The manuscript by Spang et al. uses satellite retrievals of cirrus near the tropopause for a region in Eurasia to calculate its radiative effect. Given the uncertainties in particle radius and shape, they also perform several sensitivity experiments. This is the first time that the top-of-atmosphere radiative effect of such optically thin, often unnoticed, cirrus clouds has been evaluated.

It is very valuable to determine the radiative effects of such thin clouds, which normally go unnoticed even by sensitive active sensors such as the CALIOP lidar. Not surprisingly, the study finds a very small net TOA effect compared to other cloud types. Their net effect is, however, strongly dependent on the assumption of particle size and shape.

I have several comments, large and small, that should be addressed before the study is published in final form. In particular, while the abstract reads as if the authors consider the radiative effect of such clouds worldwide, they ultimately perform the analysis on only a small subset of measurements in a region of northern Eurasia, for only one week. Climate variability can be extremely large on weekly timescales in a relatively small region of the extratropics, which is the main reason why I am not confident in the robustness of the generalization of the results in both space and time. Cirrus frequency may be different for other geographic locations and generally has a relatively strong seasonal cycle. Moreover, regions like the tropics are subject to substantially different meteorology, surface temperature, surface albedo... all of which need to be carefully considered (if they haven't already been).

In addition, while using all available CRISTA measurements would only increase the sampling to 2 weeks, I don't see why not use all data from this otherwise relatively limited dataset.

Finally, I miss more discussion of why studying such clouds might be important at all. Given the much larger radiative importance of other cloud types, I don't think their TOA radiative effect is a strong enough motivation. However, even such thin clouds may be important in modulating the radiative budget within the atmosphere and thus modifying the mean tropopause temperature. In this way, they may indirectly control the amount of water vapor entering the stratosphere (at least in the tropics) and influence circulation patterns. This may be their most important climatic role and should not be neglected in this analysis.

Additional general comments

1. Ultra Thin Cirrus (UTC) are (most likely) standard representatives of the in situ formed cirrus, which happen to be too thin to be detected by most sensors. As such, there is no need to consider them as a separate cirrus type in climate models. While climate models in general struggle to correctly represent cirrus clouds, I see no evidence of climate models being particularly bad at simulating UTC. See, for example, examples of such thin clouds for coarse global climate models (e.g., Gasparini et al., 2018), fine-resolution cloud models (e.g., Gasparini et al., 2022, appendix), and global storm models (e.g., Turbeville et al., 2022, Nugent et al., 2022). In fact, models sometimes simulate a large number of such extremely thin cirrus. Moreover, given the inability of most satellite retrievals to detect UTC (and thus validate models), and given its small radiative impact, UTC can generally be neglected in standard climate model applications. 

   Note: there may be other, better references compared to those listed

2. Is the main scientific motivation for studying UTCs really their (very small) net TOA radiative effect? At least for the tropical subset of UTC, they may be more important because of their role in stratospheric water transport (e.g., Luo et al., 2003).
3. What is the value of using aggregates to estimate radiative uncertainties, as it is almost certain that such high, thin cirrus are not composed of aggregates? Wouldn’t it be of more value to consider the more common columns, rosettes, or budding rosettes instead (as per Lawson et al., 2019)?
4. I am not convinced of the value of the mirrored February analysis. Clouds exhibit strong seasonality in mid and high latitudes. Therefore, it is hard to expect the same clouds to occur in winter. Also, in winter, the tropopause is lower, so clouds must also shift to lower altitudes.
   
   Moreover, were the possible differences in surface albedo between summer and winter considered? The selected regions are most likely snow-covered in February. This further diminishes their SW CRE.
   
   I therefore think this should be removed from the result section. Such results may better fit in the discussion section, which may be a good place for such speculative results.
   
   I would instead appreciate it if the authors showed the heating rates (units K day^-1) in the result section. I am mentioning that particularly as it is very straightforward to transform radiative flux vertical profiles in W m^-2 to heating rates. High cloud heating rate was shown to strongly modulate circulation patterns (e.g. Voigt et al., 2021, Gasparini et al., 2023 and references therein).
5. I didn’t understand whether cloud geometrical thickness can be estimated directly from the CRISTA measurements. If so (as the selected cloud in Figure 2 suggests), why did the authors use predefined cloud thickness of 0.5 and 2 km in their analysis?

Specific comments

Page 1, abstract:

Authors could improve the readability of the abstract by avoiding most abbreviations here and defining them later in the text.

Page 1, abstract:

I strongly suggest that the abbreviation "RE" be changed to "CRE" for consistency with current atmospheric and climate science literature. I don't see an urgent need for the additional abbreviation "cirrus radiative effect".

Page 2, line 28:

Foster et al., 2021 may not be the best reference for that statement. If I don’t miss something, it mentions the uncertain cirrus response to global warming only related to their very uncertain anvil area cloud feedback, which is not relevant for this study.

Page 2, line 44:

…they computed cirrus cloud radiative effect…

I think the cited study considered all ice-containing clouds, not only high-altitude clouds; I therefore suggest replacing it with “ice cloud radiative effect” instead of “cirrus radiative effect”.



Page 2, line 48:

CALIOP can probably detect some of the thinnest cirrus considered in this study. There seems to be some overlap at cloud optical depths of ~0.01. I would rewrite the sentence with a weaker statement.

Page 2, lines 50-52:

Beyond Davis et al,, 2010, Balmes and Fu, 2018 show a comparison of 2 ground-based Raman lidars with CALIPSO, showing the limit to CALIPSO detection.

In addition, Matus and L’Ecuyer, 2017 is another good CloudSat-CALIPSO reference for ice cloud radiative effect.

Page 3, section 2.1:

A global map with CRISTA measurements may help in qualitatively understanding the spread of its retrievals



Page 3, line 81:

I don’t understand the meaning of the sentence that begins with “Where along the line of…”

Page 4, Fig. 1 caption:

What are the contour lines representing? Is the region of interest the whole map section or only the three pink rectangles?

Page 4, Fig. 1:

The detected thin cirrus seem to be occurring above mountains, likely connected to orographic mountain waves? Is this true? Are mountains a preferred location of very thin cirrus near/above the tropopause?

Section 2.2:

Do you consider clear-sky conditions below the detected cirrus when calculating radiative fluxes? Please clearly state that and discuss the limitations of such an assumption!

Data preparation section:

1. Is there a lower (altitude) bound to cirrus detection? Is there any lower bound considered to avoid clouds that occur at temperatures, where water can also co-exist with ice (e.g. T>-38°C) ?
2. How are thicker clouds treated? Are only their tops considered, or are they fully discarded from the data?

Page 5, Fig. 2:

The chosen cirrus seems pretty thick in terms of cloud geometry, given that the UTC are expected to be very thin.

Also, for better visibility, I suggest cutting the panel (a) at 200 K, and the panel (b) at +/- 300 (or 350) W m^-2.

Page 6, line 154:

Are the cloud properties assumed to be constant in vertical? Please mention that explicitly and explain the rationale for it.

Page 6, lines 155-156:

The thicker of the considered clouds overlap with a range of cloud optical thicknesses accessible also by CALIPSO data (at least during night time). This overlap gives an opportunity to test whether the mirroring of August clouds to February makes sense, based on the long and climatologically more relevant CALIPSO measurements.

Page 6, lines 157-164:

I don’t understand fully understand why two methods are used to estimate cloud optical depth.

Page 7, Figure 3:

-For clarity, I suggest using values instead of log(IWC) or log(COD)

-This was where I first realized that CRISTA doesn’t allow for an estimation of the vertical extent. Please state that more clearly in the text.

-Why are some cloud layers at temperatures above 240 K? Those are likely several km below the tropopause.



Page 8, lines 184-185:

The mentioned point is one of the main reasons why I think there is only little value in estimating the radiative effects of UTC in February.

The clouds analyzed seem to end up high above the tropopause altitude if I understand correctly. However, even if one were to correct for this bias, I still think that mirroring the results to February is not much better than guessing.

Page 11, Fig. 5:

Why are just results for aggregates shown for February conditions, and not the more relevant spheric shapes? Mention CRE = 0 line.



Page 12, line 231: Why?

Page 12, lines 237-238:

LW CRE is related to the difference between the surface and cloud top temperatures, not necessarily the cloud temperature itself.

Page 13, line 238:

Lower temperatures at the tropopause -> compared to surface temperatures.

This is why high clouds in the tropics (to a first approximation) have larger LW CRE than clouds of equivalent optical depth and cloud top temperatures in the extratropics. [consider surface temperatures of ~28°C in the tropics vs. 10°C in the extratropics].

Page 13, line 261:

Don’t forget the impact of surface albedo! (or the albedo of lower-lying clouds)

Page 14, lines 290-292:

It may make sense to estimate CRE for 3 cloud thicknesses instead of only 2: the lower plausible thickness (as detected by WISE or other observations), the most frequent one (500 m), and the largest one (2 km).

Page 15, lines 318-321:

Cirrus cases studied are most likely not representative of clouds all over the world, but only of a tiny region in Eurasia. As previously mentioned, I don't think one can easily extrapolate a global effect from that sample of clouds.

References:

Balmes and Fu, 2018, 10.3390/atmos9110445

Gasparini et al., 2018, 10.1175/JCLI-D-16-0608.1

Gasparini et al., 2022, 10.1175/jcli-d-21-0211.1

Gasparini et al., 2023, 10.5194/egusphere-2023-1214

Lawson et al., 2019, 10.1029/2018JD030122

Luo et al., 2003, 10.1029/2002GL016737

Matus and L’Ecuyer, 2017, 10.1002/2016JD025951

Nugent et al., 2022, 10.1029/2021EA001965

Turbeville et al., 2022, 10.1029/2021EA001978

Voigt et al., 2021,  10.1002/wcc.694