on whether the word “modeled” and “modeling” should be spelled with one “l” or two “ll’’s.

3) Line 7: Remove “however”.

4) Line 9: Suggest replacing “besides” with “in addition to”.

5) Line 20: Suggest rewording from “Polarimetric radar observations provide ZDR,…” to “Besides ZDR, polarimetric radar observations provide…”

6) Line 32: Suggest changing “thus e.g. provides insight on new snow generation” to “thereby providing insight into the generation of new snow”.

7) Line 33: Suggest changing “measure for the diversity” to “measure of the diversity”.

8) Line 34: Change “These informations” to “This information”.

9) Line 47: Remove “e.g.”

10) Line 49: Suggest changing “operator and due” to “operator due”.

11) Line 50: Remove “e.g.”

General comments:

This study utilizes ensemble Terrestrial Systems Modeling Platform (TSMP) simulations with forward dual-pol radar operator to evaluate the performance of simulated cloud microphysical processes for three summertime convective storms over northern Germany using bias corrected X-band radar observations. The paper presents some interesting results and contributes to scientific community in this field. However, the methodologies (section 2 and 3) needs to be more clearly written and reorganized to be published. In addition, there are many grammatical errors and typos that require corrections. Thus, the reviewer suggests the manuscript to be reconsidered after major revisions are made with the following conditions.

Major comments (scientific questions/issues):

1) The abstract is too general and does not provide a solid conclusion based on the study results. Please revise the abstract.

2) The authors introduce many notations in sections 1 and 2 without proper explanation or actual use of the equations. e.g. backscatter differential phase (delta), aerosol size distribution (R2), the logarithm of its geometric standard deviation log(sigma), solubility (epsilon), aspect ratio (AR), width of canting angle distributions (sigma), etc.

Please provided proper equations to these notations.

3) In sections 1 and 2, authors cite too many online references or unpublished (not peer-reviewed) articles. The reviewer is skeptical with some of the research results mentioned in the paper. Please correct them or update them to more recent peer-reviewed papers. 

4) This study uses TSMP to study the evaluation of modeled summertime convective storms using polarimetric radar observations. However, in sections 3.1 and 3.3 the paper discusses about land cover types and root zone soil texture without any context. Based on the three cases which are only for short convective periods, the reviewer finds little purpose of the coupling nor its relation to dual-pol study.

In addition, lines 201-205, evaluation of streamflow and discharge in the model has little to do with the research purpose. Again, section 5.2 river discharge serve little purpose in the paper. The locations of streams are not even shown in Figure 3, which makes it impossible to understand the results.

The reviewer suggests either to change the title and research purpose or delete the above mentioned descriptions.

5) Section 4.1 discuss synoptic situations of three cases but the reviewer finds it difficult to follow the descriptions. Some figures depicting synoptic conditions are needed.

6) The description in lines 260-262 is incorrect. Observations do not gradually increase as modeled CAF. In figure 4, there is no explanation of what (a), (b), (c) are. Further, the snapshot time periods (boxed area) between observations and model runs are different. The paper does not mention this issue or provide reasons of different time period selection. “Optimal” is not enough.

7) In sections 5.3.1, 5.3.2, and 5.3.3, there are many leaps in steps and logic in terms of how locations, elevations, and time is determined for analyses (figures 5 – 10). How can you compare the radar signatures with model simulations when the locations, elevations, and time periods are different? The range plots in Fig. 5 and 6 is also different and very difficult to compare. In Fig. 9 why is 8.2 degree elevation used?

8) Captions in figures 2, 4, 5-10 need to be improved. For instance, box plots of ensemble members in Fig. 2 is not explained well.

9) The first sentence in section 6 (discussion) is improperly placed. Further, in order to address the change in IC/BC and its influence on the simulated cases, the experiment needs to be conducted on the same cases. But, this paper does not. Thus, lines 424-429 should be deleted.

Minor comments (technical corrections):

1) There are many acronyms used in this study without spelling them out properly (e.g. EMVORADO,  DE, GME, JUWELS, a.g.l., etc.

2) There are many typos and unnecessary use of “e.g.” used in this paper. Please correct them.

3) L59: There are many more recent X-band study results that are published. Please include them as references.

4) L75: Runge-Kutta is not a “dynamical core” but “numerics” to solve PDE equations.

5) L90: No à No_x

6) L175: The two X-band radar used in this study is calibrated based on GPM DPR (Ka band). However, there are many literatures that show ground based and airborne radar signatures are much different in characteristics and have lots of biases. Please provide some evidence of how DPR can be used as reference and not the other way around.

7) L224: How does anticyclonic rotation of the warm front produce the necessary lifting mechanism?

8) L231: Between “ensemble average.” and “For the first case,” there needs to be a sentence explaining Figure 2.

9) L237: northwestern domain à northeastern domain

10) L278: How can you tell where the melting layer is based on Fig. 5a. Explain in more detail.