Review of “Observed modulation of wintertime Western Arctic mixed-phase cloud properties by sea ice conditions, their long-term variabilities and trends” by Pablo Saavedra Garfias and Heike Kalesse-Los.

In this study, the authors provide observational evidence that Arctic cloud properties are strongly related to sea ice concentration (for coupled cases). The analysis also present the long-term trends of Arctic cloud properties, and reveals that the evolution of cloud properties presents cyclical characteristics. The results are interesting and the paper is generally well written, my specific and minor comments are listed below for the authors to consider.

Specific comments:

1. More physical explanations would be helpful to better understand the relationships between the cloud properties and SIC. For example, for coupled cases, LWP and IWP both increase as SIC decreases, but the cloud depth decreases as SIC decreases. That means a shallower cloud may have a higher LWP and IWP, which is hard to understand. Is this finding statistically robust, and is there an explanation? In addition, the IWP increases as SIC decreases, and the cloud top temperature also increases as SIC decreases, which suggests the IWP is higher in a warmer and shallower cloud. This is also difficult to understand and inconsistent with previous studies. Some explanation is necessary.
2. SIC is well correlated to temperature according to Fig. 1, while all the cloud properties are plotted as a function of SIC in Fig. 5. I’m wondering if all the cloud properties are really directly controlled by SIC, or are they actually controlled by temperature, especially for the cloud top temperature and lapse rate.

Minor comments:

1. A radius of 50 km and a conical sector of ±3 degrees is used for the upstream area. How are the results sensitive to the choice of radius and conical sector?
2. Figure 2c and d. The observed LWP is high at 2-4 UTC, but the cloud is mostly classified as ice.
3. Line 237. The argument of “plateau where the property levels off” is probably based on the fitted model. For many properties, there is no “plateau” according to the actual observational data.
4. Figure 5c and 4g. It is obvious that the model used to fit the data is not the best one for the decoupled cases in these two figures.
5. Line 290. If there is an inversion in the cloud layer, it may be inappropriate to calculate the lapse rate using the equation in Line 283. Have you tried to calculate the lapse rate using the temperatures at any two adjacent altitudes in the cloud, and then average over the entire cloud layer?
6. Another important macro-property is the cloud occurrence frequency. Is it affected by SIC?
7. Line 463. I think the analysis of long-term ice properties should be mentioned in the text in Section 3.2.3.

In this manuscript, the authors analyze observations from the ARM North Slope of Alaska site and try to relate the local cloud observations to the sea ice concentration (SIC) upstream. While the research question could be of interest, the study is not performed in a rigorous way, at least not based on the information provided, and therefore, the results can not be trusted. Even if the analysis is correct, the way they are presented only provides an incremental step in the understanding of the Arctic system. The authors need to collaborate with experts on atmospheric processes in the boundary layer, clouds and the large-scale circulation to formulate relevant questions and design of the analysis.

The manuscript is full of uncertainties of what is actually done in the analyses, here, I only mention a few. First, Table 1 lists all datasets and the time resolution varies from 1 s to 1 day. In the following analysis, there is no information on what time resolution of the data is used. It seems like the data points in Figure 3 are monthly although the caption says weekly averages. Turning to Figure 2, what is the resolution of the data points that are behind this analysis? It is also very strange to have a so non-linear SIC division, does the distribution really support that? How can the confidence interval be so small when the bars show 50% of the data points? In this figure, I see no motivation for the chosen function – it could just as well be fitted with a straight line. To me it suggests that you have decided that the cloud parameters should be a function of the SIC and then you do your best to divide the dataset in such a way that you think that is what you see. The relationship is much more complex than that and the seasonality in the airmasses entering the Arctic over the season have a strong cycle, thus the temperature has a strong cycle and so has the LWP and IWP along with the SIC.

In conclusion, the study is not well motivated, nor well performed and the results, if they are correct, are not of interest to the community. Statements towards the end of the conclusion section illustrates my point. ”This study put into consideration empirical mathematical models derived from long-term observations as parameterization for the modelling of atmospheric boundary layer MPC ascribed to Arctic sea ice conditions. We suggest that the sea ice-atmospheric feedback processes that are represented in Arctic weather and climate models certainly should aim to mirror the complexity of relationships that the present study has uncovered. This implies the consideration of not only small-scale horizontal moist air mass transport that serves as physical link between sea ice openings and clouds downstream, but also weather patterns, atmospheric circulation processes, intra- and multi-decadal climate oscillations, and teleconnections between the Arctic and lower latitudes”