This paper begins by reviewing the literature on a range of aerosol perturbation events (‘opportunistic experiments’) in which aerosol-cloud interactions can be observationally explored. The paper also provides a table of references for the ‘opportunistic experiments’, with the observation/modelling type, cloud regime and location of each, along with a summary table of the available observational databases for these experiments. The review then summarises the satellite observations, field campaigns and modelling studies used to target these experiments. The review synthesises figures summarising the change in cloud properties across the range of experiments included. The review finishes by noting the challenges in establishing causality in these observed relationships, including factors that control the cloud response, and the representativeness of these opportunistic experiments.

Overall, the review paper does a good job of summarising the research within this area, and is a useful contribution to the field. I think the review is suited to publication in ACP following some modifications.

Specific comments:

I found the first half of the introduction quite difficult reading, and I think this needs some restructuring or additions for improvement. I would like to see a few more sentences clearly introducing the topic of aerosol-cloud interactions, their importance and why/what observational constraint is required so that the review begins more accessible to readers from a wider background and can be put into context. I think the following suggestions to the current text may also help improve the flow of the introduction.

P2, L20: The sentence starting with “A reduction in precipitation…” is  long, and it could be made clearer which adjustments increase cloudiness and reduce cloudiness.

P2, L15: The first and second paragraph up to “A complication is that…” could be combined. Then, a new paragraphed started from “A complication is that…” by something like “A difficulty in understand these aerosol-cloud interactions is that…”

P3, L30: Change to “However, when emissions perturb aerosols…”

P3, L35: Change to “In this review, we will use the term ‘opportunistic’” rather than “So…”

P3, L35: Why define ‘opportunistic’ and then use ‘natural’ for both natural and anthropogenic perturbations instead of solely using ‘opportunistic’?

P3, L40: The last sentence of this paragraph explains why warm liquid clouds are the focus of the review, but currently the first sentence of this introduction mentions warm liquid clouds. Could that be introduced at the same point?

The strengths and limitations of each ‘laboratory’ could be more clearly defined at the end of some of the subsections in Section 2. For example, for the longer subsections a summary at the end of the section drawing on what has been discussed would be useful. Particular events (Section 2.8) is missing the strength and limitations.  

In Section 4, I think it could be useful to first briefly describe what types of observations can be obtained from satellites and flight campaigns respectively, before going into the methodology and summarising the literature.

In Section 5.7.1, Perturbation Concentration, the second paragraph discusses the over-representation of ship track studies in the literature. As noted, this is an over representation and I don’t think it relates to perturbation concentrations. Could this over representation to be presented elsewhere or in its own sub-section instead?

Section 5.7.3 describes that models can be used to extrapolate from specific situations to climate more broadly. I think more focus is needed on how representative opportunistic experiments are to larger scale climate when applying an observed constraint to a different location/time/scenario etc.?

The title of this review paper is “Opportunistic Experiments to Constrain Aerosol Effective Radiative Forcing”, yet not much is mentioned about how the opportunistic experiments follow through to constraint on aerosol ERF. For the title to be appropriate more needs to be discussed on this, at least in the summary.

P29: The summary section mentions RFaci in a few places, why is this not ERFaci?

P29, L915. The paragraph about the potential changes of DSD comes in a bit out of the blue in the summary section. I think it would help here to have a clearer section in the summary that suggests such future directions.

Figure 3a: Figure caption says the wind is bowling from the North, but plumes look to be travelling South-West not just South?

Technical corrections

P3, L35: Change to (e.g. volcanoes) (e.g. industrial plumes, ship tracks)

P4, L70: “Although modelling studies…” this sentence is a bit contradicting, and needs better defining

P14, L90: Define what “cloud reductions around circulations” means

P9, L250: Should it be “Ship tracks and volcanic plume both show variable cloud adjustments depending on…” rather than “that”?

P12, L345: Include reference in sentence “This is true of the LWP trend from UKESM1…”

P4, L420: It’s not clear that the sentence “Systematic changes in anthropogenic emissions…” relates to the Spring Festival  

P16, L475: Include reference for “see AeroCOM ACI experiment”, or does this relate to AEROCOM – VolcACI in table S2?

P16, L480: For the sentence “Shipping perturbation results are from…” all references include the type of perturbation apart from Peters et al. 2013. Include perturbation type here too

P17, L495: Include references in this paragraph

P17, L510: Delete ‘because’ from the sentence starting with “As the cross-section for scattering…”

P18, L550: Define the “SECA” acronym again here, it was a long way up the paper that this was mentioned

P18, L555-560: Is there a reference for the ACRUISE flights and the future plans?

P20, L625: Specify if “these two approaches” refers to different scale modelling, or to modelling and observations

P24, L730: Should “Like industry” be “Like volcanoes”? Diversity in industry emission rates have not been mentioned in that paragraph

P24, L750: Is the latter part of the first sentence in the dilution paragraph missing? “From local-scales to…”? Otherwise tens to hundreds of kms is a bit confusing.

P27, Section 5.7.2 Timescales: The point about timescales is made within the first paragraph of this subsection, I think the second paragraph could therefore be more concise

Double check references - some have strange characters

Figure 10: I don't think the caption is clear enough on whether the numbers in parenthesis refer to the number of studies averaged over, it could easily be mistaken for reference number or relating to a table. I’m also not clear on why some studies are referenced in the caption and others are not.

SI, Text S1, L5: Figures S2 should be S3

SI, Figure S4 Caption: The number of studies are not shown in the parenthesis as indicated by the caption. Same comment regarding references also applies to here as Figure 10 in main text.

SI, Table S3: There’s not any numbers in parenthesis in the laboratory in this table as the caption suggests.

This paper aims at reviewing the advantages and limitations of various kinds of “opportunistic experiments” that nature and/or mankind offer to those studying aerosol radiative forcing of climate change. The paper is focused on aerosol-cloud interactions and their rapid adjustments. The authors highlight the important scientific findings that the “opportunistic experiments” have allowed, but caution that scaling those findings to the global scale remains difficult.

The paper is well written and its aim of providing a “common footing” in which to understand and interpret the opportunistic experiments is laudable. However, the paper could fulfil that aim more sharply. For that reason, I recommend revisions to address the weaknesses listed below. This could involve a sizeable restructuring of the paper, so major revisions may be needed.

Main comments:

• The paper does not give a tight definition of what an “opportunistic experiment” is. There is an attempt at lines 34-35, but it is unclear whether that definition covers all the types of experiments that are discussed in the paper. Some types do not allow to “know the unperturbed state”, for example – more on that below. In addition, that definition is immediately thrown away “for convenience” in lines 38-39. It should be possible to use a consistent vocabulary throughout.
• The paper presents two main types of opportunistic experiments. The first type is essentially based on in-plume/out-of-plume comparisons (sections 2.1, 2.2, and to some extent 2.3 and 2.4). The second type covers events that are either much larger or comparing situations that are very distant in time (sections 2.5, 2.6, 2.7, 2.8). But the two types sit uneasily together. There is a clear tension between ability to determine the unperturbed state (that would favour the first type) and relevance to climate scales (that would favour the second type). Figure 10, the only figure that attempts to mix the two types of results, illustrates the tension well. But the paper never attempts to resolve that tension. Tellingly, the very good section 5 is mostly based on the first type of experiments – it is the ability to determine the unperturbed state that has brought the best insights. Are some experiments too large in scope? Is there a sweet spot somewhere? The paper could offer guidance there.
• Section 4 on Methods is the weak point of the paper and compares poorly to the very good sections 2 and 5. I think that is because it lacks a clear focus and too often reads like an advert for upcoming campaigns/instruments/models. I am not sure that the section is needed. Pointing to key datasets is the role of section 3. Listing key insights brought about by specific campaigns or models should be done within the context of sections 2 and 5.

Other comments:

• Lines 11-12: I would move that sentence before the previous one, to have the statement on instantaneous forcing preceding that on rapid adjustment.
• Line 83: I would replace “as evident in” to “according to”, because EDGAR is not an observational dataset.
• Line 122: Missing word between evolution and several.
• Line 145: I do not understand the “therefore”. Is a large fraction of the ship tracks simulated by Peter et al. undetectable?
• Lines 161-180: This paragraph lacks a clear conclusion. The difficulty of model-data comparison is one thing, but how does that relate to the use of corridors as opportunistic experiments?
• Lines 238-239: That statement does not add much and could be deleted.
• Lines 253-257: Are those two sentences in the right place? The section would end better without them. (And the sentence on lines 254-255 is grammatically incorrect.)
• Lines 266-267: That statement seems to miss the point that ships increase aerosol loading over a normally low baseline.
• Lines 273-274: Could refer to section g3 of the Global Climate section of the BAMS State of the Climate 2020 report https://doi.org/10.1175/BAMS-D-21-0098.1 to support that statement.
• Line 312: It would help readers to clarify the link between economic restructuring and cloud reflectance.
• Line 312: “co-incident upward trend” combined with the previous sentence suggests that changes in cloud reflectance caused the trend in surface radiation. That is not so sure. Trends in surface radiation are much more robust in cloud-free sky than in all sky conditions.
• Line 316: “CMIP6 emissions database” needs a reference, especially as different versions of CMIP6 emissions have very different histories for sulfur dioxide.
• Lines 381-383: Would need references that attempted such approaches to support the claim that they hold promise. They sound very hard to do in practice.
• Lines 439-440: What caused that shallowing? Anomalous meteorology?
• Lines 447-449: Is that result linked to previous discussions that variability in cloudiness can “hide” substantial radiative forcings?
• Line 458: It would be useful to link contrail reductions to the subject matter of the paper, aerosol ERF. Perhaps a short reminder of the mechanisms of contrail formation?
• Lines 471-475: Aren’t there ship track databases for the Southeast Atlantic stratocumulus deck? There are a few studies that focused on that region and some of that data seems to be used in Figure 10.
• Line 522: “can be performed” – reference needed there.
• Line 576-577: What are the implications of these findings for aerosol ERF?

Technical comments:

• Page 3, line 48: human -> human-caused  
• Line 512: Delete “because”
• Lines 586 and 656: Character encoding problems.