This study proposes a methodology to infer precipitation-forming processes based on satellite retrievals of the vertical profiles of convective clouds' particle effective radius and phase. The satellite is the METEOSAT Second Generation (MSG) over the Indian Ocean. The target clouds are over the UAE. The objective is to determine the clouds’ seedability based on conceptual models.

The general idea and objectives are of great interest, but major methodological issues require a major revision before the paper can get accepted.

Major comments:

The satellite retrieved cloud drop effective radius product (RE) has three types: water, ice, and total cloud. However, these retrievals, as presented here, raise many questions:

How can ice retrievals of RE exist for clouds with temperature > 0C (e.g., Fig. 7b)?

Why are the total and water cloud RE values different at T>0C, where ice cannot exist?

How can water retrievals of RE exist for clouds with temperature < -40C (e.g., Fig. 10d), where water cannot exist?

Why are the total and ice clouds RE different at T <-40C, where water cannot exist?

The methodology depends critically on the definitions of the microphysical zones, as defined in Figure 8. The cloud phase is used for the zones' definitions, but it is not explicitly given by the OCA EUMETSAT cloud data, as all three types (water, total, and ice) have the effective radius (RE) at all temperatures.

Zone 1, diffusional droplet growth is defined by water phase cloud at T>0C and RE < 10 um. However, clouds can have diffusional growth at temperatures well below 0C, and at RE >10 um, until significant coalescence starts increasing RE beyond the drop growth rate by condensation only. This happens at RE>12 um, or even higher.  A case in point is clouds with cold bases, as typical to the UAE.  The lower parts of these clouds must be in the diffusional growth zone, as they are composed of very small water droplets.

Zone 2, droplet coalescence growth zone. The description in Figure 8 has details that are not described in the text (lines 494-500) and are not understandable. What is T21? Why is the 75^th percentile taken and not the 50^th percentile?  Anyway, the 25^th percentile should be the relevant number because this pertains to the smaller ER in the growing convective elements. In Figure 8 there is the condition of 20>ER>15 um.  There seems to be an uncovered gap between Zones 1 and 2.

Zone 3, Supercooled water zone. Why should the supercooled water zone be independent of the diffusional growth or coalescence zones? Both can occur at temperatures < 0 and contain supercooled water. It just does not make sense in both logical and physical ways. There is a potential overlap in the conditions for the zones because the same scene can have conditions that fulfill both Zone 2 and 3. This should not happen.  The extension of mixed-phase to temperatures colder than -35 C in all presented cases with identification of that zone is unreasonable physically. Such cold glaciation temperatures were documented by aircraft only in storms with severe updrafts that caused large hail. Furthermore, VIIRS views of similar clouds in the same region generally resulted in a glaciation temperature of -20 C, due to the dusty conditions. This determination is based on the unique channel combination of VIIRS that allows us to detect unambiguously the glaciated clouds.

Zone 4, Mixed phase zone.

The mixed phase is defined only for clouds colder than -10C, but it can occur at higher temperatures. According to the text, the total cloud ER is used here, but this is not mentioned in Figure 8. Again, since different kinds of ER are used for zones 2 and 3, there can be much overlap in conditions not recognized here, causing definition ambiguity.

Zone 5, Glaciated cloud. Its definition relies on the ice ER. But ice ER seems to be unreliable in many ways. It exists for warm clouds with T>0C. In Figure 10g it has the same large ice ER>25 um at all temperatures. Its ER in some cases (e.g., Fig 10c) is extremely small, much smaller than is almost ever detectable at -45 C by MODIS.

The criteria selection is based on and validated by comparing aircraft in situ cloud measurements. However, the aircraft did not penetrate clouds colder than -13C, so there is no aircraft validation to the zones below that temperature. But the authors claim, and rightly so, that seeding potential is best at clouds with supercooled water that extend deepest, which pertains mostly to the unvalidated temperature range.

Therefore, I recommend that the authors calibrate and validate the MSG retrievals against VIIRS and then redo all their calculations and revise their inferences as necessary.

Minor comments:

Line 87: Growth of precipitation particles cannot occur “through collision and coalescence of the ice multiplication process”.

Line 88: Raindrops cannot form by diffusional growth alone in convective clouds with any cloud base temperature.

Line 371: Please be aware and discuss the gap between the satellite-retrieved rain threshold of 15 um (e.g., Lensky and Shiff, 2007) and the aircraft-retrieved threshold of 12 um (Freud and Rosenfeld, 2012).

This paper introduces a framework that categorizes convective cloud features from UAE into 5 different microphysical zones using satellite data. The effectiveness of this framework is then evaluated using aircraft observations from UAE 2019 Airborne Campaign. Overall, this paper is easy to follow. However, while the framework is somewhat interesting, there are several flaws in the methodology that requires some re-work.

Major comments:

1. If I am understanding correctly, this paper targets the microphysics of continental convective clouds. However, it is surprising that for deriving the in situ measured cloud droplet size distribution and cloud effective radius (Reff), this paper is only using the FCDP. The author claims that this choice is based on the fact that the LWC derived from FCDP has higher correlation with the Nevzorov probe measured LWC. However, if you are studying the different microphysical processes (diffusional growth vs. collision coalescence), are having both liquid and ice phased clouds in your sampling, it is almost common practice to combine the FCDP/FFSSP with 2DS (or even HVPS), (see Rosenfeld and Lensky 1998, Painemal and Zuidema 2011, Kang et al. 2021). By using FCDP only, the in situ Reff will be biased towards the smaller droplets.

Rosenfeld, D., and I. M. Lensky, 1998: Satellite-Based Insights into Precipitation Formation Processes in Continental and Maritime Convective Clouds. Bull. Amer. Meteor. Soc., 79, 2457–2476, https://doi.org/10.1175/1520-0477(1998)079<2457:SBIIPF>2.0.CO;2.

Painemal, D., and P. Zuidema (2011), Assessment of MODIS cloud effective radius and optical thickness retrievals over the Southeast Pacific with VOCALS-REx in situ measurements, J. Geophys. Res., 116, D24206, doi:10.1029/2011JD016155.

Kang, L., Marchand, R. T., & Smith, W. L. (2021). Evaluation of MODIS and Himawari-8 low clouds retrievals over the Southern Ocean with in situ measurements from the SOCRATES campaign. Earth and Space Science, 8, e2020EA001397. https://doi.org/10.1029/2020EA001397

2. The authors use the in situ probe measured Reff to characterize the cloud microphysics, but then use the satellite measured Reff to build the framework to identify the different microphysical zones of convective clouds. This works under the assumption that the satellite measurements are representative of the in situ measurements. However, the authors need to keep in mind that passive satellite retrieved Reff is a cloud top effective radius (which is not vertically resolved in any way), and the in situ Reff could be sampled from anywhere within a cloud (depending on where the cloud pass/penetration took place). These are conceptually different, and the authors need to be careful when using the different Reffs to characterize cloud microphysical processes.

3. I don’t think the authors elaborated on how the collocation between the aircraft measurements and the satellite observations were done.

4. The thresholds used in the framework is somewhat confusing. What is the rational of choosing the 50th percentile Reff or the 25th percentile Reff in the different zones? The mixed-phase zone also seems to overlap quite a bit with the collision-coalescence zone and the supercooled water zone. Is it really necessary to have 5 different microphysical zones?

Minor comments:

Line 28: precipitation-producing clouds -> precipitating clouds

Line 208: Any possible reason why the LWC from FFSSP has a much lower correlation with the LWC from Nevzorov?

Line 210: It is surprising that you are using FCDP alone to derive effective radius for convective clouds…

Line 220: How is this “mean effective radius” defined? Is the range indicating the range of “mean effective radius for each CP”?

Line 371: If I am understanding this correctly, this 15 μm threshold was derived from AVHRR measurements, is this applicable to the in situ measured Reff?

Line 377: (d) total water content from Nevzorov.

Line 450: why are there ice cloud effective radius retrievals at T~ 10°C from in situ probes?

Line 464: why choosing to use FCDP to derive cloud effective radius when you are targeted at convective clouds?

Line 468 to 470: I cannot agree with the statement that “the ERs from aircraft and satellite datasets have a fair agreement”, your Figure 7 is suggesting otherwise…which really questions the validity of using ERs from satellite data to build the different cloud zones.

Line 497: Why in some of the zones the ER thresholds are using the 75th percentile, and in others the 50th percentile is used?

The criteria used in the 5 zone framework is somewhat ambiguous. If I am understanding this correctly, the authors are trying to formulate the 5 zone framework using thresholds of ER, dER/dt, BT, wouldn’t this cause some overlap between different zones? I would suggest sticking with the same percentile (whether 25th, 50th, or 75th) and be consistent.

Line 540: Figure 9 is suggesting that water cloud Reff is used in zone 1 and zone 2 and zone 3, total cloud Reff is used in zone 4, ice cloud Reff is used in zone 5. This was not mentioned in the main text.

This study combines in situ aircraft data collected in summertime convective clouds over the UAE in 2019 with satellite data, ground radar, and reanalysis to adapt the development of a 5 microphysical zone categorization to said region and season.  The 5 zones are: (1) diffusional droplet growth zone, (2) droplet coalescence growth zone, (3) rainout zone, (4) mixed-phase zone, and (5) glaciated zone.  A case study is then presented for evaluation.  The study presents a novel adaptation of the proposed categorization that presents a useful framework for future analyses.  Said framework is specific to the season and environment of the aircraft data, however, so further efforts should be made to better categorize that environment to improve potential applicability to other seasons and regions.  With revisions, the manuscript could be suitable for publication.

Specific Comments:

1.  The cloud droplet effective radius looks to be determined solely from the FCDP, when it should be determined as a composite from multiple cloud probes, not just the FCDP. Larger particles seen by the 2DS (and HVPS) should also be included, for both the liquid and the ice clouds.

Discussion following from line 465 suggests that the FCDP is not sensitive to ice particles, when in fact it does measure ice crystals, albeit with more uncertainty in sizing than of water droplets.  However, were the 2DS and HVPS measurements in the ice clouds included in the calculation of effective radius, one might anticipate improved agreement between the in situ and satellite estimates.  The current statement that they disagree because the FCDP is insensitive to ice particles is insufficient. 

2.  The analysis of the in situ data is lacking in qualifying the environmental conditions and the cloud evolutionary stage of the selected cloud passes. For example:

Lines 591-594: Do we know that they are not all young turrets being sampled in SF07?  Is there any indication of cloud age or development stage of these case studies?

Lines 597-599: Do we know if the cloud passes at these levels were in a similar age of cloud life cycle?  Were they at similar distance from cloud top and cloud base?

Line 608: “..implies that the droplet growth in the cloud cases SF01 and SF07 is suppressed”, are the cloud passes under consideration consistent enough to make this conclusion? What were the cloud base temperatures?  What are the cloud top temperatures/heights?  What are the environmental conditions for the various days?

Figure 12 and associated case studies: were the temporal measurements of these cloud cases all from the same convective turret, or could they have been different turrets (differing potentially in cloud top height, cloud age, etc.)?

Line 603 states the coldest observed temperature was -12 C in all four flights.  Is this the coldest temperature because it was near cloud top, or was there another sampling reason?  How close was sampling performed to cloud top?

All of the utilized cloud passes should be better qualified to improve usefulness and applicability of the analysis, and perhaps the analysis revised to include only cloud passes of comparable nature (it is currently unclear if they are comparable or not from the lack of context for the chosen cloud passes).

Minor Comments:

Figure 10 and 11 are cramped and very difficult to read

Figure 12. The font of the x-axis needs to be bigger to be able to read the times. 

Figure 12.  Black text on purple background is nearly illegible.

Figure 14. For clarity, would suggest numbering the zones in the figure.

Lines 692-695. For clarity of discussion, would suggest using zone microphysical names rather than numbers here.