| 10 Jul 2026
Design of a Cloud Chamber and Quantitative Seeding Experiments for Warm and Cold Clouds
Abstract. Cloud chambers provide controlled thermodynamic environments for investigating cloud microphysical processes and their responses to aerosol perturbations. In this study, we developed a cloud chamber system for controlled observations of cloud microphysics and weather-modification processes. The system comprises two vessels with volumes of 2.7 and 9.0 m³. It provides adjustable temperature and pressure over approximate ranges of −40 to 40 °C and 30 to 1110 hPa, respectively, and is designed to regulate relative humidity across nearly the full 0 to 100 % range. The measurement system covers particles from submicron aerosols to cloud and precipitation particles, with approximate size ranges of 10–700 nm for submicron aerosols and 2–1550 μm for cloud droplets, larger hydrometeors, and ice particles. The chamber performance was evaluated using temperature-control, pressure-control, and humidity-control tests. Representative paired background–seeded experiments were then conducted in chamber B under warm-cloud and cold-cloud conditions. In the warm-cloud experiment, the introduction of 10 g of hygroscopic seeding powder increased the aerosol number concentration and was accompanied by a rapid increase in 3–10 μm droplets. Relative to the background case, the seeded case had a lower concentration of droplets ≥12 μm, with the mean concentration decreasing from 1357 to 478 cm⁻³, and showed shorter cloud persistence.
The relative-humidity decline rate was higher in the seeded case than in the background case, increasing from 0.47 to 0.75 % RH min⁻¹, consistent with enhanced water-vapour uptake after the introduction of hygroscopic particles. In the cold-cloud experiment, 0.5 g of AgI was ignited, and the resulting AgI-containing combustion aerosol increased the submicron particle concentration to above 2 × 10⁴ cm⁻³. The AgI-seeded case showed identifiable ice-crystal signatures together with a rapid decrease in liquid-droplet concentration. The total droplet concentration reached a short-lived peak of approximately 2.1 × 10³ cm⁻³ and decreased to less than one-third of this peak within about 1 min, while droplets larger than 10 μm were strongly reduced within about 2 min. These representative experiments show that the chamber system can support controlled observations of aerosol, droplet, and ice-particle responses associated with warm-cloud and cold-cloud seeding materials.
GENERAL ASSESSMENT
This manuscript describes a two-vessel expansion cloud chamber at CUIT and presents one paired background-seeded experiment each for warm-cloud (hygroscopic powder) and cold-cloud (AgI) conditions. A facility of this kind is a welcome addition, and the engineering description of the thermal and pressure systems is useful. Several technical items from the access-review stage have been corrected (pressure range in the abstract, the heating-rate value, author-contribution initials, table layout), which is appreciated.
However, the scientific core of the manuscript does not currently meet the standard expected of an AMT instrument/facility paper. Of the seven major points raised at the access-review stage, essentially only the background-aerosol discussion received a partial response; the reproducibility and "quantitative" claims, the qualitative ice identification, the AgI temperature question, the humidity-sensor problem, the aerosol size-range inconsistencies, and the implausible 1357 cm^-3 value all persist in the present version. In addition, a closer reading reveals several new and serious problems: the newly added saturation-ratio analysis reports a mean in-cloud water saturation ratio of 1.222, which is physically impossible and contradicts the co-plotted RH sensor; the headline warm-cloud RH-decline result is confounded by the injection of dried carrier gas; the paired cases are not thermodynamically matched; and the observed cooling rates are only 20-30 % of the adiabatic value, contradicting the "adiabatic expansion" framing used to interpret cloud lifetime.
My recommendation is major revision. Substantial new experimental work - replicate runs, blank-injection controls, and quantitative ice detection - is in my view required; if that is not feasible, the manuscript should be reframed as a facility-description paper without seeding-effect conclusions.
MAJOR COMMENTS
1. Reproducibility and the term "quantitative" (carried over from access review, point 1 - not addressed)
Each seeding condition is still supported by a single paired run. No repeat experiments, no run-to-run variability, and no uncertainty estimates are provided anywhere in the manuscript. Yet the title still announces "Quantitative Seeding Experiments", and the conclusions (L367) still claim "a reproducible framework for quantitative seeding experiments". All headline numbers (1357 -> 478 cm^-3; 0.47 -> 0.75 %RH/min; 2.1 x 10^3 cm^-3; Sw = 1.222; etc.) are quoted to 3-4 significant figures with zero uncertainty, from single realizations.
Required: either (a) provide replicate paired experiments with run-to-run spread for at least the two demonstration cases, or (b) remove "Quantitative" from the title, remove "reproducible framework" from the conclusions, and attach uncertainty estimates (instrument + repeatability) to every quoted number. As written, the central claim of the paper is unsupported by its evidence base.
2. The seeding signal is confounded by the delivery method - no blank-injection control (new)
Both seeded cases involve injecting substantial volumes of conditioned carrier gas into the chamber, and the paired "background" cases involve no injection at all. Consequently, every reported background-seeded difference conflates (i) the microphysical effect of the seeding material with (ii) the thermodynamic and dynamical effect of the injection itself.
This is not a hypothetical concern; it is quantitatively decisive for the warm case:
- The powder is delivered with dried compressed air at 1.17 m^3/min (L254-255), passed through cold and adsorption dryers (Fig. 1a). The injection duration is not reported. At this flow rate, even ~10 s of injection introduces ~0.2 m^3 of desiccated air into a 9 m^3 chamber, i.e. ~2 % dilution, which by itself lowers RH by ~2 %RH.
- The entire claimed seeding signal in the RH record is (0.75 - 0.47) %RH/min x ~7 min, i.e. ~2 %RH.
In other words, the dry carrier gas alone can plausibly account for the whole "enhanced water-vapour uptake" signal (Abstract L20-21; L296-299). The same concern applies to the cold case, where hot combustion products mixed with compressed air are injected (L305-314): combustion releases both heat and water vapour, directly perturbing the very quantities (T, H2O, RH) used as evidence.
Additionally, the paired cases are not thermodynamically matched: the warm seeded case cooled 24 % faster than its background (0.61 vs 0.49 degC/min, L295-296) despite nearly identical pressure-drop rates - so even a perfect, artifact-free RH sensor would show different RH tendencies without any seeding effect. Note also that the interpretive logic is inverted between sections and never reconciled: in the warm case a faster RH decline is presented as the seeding signature, while in the cold case the seeded run shows a slower RH decline (1.46 vs 1.63 %RH/min, L348) without comment.
Required: (a) report injection duration, injected volume, carrier-gas dew point, and temperature for both experiments; (b) perform and present blank-injection controls (carrier air only; and for the cold case, a burnt blank without AgI, or at minimum carrier-air-only) - this is the obvious and necessary control for the paired design; (c) discuss the mismatch in cooling rates and its consequences; (d) reconcile the opposite-sign RH-decline logic between Sects. 3 and 4.
3. The reported saturation ratios are physically impossible and internally contradictory (new; arises from the response to access-review point 4)
Section 4 (L348-353) now derives saturation ratios from the LI-COR H2O concentration and reports a mean Sw = 1.222 (and Si = 1.323) for the AgI-seeded case. This cannot be correct:
- In the presence of the reported droplet population (~2 x 10^3 cm^-3 at ~3-8 um), the phase-relaxation time, tau ~ (4 pi D N r)^-1, is about 0.5 s. Any water supersaturation decays toward its quasi-steady value on sub-second timescales; for the reported cooling rate (0.81 degC/min) the quasi-steady supersaturation is of order 0.1 % or less. A sustained mean supersaturation of 22 % in a dense cloud is off by roughly two orders of magnitude and is, to my knowledge, beyond anything ever reported from any cloud chamber.
- The same figure panel (Fig. 9f) shows the RH sensor at ~86-94 %, i.e. sub-saturated, during the same period. The manuscript thus simultaneously reports Sw ~ 0.9 (sensor) and Sw = 1.222 (LI-COR-derived) for the same air, a discrepancy of 30+ percentage points that is not even acknowledged. Reading Fig. 9f directly makes it worse: H2O ~ 3.5 g/m^3 at the -8.7 degC plateau (where the saturation vapour density is ~2.35 g/m^3) implies an instantaneous Sw ~ 1.5. The background case exhibits the same pathology: in Fig. 9e, H2O ~ 3.0 g/m^3 at -7.8 degC (saturation vapour density ~2.6 g/m^3) implies Sw ~ 1.15 at the very time the RH sensor reads ~84 % and the droplet cloud in Fig. 9c is dissipating - three mutually incompatible records, of which the LI-COR-derived humidity is the outlier in both experiments.
- The likely cause is well documented: an open-path NDIR gas analyzer inside a dense fog does not measure vapour alone - liquid water in and on the optical path contributes to absorption, and LI-7500-class instruments are not specified for in-cloud operation. Combustion-derived H2O from the AgI smoke injection (see Major comment 2) may contribute further. Sensor-temperature mismatch (which T sensor, at which height, was used in the Murphy and Koop conversion?) adds ~7 % per degC to the uncertainty in Sw.
- The background-case values are equally problematic in the other direction: a mean Sw = 0.942 is stated while 5-20 um droplets at up to 3.1 x 10^3 cm^-3 are said to persist for minutes (L332-334). At Sw = 0.942, 10 um droplets evaporate on timescales of tens of seconds; and >30 % activation of the ambient aerosol into droplets is impossible without transient supersaturation. A "mean saturation ratio over the run" is not a meaningful diagnostic.
Required: remove or completely rework the saturation-ratio analysis. Characterize the LI-COR's in-cloud response (or restrict it to cloud-free periods), state which temperature record enters the calculation, propagate uncertainties, and reconcile the result with the co-plotted RH sensor. As it stands, this addition undermines rather than supports the paper's credibility - and the access-review request for a discussion of the supersaturation limitation remains effectively unmet.
4. The 1357 cm^-3 (droplets >= 12 um) value fails a basic water budget (carried over from access review, point 7 - not addressed, now sharpened)
The abstract (L18-19) and Sect. 3 (L284-285, L289) still state that the average concentration of droplets >= 12 um reached 1357 cm^-3 in the warm background case, with no statement of instrument, averaging window, or bin definition.
- Even if every such droplet were exactly 12 um, 1357 cm^-3 corresponds to a liquid water content of at least 1.2 g/m^3 in the >= 12 um class alone - on top of the "majority" of droplets stated to lie at 3-10 um (L284).
- The available condensate cannot support this. Starting from ~94 % RH at ~18.5 degC with ~3.4 degC of cooling, the absolute ceiling (ignoring pump-out dilution entirely) is ~2 g/m^3; accounting for the ~16 % removal of chamber air (and vapour) during the 950 -> 800 hPa pump-down, the sustainable bulk condensate is far smaller. A sustained average of >= 1.2 g/m^3 in large droplets alone is not physically plausible under the reported P-T-RH trajectory, even allowing for some vapour resupply from wetted walls.
- The value is also difficult to reconcile with the authors' own Fig. 7c, where the >= 12 um region shows per-bin values mostly well below ~1000 cm^-3 (pale blue on a scale saturating at 3000 cm^-3); whether a bin-summed average could reach 1357 cm^-3 cannot be judged from the figure - which is precisely why the underlying time series, bin definitions, and the (unreported) total droplet concentration must be provided.
- At implied totals of several 10^3 cm^-3, FM-120/CDP-class probes suffer significant coincidence and dead-time errors, which are nowhere discussed.
Required: state the instrument, time window, and size bins behind 1357 and 478 cm^-3; provide the corresponding time series and total concentrations; provide an LWC closure estimate; and correct or withdraw the value. Since 1357 -> 478 is the abstract's headline warm-cloud result, this must be resolved with raw data, not adjusted wording.
5. Ice-phase identification remains qualitative, and the shown data are internally inconsistent (carried over from access review, point 2 - not addressed, now stronger)
The only ice evidence is still Fig. 8: two single-snapshot HCDS frames with hand-drawn circles, classified "mainly from their non-spherical morphology and optical appearance" (L322-323). The concerns raised at access review stand in full: these appear to be unreconstructed holograms in which out-of-focus droplets produce exactly the concentric ring patterns being circled; no reconstruction, no quantitative classification criterion, no counting statistics, no ice number concentration, and no corroborating CUIT-CIP data are provided (the CIP is listed, its 25-1550 um range is claimed in the abstract, yet not a single CIP datum appears in the paper). Beyond this, the present version adds new problems:
- The control frame fails. Fig. 8a ("Unseeded after 1 min") shows essentially no particles at all - yet at that time the background cloud is stated to contain up to ~3.1 x 10^3 cm^-3 droplets of 5-20 um (L332-334). If the HCDS can image 2-50 um droplets (Table 2), the unseeded frame should be dense with droplet holograms. Either the time label is wrong, the instrument does not detect the droplets it is specified for, or the frames are not comparable. In any of these cases, Fig. 8 cannot carry the paper's cold-cloud conclusion.
- The scale bar itself refutes the classification. Using the 200 um scale bar in Fig. 8b, the dashed-circled "droplet" features are concentric ring patterns roughly 300-440 um across - an order of magnitude larger than any droplet the HCDS claims to measure (2-50 um, Table 2) - i.e. they are unmistakably out-of-focus diffraction patterns, not resolved particle images. The solid-circled "ice" features are dark cores of roughly 50-120 um with surrounding fringes, at or beyond the instrument's stated upper size limit. Classifying particle phase from the morphology of diffraction fringes, without reconstruction, is not defensible. Also "after 1 min" - after what (expansion start? AgI injection?) - is undefined.
- No sample volume is given for the HCDS (Table 2), so the handful of particles in Fig. 8b cannot even be converted into a concentration for a consistency check against the CDP/FM-120 record.
- The claimed Si is incompatible with the shown size distributions. If Si were really ~1.3 at -8 to -9 degC (L353), ice crystals would grow by vapour deposition at rates of order 1 um/s and reach >= 20-50 um within tens of seconds - a growing large-particle mode that Fig. 9d conspicuously does not show (everything simply disappears within ~2 min). The reported Si and the absence of any depositional-growth signature cannot both be true.
- A liquid-only alternative is not excluded. Injecting > 2 x 10^4 cm^-3 of hygroscopic combustion aerosol (AgI pyrotechnics contain soluble iodide salts) quenches supersaturation, shifts the droplet population to smaller sizes, and - under the strong wall heat flux (see Major comment 8) - accelerates evaporation. This alone reproduces the qualitative behaviour of Fig. 9d (fewer, smaller, shorter-lived droplets) without any ice. The repeated hedge "consistent with enhanced ice-phase development" (used throughout Sects. 4-5) is not a demonstration.
Required: reconstructed hologram images with a stated focus/shape classification criterion; the number of holograms and particles analysed, with a false-positive assessment; an ice number concentration time series; CUIT-CIP images/concentrations; a consistency check of HCDS-derived concentrations against the droplet probes; and a serious treatment of the liquid-only alternative. Without quantitative ice detection, the cold-cloud section demonstrates aerosol injection followed by cloud decay - not glaciogenic seeding.
6. AgI activity at the achieved temperatures, and the injection timeline (carried over from access review, point 3 - not addressed)
The chamber was precooled to -4 degC and reached roughly -8/-9 degC in-cloud (Fig. 9f). The manuscript still does not state the minimum in-cloud temperature at which the claimed glaciation occurred, nor compare it with published AgI activity spectra (DeMott, 1995 - cited in the introduction but never used). At -8 degC, AgI-type aerosols are orders of magnitude less active than at -12 to -16 degC; the plausibility of the implied active fraction should be computed explicitly (mass of AgI, aerosol number, expected ice-nucleus yield at -8/-9 degC vs the droplet-loss rate attributed to ice).
Moreover, the injection timing is given only as an absolute clock time, "introduced into chamber B at 16:41" (L320), with no reference to the expansion start - while in Fig. 9b the 0.4-0.8 um band is already at the colour-scale maximum (>= 2 x 10^4 cm^-3) at t = 0 and remains essentially constant over the entire 6-min record: no rise is captured at all. The injection clearly preceded the displayed window, contradicting the narrative "After introducing AgI, the number concentration ... quickly rose" (L328-329). Note also that the two experiments used different protocols relative to cloud formation (warm: powder released 10 s after cloud onset, L272; cold: AgI apparently before expansion), which undercuts the claim of a unified paired framework. Finally, clarify whether "0.5 g of AgI" is the mass of AgI proper or of the pyrotechnic stick.
Required: a full timeline (t_injection vs t_expansion vs t_cloud) for all four runs; the minimum in-cloud temperature; a comparison with DeMott (1995); the AgI dosage definition and stick composition.
7. Humidity sensing near saturation (carried over from access review, point 4 - not addressed)
All RH-based conclusions still rest on a capacitive sensor that reads ~88-94 % (Fig. 7e-f) and ~83-95 % (Fig. 9e-f) while a cloud is present, i.e. it never registers saturation even when the air is demonstrably at or above it. Response time and accuracy near saturation remain uncharacterized; sensor placement is unstated (wall-mounted per Fig. 1b, hence subject to wall thermal offsets). A sensor that is demonstrably biased low in-cloud cannot support 0.28 %RH/min slope differences as a physical signal (see also Major comments 2 and 3). Please also state which of the (multiple) T and RH sensors are plotted in Figs. 7e-f and 9e-f.
8. "Adiabatic" framing vs strongly diabatic reality; consequences for the cloud-lifetime claims (new)
The measured cooling rates are only ~20-30 % of the dry-adiabatic value for the reported pressure-drop rates: e.g. the warm background case (26.84 hPa/min at ~290 K, ~900 hPa) corresponds to an adiabatic ~2.5 degC/min, vs 0.49 degC/min observed; the cold cases likewise (~2.8 vs 0.75-0.81 degC/min). The expansions are therefore dominated by wall heat flux, and the text's framing ("expand and cool nearly adiabatically", L110; "adiabatic expansion", L118, L318) is incorrect. This matters scientifically, not just semantically: with wall heating dominating the energy budget, cloud dissipation ("shorter cloud persistence", a claimed seeding effect, L19, L291-292) is controlled largely by wall-driven subsaturation, mixing, and dilution - all sensitive to small differences in forcing between paired runs (Major comment 2). Please quantify the wall heat flux (it is directly obtainable from the measured P(t) and T(t)), state the mixing state (was the blower fan on during experiments?), and moderate the adiabatic language and the persistence-based inferences accordingly.
9. Aerosol measurements: size-range inconsistency, absent SMPS data, wet sizing, and in-cloud ambiguity (carried over from access review, point 5 - not addressed)
- The abstract still claims submicron coverage of "10-700 nm" (L13-14), which matches neither the SMPS range in Table 2 (7.4-286 nm) nor the CUIT-Aerosol lower limit (200 nm). This was pointed out at access review and remains unreconciled.
- The SMPS - a homemade instrument with a homemade CPC - never appears in the results at all; every aerosol statement derives from the CUIT-Aerosol. Worse, Fig. 2a plots SMPS data extending to ~600-680 nm, far outside the instrument's own stated 286 nm upper limit - Table 2 and Fig. 2 cannot both be right. The calibration itself raises questions rather than settling them: for the 300 nm standard the SMPS shows two peaks of comparable magnitude near ~220 and ~320 nm, and the CUIT-Aerosol response to the 400 nm standard peaks near ~420-430 nm. Please quantify sizing accuracy (mean retrieved diameter and distribution width per nominal size, per instrument) instead of overlaying curves. (Also: the Fig. 2a legend labels the CUIT-Aerosol as "PSD"; please unify naming.)
- All in-chamber aerosol sizing is optical and performed at 90-95 % RH with no drying stage described. CaCl2 deliquesces near ~30 % RH; the reported "aerosol" peaks at 0.3/0.7/1.5 um (L280-281) are therefore wet solution-droplet sizes, not dry particle sizes, and the 1.5 um mode in-cloud may simply be small activated droplets. The access-review request for careful interpretation of the "aerosol" panels during cloud (Figs. 7a-b, 9a-b: interstitial aerosol vs droplets) has not been implemented.
- Fig. 2b-e (CUIT-CDP calibration): state the particle material and refractive index, and whether a Mie-response correction to water was applied; report sizing accuracy and width per nominal size (the 15 um histogram spans roughly 8-24 um). No calibration or validation at all is shown for the CUIT-Fog, CUIT-CIP, or HCDS - the latter carrying the paper's most important claim.
10. Background aerosol and chamber cleanliness (carried over from access review, point 6 - partially addressed, still insufficient)
The added paragraph (L241-249) acknowledges that backgrounds are uncontrolled ambient air, but the specific request - site, date/season, and approximate ambient composition/loading for each experiment - is still unmet (the only time information in the paper is the bare clock time "16:41"). This matters concretely: the cold background contains 5000-10000 cm^-3 in the 0.6-1.4 um optical range (L327-328) - haze-like loadings (plausibly hydrated urban aerosol, but the reader is left guessing), and 5x the warm-case background. There is also an unresolved contradiction: L232-233 states "all air inlets are equipped with HEPA-grade filters", yet the backgrounds are explicitly formed from re-entering ambient air (L241-243) - is the bottom re-entry valve unfiltered? Finally, for a facility paper, the standard chamber-characterization elements the authors themselves cite as important (Dias et al., 2017; Schnitzhofer et al., 2014 - L58-59) are absent: particle wall-loss timescales, mixing time, leak rate, post-cleaning background counts, and - critically for paired runs - the run order and interval (was the seeded run always second? what residual particle level remained from the preceding run?).
11. Sampling representativeness of the externally plumbed probes (new)
- The CUIT-CIP samples at 5 L/min through a 40 mm inner-diameter line (L133-134): the mean duct velocity is ~0.07 m/s. Particles of 100 um (terminal velocity ~0.3 m/s), let alone 1550 um (~5 m/s), cannot be aspirated representatively into such a flow. The claimed 25-1550 um capability is not credible without an aspiration/transmission-efficiency analysis - consistent with the fact that no CIP data are shown (Major comment 5).
- All external probes sample from the bottom of the chamber (L131-132), where settling enriches large droplets - a bias directly relevant to the >= 12 um claims (Major comment 4). Please quantify or at least discuss.
- In-situ instruments (FM-120, LI-COR, HCDS) sit near the bottom sensors and dissipate heat (the LI-COR has heated optics); their thermal footprint in a sealed 9 m^3 volume is unaddressed and may contribute to the low in-cloud RH readings (Major comment 7).
- Table 2 lists "sampling flow rates" of "0.2-20 m/s" (CUIT-Aerosol) and "2-10 m/s" (CUIT-CDP) - these are velocities, not flow rates, and the actual sample flows/velocities used in the experiments are never stated.
12. The seeding materials themselves are unquantified (new)
For a paper titled "Quantitative Seeding Experiments", neither seeding agent is characterized: the warm-cloud powder is described only as "primarily composed of CaCl2, plant ash, and composite materials" (L256-257) - no mass fractions, no dry size distribution, no hygroscopicity (kappa or growth factor), no purity; "plant ash and composite materials" is not a reproducible specification. The AgI "smoke sticks" (L306) similarly lack manufacturer/composition, AgI content, and combustion-aerosol size characterization. The community standard the authors themselves cite does exactly this (Tajiri et al., 2015; Kim et al., 2025 - powder-type NaCl/CaCl2 characterization in K-CPEC). Without material characterization, neither experiment can be reproduced, in this chamber or any other.
MINOR COMMENTS
1. Sect. 2.2 / Fig. 3b: nearly every number in the text is contradicted by the figure. The text (L186-189) gives 102 min at 0 degC + 110 min (0 -> -10 degC) + 4 min hold at -10 degC + a 189-min descent to -20 degC + 30 min at -20 degC, i.e. ~435 min in total, but the Fig. 3b axis ends at 200 min. In the figure: the 0 degC plateau does last ~103 min (the only matching number); the 0 -> -10 degC transition takes ~5 min, not 110 min; the -10 degC plateau lasts ~25-30 min, not 4 min - and sits at about -10.5 to -11 degC, i.e. at or outside the stated -10 +/- 0.5 degC, with an Upper/Lower spread of ~0.5-0.8 degC that contradicts the "< 0.3 degC" inter-layer claim (L190-191); the descent to -20 degC takes ~50 min, not 189 min; and -20 degC is reached only near t ~ 190 min and held for ~10 min at most, not 30 min (the Lower sensor bottoms out near -18.5 degC). Please re-derive this entire paragraph from the raw logs and restate the plateau-stability claims accordingly.
2. Thermal-uniformity statements need tightening: L190-191 claims vertical differences < 0.3 degC, L192 claims < 1 degC; state clearly which applies under which conditions, and translate the +/-0.5-1 degC spread into the corresponding equilibrium-RH inhomogeneity (~+/-3-7 % near saturation), which is directly relevant to Major comments 3 and 7.
3. Chamber A / differential-pressure mode is described at length (Sect. 2.1, Table 1, Fig. 1) but never demonstrated (L115-116: all experiments in chamber B, valves closed). State explicitly that the dual-vessel mode is untested and defer performance claims for it.
4. L285 "Larger droplets then formed through collision-coalescence and condensational growth": collision-coalescence is negligible for a quiescent 3-10 um population on ~5 min timescales; remove or substantiate.
5. Figs. 7/9 (a-d): define the plotted quantity (per-bin concentration vs dN/dlogDp), the bin widths, and the instrument used for the droplet panels (c-d) - FM-120, CUIT-CDP, or CUIT-Fog? This information is still missing (access-review point 7 asked for it explicitly).
6. The cold and warm backgrounds differ by a factor of ~5 (<= 2000 vs 5000-10000 cm^-3); a sentence comparing the two ambient situations (and their dates) is needed for the paired-design logic to be assessable.
7. Fig. 5: annotate the shaded bands; y-axis "Humidity (%)" -> "Relative humidity (%)"; the caption lacks a final period. The abstract's "nearly the full 0 to 100 % range" should be reconciled with the demonstrated 10.3-98.4 %.
8. Data availability: the Mendeley record (Luo, 2025) should be verified to contain the raw data behind Figs. 2-9, including the raw holograms and the processing/reconstruction code; at present the paper's key claims cannot be independently re-derived.
TECHNICAL CORRECTIONS
- Title: remove "Quantitative" (see Major comment 1).
- Sect. 2.1 (L92-108): words are broken across lines without hyphens ("chambe r B", "optic al windows", "differentia l-pressure", "mechanica l vacuum pump", "botto m valve") - please fix the typesetting throughout.
- Terminology: "catalyst/catalytic/flame-based catalysts" persists throughout (L106-107, Fig. 1a labels, L252-266, L363, L371) despite the access-review correction; the accepted terminology is "seeding material/agent". The section titles were fixed; the body text was not.
- L273-274: a figure caption has been pasted into the running text ("Fig. 6 Sequential images of warm-cloud formation ... after cloud formation."), producing a broken sentence; the actual Fig. 6 caption still lacks the (a)-(c) panel descriptions.
- Fig. 2: legend "PSD" vs caption "CUIT-Aerosol" (unify); double space in "(b-e) Calibration"; state the calibration particle material (see Major comment 9).
- Table 2: "Manufacture" -> "Manufacturer" (and the column mostly lists instrument acronyms, not manufacturers - the CUIT instruments should name the developing laboratory); "0-60 mmol" -> presumably mmol/mol (and give the CO2 range or drop the CO2 mention); the HCDS sample volume is missing; velocity vs flow-rate units (see Major comment 11); "-60-160 degC" formatting.
- Fig. 4 caption: "Figure 4," -> "Figure 4."; the caption ("Wall Temperature stability") mislabels panel (a), which shows air temperature; capitalization.
- Fig. 8: define the reference time of "after 1 min" in the caption; move the solid/dashed in-image circle legend into the caption; state the hologram frame dimensions, pixel resolution, and depth of field.
- Overprecise single-run averages (117.44 / 72.60 hPa/min; 26.84 / 27.72; 32.63 / 32.91; 1357 / 478): round to a defensible precision and define the averaging windows (the P(t) curves are visibly nonlinear).
- L320 "at 16:41": replace with time relative to expansion onset (see Major comment 6).
- L360-361: "temperature, humidity, pressure, and water vapour" - humidity and water vapour are redundant here.
- References: Schnitzhofer et al. is cited as 2013 in the text (L59) and dated 2013 in the list, but the DOI (amt-7-2159-2014) is from 2014; fix both. Khain et al. (2024) is given an SSRN preprint DOI for a J. Weather Modif. article; verify the final metadata. Data availability: "(Luo,2025)" is missing a space.
- Acronyms AIDA, CLOUD, MRI, K-CPEC are undefined at first use (L53-63).
- L92-93: the stated chamber-B dimensions (1.8 m diameter x 3.6 m height cylinder) give ~9.16 m^3, outside the stated 9.0 +/- 1 %; presumably head geometry - please clarify.
- Financial support: missing spaces ("Project(2025ZYD0179),the", "Administration(CMA ...").
- Redundancy: the "initialized from the same ambient aerosol source" statement appears three times nearly verbatim (L244-249, L267-269, L316-318); condense.
- Supplement: if the uploaded supplement is still the response letter to the previous editorial comments, it should be posted as an author comment, not as supplementary material (carried over from access review).