The article evaluates the potential of the ESA Earth Explorer 11 candidate mission, WIVERN (WInd VElocity Radar Nephoscope), to improve snowfall measurements compared to CloudSat. Using simulations based on ERA5 reanalysis data, it compares the performance of the two radar systems. WIVERN's conically scanning geometry and wide swath coverage significantly reduce sampling errors, providing higher temporal and spatial resolution snowfall estimates. While CloudSat provides reliable estimates only at large spatial and temporal scales (e.g., annual zonal averages), WIVERN achieves reliable measurements at finer scales (e.g., 0.5° x 0.5° grid for 10-day intervals). The study identifies the main sources of error, including sampling errors, radar sensitivity, and uncertainties in the reflectivity-snowfall relationship, and demonstrates WIVERN's superior performance in capturing snowfall variability, particularly in polar regions.

The paper effectively sets out to compare WIVERN and CloudSat and provides a clear methodology using ERA5 data. However, there are several points that need to be addressed before the publication and I recommend a major revision due to some issues in the methodology.

Major points

• Zonal and Regional Analysis:
  
  The paper highlights WIVERN's superior sampling capabilities, but the interpretation of regional results (e.g., Antarctic basins) could benefit from more detailed discussion. For example, if the temporal resolution WIVERN offers is crucial for mass balance studies in these regions.
• Clarity of Figures and Tables:
  
  While figures support the findings, some lack detailed captions or sufficient detail to differentiate WIVERN and CloudSat results effectively. Explaining key trends (e.g., differences in RMSE across snowfall classes) in the text accompanying each figure would improve clarity.

• Future Prospects and Limitations:
  
  While the article touches on potential future research directions, it does not fully address the limitations of the current analysis (e.g., assumptions about the unbiased nature of the reflectivity-snowfall relationship). Adding this discussion would provide balance.
• Improved Approach for Error Statistics

The methodology for computing error statistics (e.g., RMSE, Absolute Bias) appears to be based on instantaneous measurements, i.e., snowfall rates derived from ERA5 at specific times corresponding to satellite overpasses, with modifications to account for uncertainties in the Z-S relationship. If this interpretation is correct, the resulting error distribution will follow a Gaussian distribution with zero mean (assuming unbiased error) and a standard deviation equal to the uncertainty in the Z-S relationship. Consequently, the RMSE decreases as 1/sqrt(N) by definition due to increased sampling, but this approach does not directly assess how sampling impacts the derived climatology. To evaluate the sampling's effect on climatological estimates, a different approach should be considered:

• Define the Observing Period: For example, consider a one-month observation period.
• Compute Monthly Accumulations: Use the ERA5 hourly dataset to calculate the "true" monthly snowfall accumulation (or mean snow rate) for each grid cell.
• Compare with Sparse Spaceborne Measurements: Derive monthly accumulations (or mean rates) from the spaceborne measurements using their more sparse sampling.
• Compute RMSE on Monthly Estimates: Calculate the RMSE by comparing the monthly ERA5 estimates with the monthly estimates derived from the spaceborne measurements.

This approach would more accurately quantify the impact of sampling on the derived climatology, as it evaluates errors at the monthly scale rather than relying solely on instantaneous measurements.

• Misleading Implications About Shallow Snowfall Detection:
  
  The article primarily focuses on snowfall estimation over land, which is critical for ice sheet mass balance studies. However, the introduction gives a misleading impression that WIVERN will observe more shallow snowfall events than CloudSat. While this may hold true over open oceans due to reduced surface clutter at slanted incidence angles, it is not the case over land or sea ice. This distinction is crucial and should be clarified to avoid overestimating WIVERN's capabilities in these contexts. Addressing this limitation upfront would align reader expectations with the radar’s realistic performance in various environments.
• Complexity and Clarity in Figure 4:
  
  Figure 4 presents complex data, and the description lacks sufficient detail to make it accessible to readers. Key concepts, such as the definition of accumulation classes, need clarification. For instance:
• Do the accumulation classes (e.g., snowfall between 36 and 108 mm per year) correspond to snowfall accumulation over the specified period (10 days, one month, or one year) in the ERA5 dataset?
• How is the ERA5 variability computed for these classes? Is it the standard deviation of snowfall rates for grid cells corresponding to the specified range or maybe it’s a mean of the normalised standard deviations?

Although the concept behind the figure is straightforward, the lack of a detailed explanation makes it harder to follow. Additionally, referencing the central limit theorem (https://en.wikipedia.org/wiki/Central_limit_theorem) could greatly simplify the discussion. The explanation could say that the PDF being sampled is the ERA5 hourly snowfall product for each pixel separately, and the difference in sampling (WIVERN with n1​ samples vs. CloudSat with n2​, where n2<n1​) leads to RMSE convergence as std(snow rate)/sqrt(n)​ when n is large. This statistical insight could make the sampling error analysis more intuitive. As the domain size or sampling time window grows the value of n grows too and RMSE decreases. The RMSE will be additionally inflated by the S-Z relationship uncertainty but this will affect both instruments in the same way as n gets larger.

• Blind zone effect

To provide a complete assessment, the paper should include an analysis of how ground clutter affects snowfall statistics for both WIVERN and CloudSat. Currently, this aspect is not addressed, which leaves a significant gap in understanding the limitations of these radar systems. While deriving these statistics directly from ERA5 data would be ideal, it would require extensive effort to analyse the vertically resolved precipitation product. A practical alternative would be to use statistics from the DAR-DAR (Radar-Lidar) A-train product. By deriving a 2D PDF of surface precipitation rate versus cloud top height, the authors could simulate the probability of an event being captured by both radars. This could be done by randomly sampling from the derived PDF. Events with weather system tops falling below the clutter height should have their precipitation set to zero, similar to the treatment in the radar sensitivity discussion. This approach would provide an insightful comparison of WIVERN and CloudSat performance, accounting for ground clutter effects. Obviously, it would not account for processes below the ground clutter height but it will provide a more comprehensive picture of radar limitations in snowfall detection.

 Minor points:

• Due to the orbit repeat cycle of 25 days, the analysis of CloudSat data blow this time resolution period has limited value.
• L106: pencil beam term is used for the delta distribution, use “nadir observations” instead.
• L190: units of snowfall rate should be mm/h
• Figure 7. Some of the colours in the colour bar are repeated or too similar to be distinguishable (e.g. two shades of grey: G7.2 and G3.3 or red G1.2 and G4.1, please use hatching or another way to make them less alike)

This paper documents the expected improvements in snowfall sampling that may be obtained should the ESA WIVERN mission be selected and implemented. ERA-5 snowfall is used as a reference, and comparisons are made between the expected WIVERN 94 GHz radar retrievals of snowfall and simulations of the CloudSat data record. It is shown that, primarily due to the conical scan strategy (vs solely nadir view of CloudSat), WIVERN is capable of returning robust snow climatology information down to the 10 day period. 

The paper is well written and straightforward. The authors acknowledge the uncertainty present in Z-S relationships, and account for this (at least in part - more in my comments below). I think this paper is worthy of publication subject to minor revisions. I enumerate my comments below.

1. p3, line 63 - The authors refer to a snowfall rate in mm per hour. Is this liquid equivalent? If so, please be specific.

2. p3, line 70 - Here you report an angle of 38 degrees, but above you report an incidence angle of 42 degrees. Please resolve this apparent discrepancy.

3. pp4-5, lines 88-103 - Do you draw uncertainties from a Log-normal distribution because of the assumed power law relationship? Please add more detail as to how you include this noise. Do you run a Monte Carlo experiment? If so, how many samples do you use?

4. pp4-5, lines 88-103 - Would you expect there to be bias in the Z-S relationship?

5. pp4-5, lines 88-103 - The distributions in Hiley et al., 2011 were generated by sampling multiple crystal shapes and also allowing a limited range of PSD variability (via the temperature dependence of the Field et al. 2005 parameterization). Can you comment on whether there might be other sources of uncertainty (e.g., the fact that rimed aggregates and graupel were not considered in the scattering calculations)?

6. pp10-11, lines 187-196 - Is there also an effect of the increased attenuation / path length due to 38 degree view angle (vs nadir view from CloudSat) on the WIVERN snowfall estimates?

7. pp. 14-18 - Discussion/Conclusions: A significant advance in WIVERN is the Doppler capability, providing both horizontal and vertical (line of sight) motions. I imagine that the combination of cloud structure (and snow mass retrievals) with dynamics could improve snow estimates even further (above the already substantial benefit of improved sampling). I suggest including some text along these lines.

This submission presents a comparison of sampling errors between a CloudSat/EarthCare type instrument and the proposed scanning WIVERN radar. The authors make a compelling case for the benefits of a scanning architecture. Caveats and shortcomings in the analysis are offered and well described. I recommend publication following some minor revisions. Specific comments are detailed below.

Line 6: Citation(s) for the statement that CloudSat is considered the reference?

Line 7: inability to see shallow or retrieve heavy precipitation.

Line 13: Reword "This radar measurements" - suggest "The proposed radar measurements"

Line 29: Reword to "snowfall not only removes moisture..."

Line 76: Please give some background to ERA5 snowfall. Validation, assimilated data, etc.

Figure 4: These plots are a bit dense and need more explanation or possibly simplification

Line 215: Why not shown? 

Line 220: Why not shown? Maybe not worth mentioning the 10-day time scale given CloudSat's sampling

Line 250: Please add more explanation/narrative here regarding the local variability plots

Figures 8 and 9 would benefit from a legend