We thank the reviewer for their encouraging and positive comments on our work.

We thank the reviewer for their positive and valuable comments which have greatly improved the manuscript. In the following, we address their concerns individually and provide the line numbers where the corresponding changes in the revised manuscript are made.

Answers regarding Main Suggestions

(1): The reviewer is correct that this statement was overconfident. We have now softened the language considerably to reflect that we cannot rule out differences also in temperature and cloud cover. However, we maintain that given the observations that are fed into the ECMWF-AUX data set (temperature, winds, humidity, etc), large errors in temperature and cloud macroscopic variables are much less likely than errors in microphysical variables that are largely unconstrained in the analysis. We have created the zonal mean 2m temperature of ECMWF-AUX and ERA5 and it can be seen, that the temperature of both datasets are quite close together, the latitudinal average difference is added to the text and is 0.24K±0.22K. We chose to compare the 2m temperature as the cloud cover is not included in the ECMWF-AUX data and added the figure to the appendix.

Figure D1: Zonal mean 2m temperature of ECMWF-AUX (black) and ERA5 (dashed-orange) for mid-to-high latitude.

Lines 304 - 307: “It is reasonable to assume that the temperature in the ECMWF-AUX product used in CloudSat-CALIPSO is quite similar to the ERA5 daily mean. A comparison of the 2m temperature between ECMWF-AUX3 and ERA5 shows a latitudinal average difference of 0.24K ± 0.22K (Fig. D1). While atmospheric circulation and overall cloud cover should be well constrained by the observations used in the ERA5 reanalysis.”



(2): Thanks for raising this point. Although it is possible that the higher minimum height bin of the radar used over land may lead to the 2C-SNOW-PROFILE product missing some of the precipitation over land, the fact that we see variability in the fsnow variable in regions where we expect similar cloud types (e.g. over North Central Russia) and to a lesser degree, that the models also show a lower fsnow over land, suggests that this land sea contrast is likely a representative feature. This is consistent with certain weather features such as cold air outbreaks, which once they reach a certain distance from the sea ice are frequently precipitating (e.g. Abel et al, 2017), while during the winter months, Northern Europe is frequently covered by stratus clouds, which are non-precipitating and supercooled (Cesana et al., 2012; McIlhattan et al., 2017).

We have now elaborated on this by stating:

"While the influence of CAOs is also visible in the fsnow patterns across various seasons, with the exception of boreal summer, the fsnow values over land areas tend to be lower. Although it is possible that the higher minimum height bin of the radar used over land (∼ 1000m) may lead to the 2C-SNOW-PROFILE product missing some of the precipitation over land, the variability in the fsnow values in regions where similar cloud types are expected suggests that this land-sea contrast is likely a representative feature. This discrepancy can be attributed to the limited moisture content and shallower boundary layer over land during winter, resulting in lower fsnow compared to ocean areas where CAOs provide deeper, moisture-rich clouds. Once clouds embedded in CAOs reach a certain distance from the sea ice, they frequently produce precipitation (e.g., Abel et al., 2017). In contrast, during the winter months, Northern Europe is often covered by non-precipitating, supercooled stratus clouds (Cesana et al., 2012; McIlhattan et al., 2017). The models also show a lower fsnow over land, supporting the idea that the observed land-sea contrast is a genuine characteristic rather than an artifact of measurement techniques.” on lines 352 - 362.

(3) We thank the reviewer for pointing this out and fully agree. To address this omission we have now included some sentences to discuss the importance of different model versions on the potential differences between our observations and previous studies. This discussion has been added as follows on lines 475 - 488

"Previous studies not only use different metrics but also, in some cases, evaluate different generations of climate models, each with distinct cloud characteristics. For instance, many models included in CMIP5 were found to have too few sLCCs (Cesana et al., 2012, 2015; Tan and Storelvmo, 2016; Kay et al., 2016; Lenaerts et al., 2017; McIlhattan et al., 2017). This deficiency was addressed in several of the next-generation models used in CMIP6, resulting in an increase in the simulated SLF (Mülmenstädt et al., 2021). Although some CMIP6 models used in this study still over- or underestimate the traditional SLF metrics compared to CALIPSO observations, models such as CNRM-CM6-1 and CNRM-ESM2-1 are within one standard deviation of the satellite observations (Fig. F1). Studies focusing on high latitudes have shown that comparing individual CMIP5 models with CMIP6 models reveal improved cloud representation, more closely aligning with observations due to microphysical adjustments in the newer model versions (Lenaerts et al., 2020; McIlhattan et al., 2020). Indeed, McIlhattan et al. (2020) showed that when going from the CMIP5 to the CMIP6 version of CESM the frequency of LCC bias switched direction and overestimates in comparison to CloudSat-CALIPSO. Lenaerts et al. (2020) showed that cloud coverage increased and is slightly overestimated in the CMIP6 version of CESM, but still underestimates ice water path. Consequently, some of the differences between our findings and those from previous studies may be attributed to these improvements in the newer model generations."

(4) In light of this and the previous comment, we have now added specific examples of how the changes incorporated by various modelling centres lead to differences in the representation of cloud properties across CMIP generations. See the previous response to see the added text on lines 475-488.

(5) The following figures show the fLCC (a-d) as defined by McIllhattan et al. (2017) and the fsLCC (e-h) as defined in our study. The lower panel in the figures (i-l) show the difference between the two metrics. fLCC and fsLCC are especially different over the ocean.

We included the additional reference given by the reviewer and added some sentences to the discussion on model versions. Specifically, we added the following sentences in Lines 492-509:

“However, since we utilize a different metric (sLCC instead of LCC), there is no reason to expect the model biases to be identical. Nevertheless, in a further study by McIlhattan et al. (2020) it was shown that the LCC frequency in the newer CESM version is more aligned with the satellite observations except for in the summer months, where it overestimates the LCC frequency.

Similarly, we generally observe that ERA5 and the CMIP6 mean overestimate the fsLCC to various extents during all seasons. These overestimations are likely linked to the microphysical parametrizations of cloud processes that govern cloud phase. This finding aligns with McIlhattan et al. (2020), indicating that while newer model versions have advanced in representing LCC frequencies more accurately, they can still overestimate cloud occurrences, particularly in specific seasons. Precipitation in the new CESM version is more frequent but lighter overall compared to the previous version, which is similar to our findings indicating that the models’s sLCCs produce continuous snowfall, analogous to the "perpetual drizzle" problem (Mülmenstädt et al., 2020; Lavers et al., 2021).

Furthermore, while McIlhattan et al. (2017, 2020) considered a single ESM, our study considers an ensemble of CMIP6 models. Nevertheless, the insights from McIlhattan et al. (2017, 2020) provide relevant context for our finding that ERA5 and the CMIP6 model mean produce sLCCs more frequently than observed in the NH and SH mid-to-high latitudes, especially over the sea ice and land depending on the season (Figs. C1 and C2). In these regions, not only is the frequency of occurrence of sLCCs too high, but the sLCCs are too efficient at producing snowfall in ERA5 and CMIP6 models (Figs. 5 and 6). The latter finding is consistent with the findings of McIlhattan et al. (2017) that LCCs produce snow too frequently in the CESM-LE model.”

(6) We revised the text to include the reference given by the reviewer and to include the individual model variation. Lines 521 - 531 now reads:

“However, the fact that very similar biases are found in the reanalysis as in the CMIP6 models might suggest that surface conditions and weather patterns are not the main explanation for the identified biases. That being said, reanalysis products are known to be less reliable at high latitudes (Liu and Key, 2016). We therefore cannot rule out differences in weather patterns as a partial cause of the biases discussed above, but maintain that deficiencies in the model representation of cloud microphysics is the more likely culprit. The inclusion of both ERA5 and CMIP6 models in this analysis is thus beneficial as it helps highlight potential root causes of the identified biases, although it should be noted that individual CMIP6 models do not all show biases that match ERA5 (Figs. 4 and 8). As stated above, we cannot entirely rule out a modest contribution to the biases from circulation differences, even for the reanalysis. For example, according to Boisvert et al. (2020), all reanalysis products show lower precipitation amounts in the southern Ross Sea and Weddell Sea embayments near the ice shelves. They explain that this is due to persistent cold and dry katabatic wind blowing from the continent across the ice shelves and out over the sea ice, which is not beneficial for precipitation formation and not captured well by the reanalyses.”

(7) We agree that our initial implications were quite broad and general. To more directly link the implications to our study we have revised the two paragraphs starting on Lines 572 - 596 as follows:

“The overestimations of fsLCC and fsnow could stem from how models handle cloud microphysical processes, particularly the WBF process and the partitioning of cloud condensates into the liquid and ice phase. Some identified biases in ESMs may be due to the specific parametrization schemes used in the models, which could rely on simple temperature-dependent cloud phase partitioning. For example, models with a temperature-dependent cloud phase could simulate a fixed liquid fraction at a given temperature within the mixed-phase range, making it impossible for them to simulate an all-ice or all-liquid cloud in that range. Consequently, these models could tend to under- or overestimate the occurrence of sLCCs.

Interestingly, models with more sophisticated microphysics schemes (Figs. F1 and F2) do not necessarily perform better (Figs. 4 and 8). Larger biases in cloud phase can still occur, for example if these models do not accurately represent INPs. This underscores how generational advancements and implementations of observational constrains can significantly influence the comparison and interpretation of cloud and precipitation patterns.

Continuing work should be performed on the comparison between ESMs, reanalysis and observations at specific locations to better understand microphysical processes and their representation in ESMs and reanalysis data. Cloud microphysical scheme development should, therefore, focus on improving these processes in order to reduce cloud phase and surface snowfall biases. More sophisticated schemes should be explored, and ensemble studies with varied microphysical model parameterizations can provide insights into how different parameterizations affect the representation of sLCCs and snowfall in climate simulations.

Ground and aircraft observations in connection with field campaigns at specific locations should, therefore, be increased, especially to improve our understanding of cloud phase, liquid and ice water content, and cloud top temperatures in regions where sLCCs form. These field campaigns should also focus on the processes responsible for snowfall from these clouds.

Finally, as shown in the present study, satellite observations are an essential validation tool for assessing the representations of clouds and precipitation in ESMs. To ensure that future model generations are as accurate as possible, continuous improvements to satellite retrievals are required to further refine our understanding of cloud and snowfall properties, especially in high-latitude regions where limitations in sun angle and ground-track are prevalent. The accurate representation of these cloud and snowfall processes in models, validated through enhanced satellite observations, is essential for reliable simulations of weather and climate. This accuracy is particularly important for making future predictions of the hydrological cycle, which are essential for understanding and mitigating the impacts of climate change.”

Answers regarding Minor Suggestions

(8) Thank you for pointing this out. We have now added the study region as you have suggested in the title so that it reads:

“Evaluation of Biases in mid-to-high latitudes Surface Snowfall and Cloud Phase in ERA5 and CMIP6 using Satellite Observations”

(9) We agree and have now rewritten the introduction paragraph and also included some more specifics about what is meant by ecosystems as recommended below. The paragraph now reads: 

“Snowfall and snow cover have a significant impact on the Earth’s energy budget and the hydrological cycle, especially in mid- and high-latitudes, and thereby strongly impact ecosystems and human societies. Snow cover influences vegetation growth, animal populations, and ecosystem processes, while also impacting economic activities, infrastructure, and health (Callaghan et al., 2011; Bokhorst et al., 2016). Indeed, snowfall is also an important water resource globally (Barnett et al., 2005) and snow cover increases the surface albedo, reducing the absorption of incoming solar energy (Zhang, 2005). However, heavy snowfall events have the potential to negatively impact local communities and infrastructure (Eisenberg and Warner, 2005; Scott et al., 2008; Fox et al., 2023).” on Lines 16-22.

(10) In line with the previous comment, we have now rewritten the Introductory paragraph and included more details about what is meant by ecosystems and human societies. Please see the previous comment for the updated text. 

(11) Thanks for pointing this out and we agree. We have now changed the first sentence of this paragraph to “Cloud phase determines the formation process of snow” on line 23.

Additionally, we have now added a sentence that specifically states how cloud phase relates to snowfall on lines 26-28 that reads:

“The cloud phase affects not only the rate and intensity of snowfall but also the microphysical properties of the snowflakes that reach the ground (Jiusto and Weickmann, 1973; Liu, 2008). ”

(12) Good catch, we have now clarified what we meant here by adding more details from Chen et al., (2020) and Danco et al., (2016) on Lines 42-46.

“The shift in cloud phase towards more liquid and less ice leads to a reduction in the fraction of precipitation falling as snow in previously snowy areas in the Northern Hemisphere (NH), but snowfall events will happen further north in the future (Chen et al., 2020). The shift in isolation would result in an expected decrease in snowfall events and duration of the snowfall season for most regions in the NH (Danco et al., 2016; Chen et al., 2020).”

(13) We have now added some sentences about the specific uncertainties highlighted in the Stephens and Kummerow (2007) and Hiley et al., 2011 studies on Lines 90-95.

“For example, Stephens and Kummerow (2007) identified two primary sources of uncertainty in retrieval methods, errors in distinguishing between cloudy and clear sky scenes and between precipitating and non-precipitating clouds. Furthermore, the forward models used are highly sensitive to their input parameters, particularly the radiative transfer and atmospheric models. Hiley et al. (2011) demonstrated that snowfall retrievals are also influenced by retrieval assumptions and the use of different ice particle models, which can significantly affect the estimated snowfall rates.”

(14) We agree that this was unclear as written. To clarify, the 2 m temperature threshold is applied to the daily mean values rather than the instantaneous hourly values. We have now adjusted  the sentence on Lines 207-210 to:

“We incorporate tcrw with the T2m threshold below 0°C, as this threshold is used to exclude any rainwater below the melting layer. By using daily mean values of tclw and tcrw and applying the daily man temperature threshold, we can analyze the role of supercooled liquid water within clouds and the contribution of liquid water to the snowfall precipitation process in ERA5.” to make this clearer.

(15) The reviewer is correct that the CMIP6 historical simulations do incorporate prescribed natural and anthropogenic historical forcings. Unlike weather forecasts, these simulations are not periodically updated with new climate state information. Instead, CMIP6 historical simulations follow the prescribed forcings and are intended to reproduce observed multi-decadal climate statistics rather than exact sequences of weather and climate events. However, the forcing difference between 2006-2009 and 2007-2010 is negligible and would not lead to noticeable differences between the two simulated periods. The difference would instead originate from differences in internal variability between the two periods, which would in neither case resemble the actual internal variability. Therefore, we argue that the comparison between 2006-2009 and 2007-2010 for selected CMIP6 models would not be meaningful in this context.

(16) That is a great suggestion. We have now updated Fig. 4 and Fig. 8 to show the direction and magnitude of the model bias. We also adjusted the text by removing “absolute” in the corresponding sections.

Figure 4. Magnitude of seasonal area-weighted averages (between |45° − 82°|) of f_sLCCfor CloudSat-CALIPSO, ERA5, CMIP6 models (numbers). The heatmap colors correspond to the differences of area-weighted averages of f_sLCCbetween CloudSat-CALIPSO and ERA5 and CMIP6 models. Green (pink) values indicate underestimation (overestimation) of the individual model with respect to CloudSat-CALIPSO. (a) for the mid-to-high latitude NH, and (b) for the mid-to-high latitude SH. Per season, the smallest (largest) seasonal and spatial differences are outlined with blue (red) lines.

Figure 8. Magnitude of seasonal area-weighted averages (between |45° − 82°|) of f_snow for CloudSat-CALIPSO, ERA5, CMIP6 models (numbers). The heatmap colors correspond to the differences of area-weighted averages of f_snow between CloudSat-CALIPSO and ERA5 and CMIP6 models. Green (pink) values indicate underestimation (overestimation) of the individual model with respect to CloudSat-CALIPSO. (a) for the mid-to-high latitude NH, and (b) for the mid-to-high latitude SH. Per season, the smallest (largest) seasonal and spatial differences are outlined with blue (red) lines.

(17) Thanks for pointing this out and we agree that it was not clear what we meant here. We have now rephrased the sentence on Lines 352-362 to read:

“While the influence of CAOs is also visible in the fsnow patterns across various seasons, with the exception of boreal summer, the fsnow values over land areas tend to be lower. Although it is possible that the higher minimum height bin of the radar used over land (∼ 1000m) may lead to the 2C-SNOW-PROFILE product missing some of the precipitation over land, the variability in the fsnow values in regions where similar cloud types are expected suggests that this land-sea contrast is likely a representative feature. This discrepancy can be attributed to the limited moisture content and shallower boundary layer over land during winter, resulting in lower fsnow compared to ocean areas where CAOs provide deeper, moisture-rich clouds. Once clouds embedded in CAOs reach a certain distance from the sea ice, they frequently produce precipitation (e.g., Abel et al., 2017). In contrast, during the winter months, Northern Europe is often covered by non-precipitating, supercooled stratus clouds (Cesana et al., 2012; McIlhattan et al., 2017). The models also show a lower fsnow over land, supporting the idea that the observed land-sea contrast is a genuine characteristic rather than an artifact of measurement techniques.”

(18) That is a fair point. We have now revised the first paragraph in section 4 “Sensitivity tests”  (Lines 396-405), accordingly:

“In the previous section, we examined the frequency of occurrence of sLCCs (fsLCC) and the frequency of occurrence of surface snowfall (fsnow) from sLCCs in CloudSat-CALIPSO, ERA5, and CMIP6 data. We found that the reanalysis and ESMs overestimate sLCCs and the frequency of surface snowfall in comparison to CloudSat-CALIPSO. These biases in the CMIP6 mean have potentially significant implications for our ability to predict how sLCCs and snowfall might change with future warming. At the same time, it is important to note that although the CMIP6 multimodel mean has this overestimation, some of the individual ESMs are much closer to the observations (green dots in Fig. 3), suggesting that these members may have more representative changes in clouds and snowfall in the future. It is therefore important to ensure that these findings are robust, and not overly reliant on subjective decisions or limitations in the design of the comparison. In the following subsections, we investigate whether the identified discrepancies between the observations on one hand and ERA5 and the CMIP6 models on the other can be explained by sampling biases or instrument sensitivity.”

We extended our statement to a softer language. Lines 448-451 read:

“However, the conclusion that ERA5 overestimates the fsLCC does not change by changing to a higher time resolution (Fig. 9). While the sensitivity analysis for ERA5 data indicates that changes in output frequency do not significantly affect the results, the same conclusions cannot be directly extrapolated to CMIP6 models. However, based on our analysis, it is reasonable to assume that similar results could be observed for individual ESMs within CMIP6, although this cannot be confirmed.”

(19) Thanks for this suggestion, we have now added the magnitude to the text on Lines 409-411:

“Milani et al. (2018) found that applying adjustments and a temperature threshold to the CloudSat snowfall retrieval led to a decrease in the estimated occurrence of snowfall events by up to 30%, primarily in the ocean regions surrounding Antarctica.”

(20) We completely agree and were hoping to look into this more. However, it is not immediately clear from the model documentation which models use a purely temperature-dependent partitioning between liquid and ice. As a quick check though, we have investigated this by plotting the supercooled liquid fraction as a function of temperature on average for each of the models used in this study (see Figure F1 below). If the models relied on something other than temperature i.e. aerosol, then there would be a difference between the NH (solid lines) and SH (dashed lines) in the supercooled liquid fraction. As most of the models have these two lines on top of each other, it is likely that they are only using temperature-dependent partitioning.  Therefore, we assume that MIROC6, CanESM5, AWI-ESM-1-1-LR, MPI-ESM1-2-LR, UKESM1-0-LL, and HadGEM3-GC31-LL are using a more complex phase partitioning while other models like CNRM-CM6-1, CNRM-ESM2-1, and IPSL-CM6A-LR are assumed to have a simple temperature-dependent cloud phase. 

Figure F1: Supercooled liquid fraction (SLF) as a function of temperature for CALIPSO (black) and the CMIP6 models used in this study (color). Error bars on the CALIPSO SLF values correspond to one standard deviation. All values represent an area-weighted average for lat ≥ 45° in the Northern Hemisphere (NH, solid lines) and Southern Hemisphere (SH, dashed lines).

This assumption is further supported by the interquartile range (IQR) of the temperature at various supercooled liquid fractions as shown in Figure F2, where both CNRM model versions and the IPSL model, have a much smaller IQR at the given SLFs than the other models. Despite the fact that MIROC6, CanESM5, AWI-ESM-1-1-LR, MPI-ESM1-2-LR, UKESM1-0-LL, and HadGEM3-GC31-LL have a more variable SLF-temperature relationship, from Fig. 4 it is not clear if these models perform better with respect to fsLCC. This analysis is part of a follow-up study by co-author Haochi Che. 

Figure F2: Interquartile range of the temperature at given supercooled liquid fractions (e.g., SLF 50%, representing 50% ice and 50% liquid) for the different models used in this study for lat ≥ 45◦ over the Northern Hemisphere (NH, a) and Southern Hemisphere (SH, b).

Since there is no clear improvement in fsLCC (Fig. 4) whether the model has a purely temperature-dependent partition or not, we have decided to make adjustments to the text. In lines 541-548:

“A follow-up study will investigate the SLF by isotherm for mixed-phase clouds of several CMIP6 models. Figure F1 shows that some of the CMIP6 models have a more variable SLF in time and space (e.g., AWI-ESM-1-1-LR, MPI-ESM1-2-LR), which might be an indication that they could have more sophisticated microphysics and could, therefore, perform better in terms of sLCC and surface snowfall occurrence. However, it is not immediately clear which models use a purely temperature-dependent cloud phase partitioning. From the interquartile range (IQR, Fig. F2) CNRM-CM6-1, CNRM-ESM2-1, and IPSL-CM6A-LR are assumed to have a simple temperature-dependent cloud phase due to the smaller IQR of the temperature at various SLF. Interesting to note, these models with the simpler temperature-dependent schemes are the most poorly performing models for fsnow (Fig. 8).”

Answers regarding Typos/Grammatical Issues/Word Choice/Needs Citation:

(21) We substituted the phrase “among others” with “partly” in Line 29.

(22) There are several studies that have also focused on the global underestimation of LWP and overestimation of IWP. But the studies we primarily focus on are related to the high-latitude regions. Therefore we rephrased it to:

“ESMs have previously been shown to not accurately represent cloud phase by often underestimating liquid and overestimating ice, compared to satellite measurements. This has particularly been the case for high-latitude regions (Komurcu et al., 2014; Cesana et al., 2015; Tan and Storelvmo, 2016; Kay et al., 2016; McIlhattan et al., 2017; Bruno et al., 2021; Shaw et al., 2022).” on Lines 72-74.

(23) We have replaced “must” with “could” to soften the language (Line 80) and added an additional sentence in lines 81-82, following for lines 78-82:

“Because cloud phase and snowfall are tightly linked through the processes outlined earlier, any inaccuracies in representing cloud phase could lead to biases in the simulation of snow growth, formation, and the precipitation reaching the ground in solid or liquid form (Mülmenstädt et al., 2021; Stanford et al., 2023). It is important to note, however, that while such biases in cloud phase representation might exist, other compensating model biases could nevertheless lead to accurate precipitation simulations.”

(24) Thanks for catching this. It has now been clarified on Lines 87-88:

“The CloudSat-CALIPSO constellation overpasses the same regions approximately every 16 days, providing long-term periodic monitoring of cloud and snowfall characteristics across the globe.”

(25) We rephrased the sentence using softer language to make the statement more general. Lines 97-99 now read:

“Previous studies suggest that there is a notable difference in snowfall estimates between ESMs and satellite observations. For example, (Heymsfield et al., 2020) found that the Met Office Unified Model and the Community Atmosphere Model 6 produce double the amount of snowfall relative to satellite observations.“

(26) Thank you for catching this.The citation for Edel et al., 2020 is now added to the statement. (Lines 127-129)

"In the Arctic, the ERA-Interim data qualitatively represented the interannual snowfall rates and seasonal cycle well but underestimated high snowfall rates significantly during summer and overestimated weak snowfall rates over open water compared to CloudSat (Edel et al., 2020)."

(27) Thank you for pointing this out. As suggested, we have now rephrased the sentence on Lines 162-163 to:

“Our primary emphasis is on sLCCs in the mid-to-high latitudes and understanding how frequently and where they produce snowfall.“  

(28) We agree that this nomenclature is confusing so have changed “LCC” to “sLCC” in this case (Lines 216-217), thanks. 

"Nevertheless, a non-zero model daily mean LWP means there was an sLCC in the grid box at some point during the day."

(29) Thanks, the column headings have been added in Figure 9 as shown below:

Figure 9. NH mid-to-high latitude monthly averages of fsLCC for July (a-e) and November (f-j) 2007 to 2010. The columns show the CloudSat-CALIPSO (CC) monthly means (a, f), ERA5 monthly means based on daily average values (b, g), and ERA5 monthly means based on hourly values (c, h). The fourth and fifth columns are the difference plots between CloudSat-CALIPSO and the ERA5 daily averages (d, i) and ERA5 hourly values (e, j), respectively. In the last two columns, green (pink) values indicate underestimation (overestimation) with respect to the reference used. Area-weighted averages for the study area (45◦N − 82◦N) are located in the lower-left corner of each map and exclude the dotted area (in b, c, g, h).

(30) Good catch, we have now removed “depending on the region” on Lines 446-447.

"The area-weighted difference between CloudSat-CALIPSO and ERA5 hourly values is negligible in July (1%) and ∼ 15% in the inner Arctic and reduces from 12% overestimation to 7% in November."