We would like to thank Dr. Bley for his analysis and useful comments. The responses to major and minor comments are given below. We marked the reviewer’s and the author’s comments by “RC:” and “AC:”, respectively.

Specific comments

RC: The manuscript is structured in ‘Definitions’, ‘Short -term cloud dataset and ‘Long-term cloud dataset’. The ‘simulated lidar profiles over cirrus and stratocumulus clouds’ part is an important input dataset for the full analysis, but it is only a subsection of ‘Short -term cloud dataset’. I would therefore suggest to add an additional chapter between Definitions and Short-term dataset called something like ‘Simulated lidar profiles’, because it is also part of the long-term dataset.

AC: Even though the two parts share some modules, the long-term dataset was not built using CLOUD3D model, so it would not be fair to mix them up. The time of emergence was estimated using climate predictions from IPSL-CM6 and CESM2 simulations, the outputs of which were fed to the COSP simulator. The monthly grids of Copaque and Zopaque, generated by this simulator, are then spatially averaged over the tropics. In this analysis, we assume that the ATLID and CALIOP datasets can be merged to build a long-term dataset, so we consider just the length of this dataset, and not the characteristics of individual contributors.

RC: I would like to see some more discussion for the case that EarthCARE starts operation later than CALIOP stops. Could be a work around to use some typical cloud scenes characterized by CALIOP and later with EarthCARE to find the same cloud regimes to tune the long-term cloud dataset without intercalibration between the instruments operating at the same time?

AC: Normally, one cannot rely on the clouds themselves because they are highly variable. Instead, and it’s a common practice, one starts the lidar calibration with the stratospheric signals in the aerosol-free area, where the molecular backscatter is known. The second potential calibration source is very strong backscatter from ice crystals of from the surface. When both channels of HSRL lidar are calibrated, the rest of the processing chain should give the results equivalent to that of the reference lidar. As for the intercalibration procedure suggested by the reviewer, if the gap between two satellites is within a year or two, then one can use the average cloud amount for low, middle, and high clouds in different zones (tropics, mid-latitudes, and polar) to track the changes and to introduce a feedback to cloud detection algorithm. This way, the number of cases measured for each zone will be high and the uncertainty will be low whereas it is unlikely that the global cloud amount will change within such a short period. The daytime and nighttime observations should be considered separately to address the diurnal cycle and daytime noise issues. We have added this discussion to the end of Section 4.1.

RC: It is stated that a long-term cloud record can be produced when using a kind of less sensitive cloud detection threshold (based on SR and the attenuated total backscatter) which improves the agreement between CALIOP and ATLID. —> But in that case, you are missing some thin clouds which ATLID would be capable to detect. Could ATLID help quantifying how CALIOP underestimated the global cloud coverage in past datasets?

AC: At the moment, there is no ATLID data available and we have yet to confirm that the actual performance of this lidar corresponds to our estimates. If we assume that the laser will operate at full power, the alignment will be perfect, the laser induced contamination will be reduced to a minimum, and the detector will not suffer from high-energy cosmic particles, then we hope to get the cloud detection performance estimated in this manuscript. In this case, one can run the cloud detection with two SR’(532nm) thresholds, one defined by Eq.6 and the other one defined by Eq.36 in the updated version of the manuscript. The difference in thin cloud amount obtained this way could be projected back to CALIOP data observation period.

RC: Climate models have large uncertainties as shown in Perpina (2021) —> therefore a long space borne lidar record is essential to better quantify trends and understand the inter model differences.  If ATLID cannot fill the long-term gap after CALIOP because it is likely not going to operate as long as CALIOP. How could upcoming satellite missions Aeolus-2 or AOS) help overcoming this long-term challenge after ATLID? You are mentioning this aspect in L 566, but could go in some more detail.

AC: In principle, the approach we have been developing (see Feofilov et al., 2022 and this work) is universal in a sense that it can be applied to any other active optical sounder. For Aeolus-2 with its molecular and particulate channels, one can apply the same treatment as we did for Aeolus-1. If the Aeolus-2, like Aeolus-1 will be designed without depolarization channel, we will have to apply the methodology currently being developed (see the presentation of Feofilov et al., 2023). If the local solar time of new instrument will be different from that of CALIOP, the diurnal cycle correction will have to be applied (Feofilov et al., 2023). All these corrections have their own uncertainties and biases, so the less the difference in the initial design, the better for the continuity of cloud record. We added a sentence on follow-up lidars to the end of this paragraph and to the end of next-to-last paragraph of this section.

Technical corrections

RC: L 1-18: You should try shorten the introduction part of the abstract. The whole abstract is way too long. Parts of the motivation and introduction can be addressed in detail in the Introduction chapter.

AC: We have shrunk the first paragraph of the abstract to ~50% of its initial size.

RC: L 31: ATLID-ST: Please define. Or do you mean CLIMP-ST?

AC: This is true, we should have named it CLIMP-ST here despite the fact that the dataset comes from ATLID. Fixed, thanks.

RC: L 56: clouds properties —> Shouldn’t it be ‘cloud properties?

AC: We changed it to “clouds’ properties”, thanks for noticing this typo.

RC: L 106: Rephrase to: “Avoid overestimation of the cloud fraction… “

AC: Fixed, thanks.

RC: L 110: Averaging le lidar signal. Should be “averaging the lidar signal“

AC: Fixed, thanks.

RC: L 121: optically thinner “cloud”. Through the text, you always write ‘cloud’, but it should be clouds.

AC: We fixed this typo and we checked for other instances when the cloud should be in plural form, thanks.

RC: L 129: Chapter 2: Definitions (rather Methods? See Specific comment above)

AC: It’s true that the chapter has developed beyond pure definitions, but it is not the full description of the methods, either. At the moment, we opted to name it in accordance with its contents: “Two spaceborne lidars, lidar equation, and cloud detection”

RC: L 199: Rephrase to “Or if it was sampled”…

AC: Fixed, thanks.

RC: L 274: Rephrase to “tropical part of the orbit”

AC: Fixed, thanks.

RC: L 320: Voluntarily split the -> voluntarily seems not the write phrase here (maybe better artificially??)

AC: We opted to write “arbitrarily” as it was suggested by the 2^nd reviewer, thanks for noticing the odd phrasing.

RC: L 329: We set the cloud mask to 1 whenever IWC>0. Shouldn’t be the instrument sensitivity be taken into account here? Very small IWC values (<0.001) could be model specific, but does not represent what the lidar would see.

AC: This is a good methodological point, and we thought about using the suggested criterion. But, when it comes to applying the instrument sensitivity and reversing it to the lowest detectable IWC values, it adds another layer of complexity. Speaking about the minimum detectable backscatter (MDB), please, see the new paragraph on its estimation at the end of Section 3.4. Theoretically speaking, we could have used the mean IWC values from this analysis, but it has been done for 5km averages, and shorter averages will require an increase in “minimal detectable IWC”. We ended up with comparing both simulated measurements to reference cloud, knowing that they will never be able to detect the thinnest cloud from a single measurement.

RC: L 337: Better rephrase to “Fig. 4 and 5. demonstrate...”

AC: we rephrased to “In Fig. 6 and 7, we demonstrate… ”, thanks for noticing this.

RC: Fig. 4: Please improve the labelling. What is CALIOP and what is ATLID becomes not really clear here. (a), (b), (c) and (d) are explained doubled, (e)-(h) are missing. I would suggest to contrast the two instruments, always CALIOP left and ATLID right would make the differences more visible.

AC: It’s true, the labeling was all messed up here, sorry. We fixed this and we labeled the groups of panels by “CALIOP” and “ATLID” in addition to (a)-(h) labeling.

RC: L 381: particulate backscatter ?? Particle backscatter is more common , also in Line 383.

AC:  In the scientific literature, both terms are used, so we will agree with whatever the language editors of the journal will suggest.

RC: L 394: “the detectability “

AC: Fixed, thanks.

RC: L 397: “the daytime noise increases the disagreement with the reference cloud dataset”. This is unclear what is meant. Increases proportional to the disagreement to the reference dataset…?

AC: We have rephrased this part to make it clearer.

RC: L 401: during day and night

AC: Fixed, thanks.

RC: L 415: Table 2: What is the end of the caption? It is unclear where the free text continues.

AC: We added a vertical space to separate the caption from the text.

RC: L 539: Any idea what makes the big difference in the observation requirement (with respect to EarthCARE operation lifetime) between IPSL-CM6 and CESM1?

AC: The answer may be partially found in (Perpina et al., 2021), which claims that “.IPSL-CM6 overestimates the amount of weak ascendance and, in this condition, significantly overestimates the Zopaque (+2 km). This overestimation of Zopaque helps IPSL-CM6 balance other errors to get its predicted CREnet close to observations. For CESM1, it is the Copaque ∼40% and the CRESW that are overestimated”. Not being experts in GCMs, we would refrain from further speculations on this subject.

RC: L 575: What would be the required overlap period between CALIOP and EarthCARE for optimal intercalibration? Could Aeolus help overcoming the problems when CALIOP and EarthCARE are not flying at the same time (due to CALIOP and Aeolus operating synchronously and Aeolus having a HSRL lidar more comparable to ATLID than CALIOP and ATLID?)

AC: One year would be an ideal overlap period, but even one month of stable operations of both lidars would yield enough statistics for intercalibration. As for using Aeolus for intercalibration, there is at least one extra uncertainty source, which is important for continuity, namely, the lack of depolarization channel. In (Feofilov et al., 2023), we discuss the method of compensation for this, but this is not a finished and published study, the approach has to be elaborated and tested yet.

RC: L 650: probably be helpful is too weak in my opinion! “The results in this study using simulations indicate that a merged dataset between CALIOP and ATLID will provide important information… ” would be more sound. In general the final paragraph of the conclusions could be phrased a bit stronger showing the benefit of this study.

AC: We have rephrased the last sentence to make it stronger, but we didn’t want to be overoptimistic given that the lidar is still on the ground and no real measurements to be analyzed are available. Anyway, thanks for this suggestion.

References:

Feofilov. A.G., Chepfer, H., Noel, V., and Szczap, F., Towards establishing a long-term cloud record from space-borne lidar observations, EECLAT 2023 Workshop, Banyuls, France, available online at https://eeclat.ipsl.fr/2023/02/13/eeclat-2023-workshop-23-26-jan-2023-banyuls, 2023

We would like to thank Dr. Vaughan for his thorough analysis and in-depth review of our manuscript. In the new version, we tried to address all the issues identified in his review. Due to the volume, we provide the responses in the attached PDF file.