Review of “Exploring new EarthCARE observations for evaluating Greenland clouds in RACMO2.4” by Thirza N. Feenstra et al. 

Atmospheric Measurement Techniques (AMT): egusphere-2025-5623

General comments

It is well known that a polar regional climate model is a valuable tool for estimating the surface mass balance of the polar ice sheets, which governs ice sheet mass balance and, in turn, global sea level. Therefore, it is necessary to continue developing such a polar regional climate model to provide more reliable climate information on snow/ice accumulation/ablation over polar ice sheets. In this study, the authors focus on the Arctic region around Greenland and compare the polar regional climate model RACMO (version 2.4p1), widely recognized in the global cryosphere community as a reliable model, with EarthCARE observations of cloud microphysics. As stated in this paper, the authors plan to improve RACMO's overall performance by leveraging the knowledge gained from such comparisons. This is undoubtedly a novel challenge that has not been conducted in the global cryosphere community. In this paper, only two case studies are presented because EarthCARE observations are available only from May 2024; however, I believe this study has the potential to serve as a future benchmark for the polar regional climate modeling community. In this respect, this paper fits well with the scope of the special issue (SI) entitled “Early results from EarthCARE”. This manuscript is generally well written and easy to follow. Therefore, I suggest that this paper can be considered for publication in this SI once the authors address the following points. 

Specific comments (major)

L. 124: What do the authors mean by “radiative effects of clouds” calculated by the McICA (Monte Carlo Independent Column Approximation) method here? Do the authors mean heating rates by clouds? Or the contemporary clear-sky downward radiation? Please explain in more detail.

Sections 3 and 4: The authors compare EarthCARE observations and RACMO simulations and clearly explain the analyzed features. However, their agreements or disagreements are, in my opinion, mainly explained subjectively. I believe the authors must provide statistical information, such as mean difference, root mean square difference, and correlation coefficient. If such quantitative model evaluation results are provided in this paper, I think this study could serve as a future benchmark for the development of polar regional climate models.

Sections 3 and 4: In general, it is difficult for a regional atmospheric model to simulate the exact timing and location of cloud formations. In other words, the cloud appearance times and locations simulated by a regional atmospheric model often disagree with reality. Therefore, I think it is necessary to use more model-derived data (e.g., data before and after the target time, and data from east and west of the satellite paths) to obtain more meaningful insights into the model's performance.

L. 460 ~ 465 “These case studies suggest that some of our previous tuning choices should be reconsidered, such as the doubling of the snow sedimentation velocity, which now appears overestimated. Additionally, the process of conversion of ice to precipitating snow might also be overestimated, leading to overly rapid snow particle generation, resulting in ice clouds dissipating too quickly. Currently, the persistence of supercooled liquid layers might likely be suppressed by a too strong Wegener-Bergeron-Findeisen process, which converts too much liquid water into ice crystals.”: My impression from this study is that it is too early to argue so. This is because the authors made only two short-period comparisons in a single year (2025) and a limited season (spring). I assume the RACMO team must achieve good model performance throughout a year to estimate a realistic surface mass balance of the ice sheet, meaning the authors must make such comparisons across multiple years and seasons to confirm whether the argument is truly valid. 

Specific comments (minor)

Title: The polar regional climate model RACMO is well recognized in the global cryosphere community; however, I don’t know whether it is also famous among the readers of the journal AMT. My impression is that it is better to add something like “the regional climate model” before RACMO2.4.

L. 74 “a higher horizontal resolution”: Can the authors add quantitative information for the horizontal resolution of the EarthCARE measurements?

Sect. 2.2: Are the EarthCARE measurements evaluated against in-situ measurements, something like upper air observations with radiosondes? If yes, can the authors briefly introduce this point?

L. 165: The names of five EarthCARE products are listed here. Can the authors briefly explain what properties we can obtain from these products?

L. 181 “the maximum modeled atmospheric wind speeds”: At which level? Please explain.

L. 184 “Since cloud processes are relatively slow ~”: I understand that the authors want to state that the polar clouds are steady within ten minutes or so. Can the authors add a reference for this statement?

L. 274 ~ 275: To present the large-scale atmospheric flow towards the southeast clearly, I think it is better to expand the area in Fig. 2a. Please consider showing a figure with a larger domain during the target period by using the parent ERA5 data. This is also the case for Fig. 7.

L. 311 “The previous findings ~”: It is better to specify.

Technical corrections

L. 48: I had an impression that the intention of the first sentence in this paragraph is similar to that of the previous paragraph (L. 30): Both sentences state that it is challenging for climate models to simulate polar clouds accurately. Suggest rephrasing “Evaluating cloud microphysical representation in climate models is particularly challenging for polar regions, as ground-based observations are limited (Shupe et al., 2013).” to something such as “Ground-based observations that can be used for the evaluation of cloud microphysical representation in climate models are limited (Shupe et al., 2013).”

L. 72 ~ 74 & L. 76 ~ 78: I had an impression that the following two sentences explain almost the same thing. Can the authors merge them? “EarthCARE not only extends the CloudSat and CALIPSO observational record but also marks a big step forward by delivering the first exactly co-located measurements of clouds, aerosols, and radiation from space at a higher horizontal resolution than ever before.” and “By combining observations of the four different instruments, an atmospheric lidar, a cloud profiling radar, a multispectral imager, and a broadband radiometer, EarthCARE provides observations of the vertical structure of clouds, aerosols, and radiation in unprecedented detail.”

L. 95: “will” can be removed.

L. 127 “very”: It sounds subjective. It is better to be removed.

L. 140 “more accurate”: Compared to what? 

General comments:

The manuscript “Exploring new EarthCARE observations for evaluating Greenland clouds in RACMO2.4” presents an evaluation of the regional climate model RACMO using observations from the active sensors onboard the recently launched EarthCARE satellite. Evaluating the macrophysical and microphysical characteristics of clouds in weather and climate models and constraining them using observations is one of the core objectives of the EarthCARE mission. In this sense, the study is timely and well aligned with the goals of the mission.

From a modeling perspective, this work demonstrates the potential value of EarthCARE observations for model evaluation and future model improvement. From an observational perspective, it provides a useful example of how model users can practically exploit EarthCARE measurements, which may also help inform product update planning within the satellite community.

The manuscript is generally well written and scientifically relevant. It also focuses on the question of how EarthCARE observations can be used to evaluate RACMO, which fits well within the scope of AMT. However, several important issues noted below need to be addressed before the manuscript can be considered for publication. Therefore, I recommend major revisions.

Specific comments

Major comments

#1. CPR reflectivity simulation

The authors use an ATLID simulator to compare ATLID backscatter with RACMO output, which is a reasonable choice. For the CPR reflectivity comparison, however, they rely on empirical relationships (Eqs. 1-7) rather than using a scattering-based radar forward model (e.g., PAMTRA; Mech et al., 2020), which may limit the robustness of the comparison.

These empirical relationships are statistical fits derived under specific conditions and do not represent variations in cloud microphysics, particularly changes in particle size distributions and densities. For instance, the Z-T-IWC relationship from Protat et al. (2007) used in Eq. (1) is known to exhibit regional variability at reflectivities above about -15 dBZ. Similarly, the Z-LWC relationship from Matrosov et al. (2004) used in Eq. (2) was derived primarily for non-precipitating marine stratiform liquid clouds. This relationship can be sensitive to CCN conditions, and its coefficients (i.e., 2.4² here) may therefore vary with region and season. Moreover, the Z~LWC² assumption is only valid for cloud droplets and breaks down once liquid water evolves into drizzle or rain, as scattering transitions away from the Rayleigh regime. Finally, the attenuation relationship applied to snow (Eq. 3) was derived under dry snow conditions.

As a result, part of the discrepancies between the observed and simulated reflectivity shown in Figs. 3 and 8 may result from errors in the reflectivity simulation itself, not only from deficiencies in the model. These errors are expected to increase with increasing reflectivity and may therefore have influenced the authors’ conclusions.

Using a radar simulator (e.g., PAMTRA; Mech et al., 2020) together with an EarthCARE CPR instrument model (e.g., Orbital-Radar tool; Pfitzenmaier et al., 2025) would likely make this comparison more robust and reliable. This would be particularly great if the microphysical assumptions used in the simulator were aligned with those used in RACMO. If this is beyond the scope of the study, the authors should at least more clearly discuss the assumptions, limitations, and potential biases associated with the empirical relationships used, and carefully consider these aspects when interpreting the reflectivity comparisons shown in Figs. 3 and 8. Where possible, the authors could also consider whether alternative empirical relationships that are more appropriate for the cloud regimes considered here might be available.

#2. Lack of quantitative evaluation

Throughout the manuscript, the comparison between RACMO and EarthCARE observations is often described using qualitative terms like “underestimate” or “overestimate” without a clear indication of the magnitude of these biases. For example, it remains unclear whether the differences in backscatter, reflectivity, cloud top height, or ice water content correspond to systematic biases or regime-dependent behavior (e.g., stronger overestimation for higher water contents but reasonable agreement for weaker ones).

Including simple quantitative metrics (e.g., mean, median differences, relative biases, or percentile comparisons) would substantially strengthen the conclusions. In particular, such information would be very helpful when considered alongside forward model simulation errors and observational retrieval errors, as it would clarify whether the remaining discrepancies can reasonably be attributed to RACMO itself, or whether they are comparable in magnitude to forward model or retrieval errors.

#3. Curtain-based comparison

The manuscript focuses on detailed comparisons for two selected case studies. Using a limited number of cases is not a problem, and the authors provide useful information on the environmental context of each case. However, an important limitation of the current analysis is that the model-observation comparison is restricted to cross sections.

The RACMO may exhibit not only biases in cloud intensity, but also spatial displacement errors in the horizontal. For cloud systems with limited horizontal variability, this may not be a major issue. However, for more heterogeneous cloud fields, apparent underestimation or overestimation could partly reflect horizontal mismatches between the modeled and observed cloud fields. For example, in Section 3.2 (lines 304-314), the manuscript said that RACMO underestimates cloud top height and water content over the Baffin Bay region. But it is difficult to exclude the possibility that this discrepancy arises from horizontal differences in cloud location.

I therefore suggest that the authors more explicitly acknowledge this limitation in the manuscript. Alternatively, the authors could provide additional evidence that horizontal variability in cloud top height and water content (or simulated reflectivity) is limited for the selected cases, or include complementary analyses (e.g., CFAD-like comparisons) that better support the interpretation of systematic model biases.

#4. Uncertainties of EarthCARE IWC products

In the manuscript, the EarthCARE IWC products (from ATL-ICE and CPR-CLD) are used as a reference when evaluating RACMO, but their uncertainties and possible biases are not discussed. One of the key messages of the paper is that RACMO underestimates IWC. However, the relatively high IWC observations (mainly from the CPR-CLD product) may themselves be biased high. Unfortunately, at this early stage of the mission, EarthCARE microphysical retrievals have not yet been fully validated.

Given this situation, it would be helpful if the authors treated the IWC comparison more cautiously and discussed the current level of confidence in these products. Also, pre-launch, forward model based studies (e.g., Mroz et al., 2023; Mason et al., 2023) provide useful guidance on the expected uncertainties and potential systematic biases and could be referenced in this context.

In addition, I understand that the authors used the most recent CPR-CLD baseline available at the time of their analysis (i.e., baseline BA). However, CPR-CLD is a rapidly evolving product, and noticeable changes in retrieved IWC have occurred between baseline BA and the more recent BB and BC versions (see the Product Disclaimer; https://earth.esa.int/eogateway/missions/earthcare/data). This raises the question of how sensitive the main conclusions of the manuscript are to the product version used. If possible, comparing two different CPR-CLD baselines for the selected cases would be informative. If substantial differences are found, updating the analysis to the latest available version would be recommended.

Minor comments

#1. (lines 134-135)

The wording “profiles of clouds, aerosols, and radiation” is a bit misleading. While cloud and aerosol properties are treated as vertical profiles, radiation is not.

#2. (line 136)

The term “molecular” is used in connection with air density. While related, these are not strictly equivalent. Please clarify this description to avoid confusion.

#3. (lines 146-147)

The statement that “the CPR can fully penetrate through clouds” is somewhat too strong. While a 94 GHz w-band radar generally has much greater penetration capability than a lidar, significant attenuation can still occur in regions with heavy precipitation or high liquid water content, potentially leading to strong signal weakening or even signal loss. Please consider refining this statement.

#4. (lines 150-159)

Please add appropriate references for the instrument specifications mentioned here.

#5. (lines 160-161)

At the time the manuscript was written, synergy products (e.g., ACM-CAP) were not yet released. But these products became available as of 1 December. While it is not necessary to use them in this study, it may be worth briefly noting that these products have become available since 1 December (the same applies to lines 238-239 and 251-252).

#6. (line 161)

Is there a specific reason why Level 1B data were used for lidar backscatter and radar reflectivity? Level 2A products provide things like corrected reflectivity, which might be more useful.

#7. (lines 161-166)

EarthCARE Level-2 products are provided separately by ESA and JAXA. To avoid confusion, please clearly indicate whether the products used in this study are from ESA or JAXA.

#8. (line 229)

For “attenuation for liquid water,” it might be helpful to clarify explicitly whether rain water is included or not here.

#9. (lines 254-257)

As the authors mentioned earlier, lidar can only see the top in the presence of liquid water, and radar alone cannot directly detect supercooled liquid water. So, in mixed-phase clouds, EarthCARE can only identify the upper boundary of the supercooled liquid layer, while the phase below this layer remains uncertain. A brief reminder of this limitation here would be helpful.

#10. (line 259)

While it is correct that Level-2 data became publicly available in March 2025, observations are also provided for earlier periods. So, the data release date itself does not seem directly related to the choice of cases after March 2025.

#11. (line 283)

The phrase “large snowfall amounts” is unclear. Do you mean snow water content or surface snowfall rate? Please clarify.

#12. (Figure 2)

If water content values below 10^-7 kg m^-3 were excluded, it would be clearer to remove the lower-end extension of the colorbar. In addition, panels (a) and (b-e) use different units (g m^-2 vs kg m^-3). Using a consistent unit system (e.g., g m^-2 and g m^-3, or kg m^-2 and kg m^-3) would improve readability.

#13. (Figure 2 caption)

In the phrase “shown in (a), for (b) cloud ice, (c) cloud liquid water, (d) cloud snow and (e) cloud rain,” please consider avoiding the terms “cloud snow” and “cloud rain.” Snow and rain are precipitation categories and can occur both within and below clouds.

#14. (lines 313-314)

Please check whether “cloud water content” is the most appropriate term here. Snow and rain water content are also included and may occur below cloud base. In addition, the conclusion that RACMO underestimates water content is largely based on radar reflectivity, which depends not only on water content but also on particle density and size distribution. In mixed-phase clouds with supercooled liquid water, high reflectivity could also be associated with rimed ice particles.

#15. (Figure 3)

Were reflectivity values below -35 dBZ masked out? If so, it would be clearer to remove the lower-end extension of the colorbar.

#16. (line 320)

Is the term “snowfall” the most appropriate here? Snow may refer both to in-cloud snow and to precipitation below cloud base.

#17. (line 322)

It would be helpful to clarify what is meant by “snowfall” in this context. The AC-TC product distinguishes multiple ice categories (e.g., snow, rimed snow, heavy snow, snow + SLW…). Even when a radar gate is classified as snow within clouds, cloud ice may still be present but undetected. Below cloud base, however, snow is more likely to represent true snowfall (without cloud ice).

#18. (lines 322-324)

The description here does not seem fully correct. Because radar cannot reliably separate clouds from precipitation regions, AC-TC uses the term “snow” without explicitly distinguishing cloud and precipitation. This does not mean that precipitation regions are classified as clouds. Please consider rephrasing.

#19. (lines 334-337)

It may be useful to first clarify how EarthCARE distinguishes ice and snow, as this definition may differ from that used in the model. For example, optically thin ice detected by ATLID but not by CPR may correspond to very small particles, which would reasonably be classified as ice from an EarthCARE perspective. It is not clear whether there is evidence that such cases should instead be interpreted as misclassified snow.

#20. (lines 337-338)

Radar reflectivity scales with the sixth power of particle size. If small cloud ice particles grow into larger snow particles, reflectivity does not necessarily have to remain very low (it depends on density and particle sizes though).

#21. (line 341)

Please clarify what altitude range is meant by “mid-level altitudes.”

#22. (line 342)

The statement that "RACMO produces mixed-phase layers that are too shallow” is not entirely clear, as the two appear rather similar. Please clarify this point.

#23. (lines 342-343)

Could the absence of detected supercooled liquid water at this altitude be due to overlying supercooled liquid layers or thick ice clouds?

#24. (Figure 4)

In panel (a), the scattered “all ice” classification between 82-83 latitude is real? The AC-TC product may provide quality flags that could be used for quality control. In addition, regions labeled as “all liquid” are within the CPR surface clutter regions. These regions may be better described as areas where CPR observations are not reliable, rather than true “all liquid” clouds. I would recommend excluding surface clutter regions from the analysis.  

#25. (lines 356-357)

When stating that RACMO overestimates snowfall rate along the western margin, please clarify the altitude level, as this is difficult to assess visually. In addition, snowfall estimates from EarthCARE are not reliable within the surface clutter region.

#26. (line 361)

Please specify in the main text which product was used to obtain the CPR sedimentation velocity.

#27. (Figure 6)

Please check whether sedimentation velocity is appropriately visualized. I can see sedimentation velocity in regions where CPR reflectivity is not shown in Fig. 3e, even though CPR Doppler measurements are only reliable for reflectivities above -15 dBZ. In addition, both snowfall and sedimentation velocity are not reliable within the surface clutter layer.

Technical corrections:

#1. (line 125)

In the phrase “the McICA method (McRad; )”, a reference is missing.

#2. (line 226-227)

In “snow water content,” the word “snow” seems to be a typo. Please check.

#3. (lines 233-234)

Units for LWP and WVP are missing.

#4. (line 286)

In the phrase “in Fig. 2a-d,” panel (e) seems to be missing. Please check.  

References

Mech, M., Maahn, M., Kneifel, S., Ori, D., Orlandi, E., Kollias, P., Schemann, V., and Crewell, S.: PAMTRA 1.0: the Passive and Active Microwave radiative TRAnsfer tool for simulating radiometer and radar measurements of the cloudy atmosphere, Geosci. Model Dev., 13, 4229–4251, https://doi.org/10.5194/gmd-13-4229-2020, 2020.

Pfitzenmaier, L., Kollias, P., Risse, N., Schirmacher, I., Puigdomenech Treserras, B., and Lamer, K.: Orbital-Radar v1.0.0: a tool to transform suborbital radar observations to synthetic EarthCARE cloud radar data, Geosci. Model Dev., 18, 101–115, https://doi.org/10.5194/gmd-18-101-2025, 2025.

Mroz, K., Treserras, B. P., Battaglia, A., Kollias, P., Tatarevic, A., and Tridon, F.: Cloud and precipitation microphysical retrievals from the EarthCARE Cloud Profiling Radar: the C-CLD product, Atmos. Meas. Tech., 16, 2865–2888, https://doi.org/10.5194/amt-16-2865-2023, 2023.

Mason, S. L., Hogan, R. J., Bozzo, A., and Pounder, N. L.: A unified synergistic retrieval of clouds, aerosols, and precipitation from EarthCARE: the ACM-CAP product, Atmos. Meas. Tech., 16, 3459–3486, https://doi.org/10.5194/amt-16-3459-2023, 2023.