The study addresses polarimetric radar estimation of extreme rainfall using specific differential phase K_DP. The performance of different versions of the K_DP-based estimators is compared and tested using rainfall gauge measurements. The authors consider various K_DP processing techniques and possible impacts of the DSD variability and raindrop orientations uncertainty on the quality of the polarimetric rainfall estimation.

It is demonstrated that all tested methods optimized by a locally measured DSD in a particular heavy rain event tend to yield reasonably good estimates of rainfall with hourly totals below 100 mm but significantly underestimate heavier rain with rain rates approaching 200 mm/h. The authors speculate that such underestimation can be attributed to a more random orientations of raindrops than is commonly assumed in the derivation of the R(K_DP) relation or to a big difference between the radar and gauge sampling volumes.

This reviewer believes that the authors may underestimate the impact of downdrafts which are often associated with a torrential rain. They downplay such an impact arguing that it is not detected by the disdrometer. The fact is that the vertical air motion is always negligible near the surface which does not exclude high downdraft speeds just a few hundred meters above the surface where the radar samples rainfall. Under such scenario, radar underestimation of rain is inevitable because of the conservation of a precipitation flux. The authors have a unique chance to check this hypothesis. Of course, direct measurements of the vertical air velocity are usually not available with the operational radars but one can examine indirect radar attributes of downburst such as strong wind divergence near the surface which can be assessed from the Doppler measurements or rapidly descending K_DP or Z cores that are proven to be linked to the downbursts / microbursts.  

I have several other recommendations and concerns.

• What is the purpose of comparing the data from the Luoyang and Zhengzhou radars? Only the latter radar data are used for a qualitative analysis. Once the maps of rain totals retrieved from the two radar are displayed in Fig. 2, the reasons for discrepancies have to be discussed. Most likely, an apparent shift in the areas of heaviest rainfall is related to the differences in the altitudes of the radar sampling volume and strong vertical gradients of rain rates which are typical for warm rain process.
• Self-consistency factor and Clpf should be defined.
• Averaging window lengths (len) have to be expressed in km.
• Keep in mind that the LP procedure always tend to overestimate K_DP because it ignores negative radial slopes of Φ_DP. Negative K_DP is always coupled with overestimated positive K_DP in the close proximity and both shouldn’t be quantitatively used. This is a nature of the nonuniform beam filling impact on the K_DP
• I don’t see any difference between the three panels in Fig. 8, It is not clear what is displayed because the shading is not specified.

English usage has to be substantially improved. Just a few examples follow

• L 32. Should be “coefficient” instead of “ratio”
• L 60. Remove “the” before “progress”
• L 72. K_DP is a radial derivative (not derivation)
• L 120. This “may be attributed”
• L 133. “A linear interpolation”
• L 153. “At 274°”
• L 174. K_DP is less dependent on DSDs than what?
• L 196. “Earlier” instead of “early”
• L 217. “Analyze” instead of “analysis”
• L 222. Replace “decent”
• L 178. “In-depth” instead of “depth-in”
• L 298. Remove “at the”
• 304. Gridded
• 309. Replace “decent” with “significant” or “noticeable”

Review of HESS-2022-0361 

Title: Assessing KDP-based QPE for the record-breaking rainfall over Zhengzhou city on 20 July 2021

Authors: Li, Moisseev, Luo, Liu, Ruan, Cui, and Bao

Summary: This concise paper is an important application and demonstration of KDP estimation uncertainty for a devastating flooding case in China. The authors perform a robust statistical analysis of the different parameter choices for popular KDP estimation algorithms, compared to the operational algorithmn for the CINRAD radar networks. This type of work has been done for synthetic observations, but not for real cases with such dense ground-based (gauge) observations for evaluation. 

The paper is free of any major fatal flaws. However, there are a large number of mostly minor comments that need to be addressed. One structural comment -- there is not really a good concluding paragraph (see the first comment below). Additional proofreading is necessary. For these reasons, I suggest MAJOR revisions, although this is somewhere between major and minor.

Comments:

1. The conclusion section is a nice summary of the study, but it sort of ends abruptly without wrapping up. The authors need to include a concluding paragraph (it can be brief), that brings the focus back out to the broader perspective. This is sometimes referred to as the “funnel technique” or structure. What do the results of this study contribute to the community’s knowledge or application of rainfall estimation? What do you recommend for future work, and how will your efforts contribute to the main goal of mitigating losses from devastating flooding events? Some answers to those questions are needed.

2. L27: “seem falling short” should be “seem to be falling short”?

3. L31: What is meant by “parameterized reflectivity factor”? That is not standard usage, unless it means something different from the traditional equivalent radar reflectivity factor?

4. L32: remove “the” before “attenuation effects”

5. L43: “infrastructures” should be “infrastructure”

6. L67: Maybe a reference for KDP being immune to beam blockage would be helpful to readers less familiar with the dual-pol products.

7. L69: “data was” should be “data were”. Same in L134.

8. L72: “derivation” should be “derivative”. Also, maybe adding “derivative with respect to range” or something like that to clarify?

9. L78: The Maesaka reference should not be in parentheses according to the style guide…same in L147, L169, and elsewhere; also, no capitalization needed for “Linear”

10. L98: are there units for these numbers? Are they range gates?

11. L105: Is there an objective routine for the removal of spikes, or is this done manually? (Either is fine, just indicate how it is done for reproducibility)

12. L117: Is there a threshold used to define “significantly deviating”?

13. L122: “This may attribute” should be “This may be attributed to”

14. L154: “elevation angle dependence of KDP” – this confused me at first. I believe the authors are referring to the viewing angle of raindrops effect, and not any sort of vertical profile of KDP effects? Maybe some sort of clarification would help.

15. L223: the word “cry” is not correct, and I don’t know what is being stated.

16. L234-238: There shouldn’t be any downdrafts experienced at the surface owing to mass continuity (i.e., downdrafts become diverging outflow at the surface). So, this argument does not make physical sense. There almost certainly will be downdrafts above this level, perhaps in the region where the radar is sampling. Are there any low-level divergence observations from the radar or surface stations that could be used to estimate the downdraft speed at radar beam height? (This is admittedly crude, but would help indicate if the scale of the downdraft is significant or not.)

17. L242-243: I’m not sure 6-10 m/s winds would be consistent with a “tornado.” Is this a misunderstanding of the cited reference? In other words, did the storm produce a tornado, but the sampled winds were obviously much weaker? Please clarify.

18. L246-248: This seems really unlikely, in my opinion. Are there any references to support this conjecture about extremely narrow heavy rain shafts? This is important because it is alluded to in one of your conclusions (L325).

19. L258: “imposes obviously oversmoothing filter” should be revised, maybe something like “imposes an overaggressive filter that obviously leads to oversmoothing”?