| 25 Sep 2025
Lagrangian Particle–Based Simulation of Aerosol-Dependent Vertical Variation of Cloud Microphysics in a Laboratory Convection Cloud Chamber
Abstract. We investigate the vertical variability of cloud microphysics in a turbulent convection cloud chamber through large-eddy simulations coupled with a Lagrangian super-droplet (SD) model. Numerical experiments mimic the convection chamber under construction at the Korea Institute of Science and Technology (KIST), employing realistic aerosol size distributions derived from field measurements (VOCALS campaign and Seoul, South Korea). Simulations show that cloud water mixing ratio generally increases with altitude due to continuous droplet activation and growth during ascent, but this vertical gradient weakens significantly as aerosol concentration increases. Enhanced aerosol loading intensifies competition for water vapor, shortening phase relaxation times and suppressing supersaturation variability, resulting in more vertically uniform cloud water mixing ratio profiles. Lagrangian trajectory analyses reveal that upward motion substantially influences droplet growth and activation under clean conditions, but this influence diminishes sharply in polluted environments where rapid vapor depletion limits supersaturation buildup. Furthermore, droplets experiencing sustained upward motion activate more readily, although this effect is weakened considerably in polluted conditions. In a turbulent convection-chamber setting, we quantify how aerosol loading modulates the vertical distribution of cloud-water mixing ratio using LES model with Lagrangian super-droplet tracking, thereby clarifying the coupled roles of vapor competition and vertical transport. Our results emphasize the critical role of aerosol loading in shaping vertical microphysical structures and highlight the interplay between vapor competition and vertical dynamics. These findings provide important insights for improving cloud parameterizations and understanding aerosol-cloud interactions in both controlled laboratory and atmospheric contexts.
Review of “Lagrangian particle-based simulation of aerosol-dependent vertical variation of cloud microphysics in a laboratory convection cloud chamber”
This paper presents a computational investigation of cloud droplet formation and growth in a 2-meter-tall convection chamber. A unique aspect of the work is the use of Lagrangian microphysics, which allows the time history of droplets to be investigated. Some of the results are intriguing and overall the work appears to be rigorous and technically correct (with one exception described below). The connections to atmospheric processes are somewhat tenuous, so it seems the paper would be more suitable for AMT than for ACP. To be published in ACP those connections should be strengthened.
The main criticism is regarding the way some of the results are discussed and interpreted. In multiple places in the manuscript the wording suggests that supersaturation builds up during ascent. This has the potential to be very confusing because some readers may associate this with the familiar increase of supersaturation with height in a convective cloud. For example, the explanation of increasing q_c with height being attributed to “active droplet activation and condensational growth during ascent” (lines 198-199) does not make sense. Symmetry suggests that the same argument should work for plumes descending from the top surface. A cold plume mixed into a warm background has nearly the same potential to generate supersaturation as a warm plume mixed into a cold background (see multiple previous papers showing calculations and simulation results for convection-cloud chambers… usually the supersaturation profile is nearly constant with height, showing no significant bias toward the bottom or top surface). In contrast, arguments that appear later in the paper (such as on lines 380-382) make much more sense in light of the results shown. I would encourage the authors to revise the language in other parts of the paper that give the potentially misleading impression that droplet growth is tied to vertical ascent. The peak near normalized vertical displacement of negative 0.25, suggests that plumes that are efficiently mixed produce the strongest supersaturation (as opposed to a plume that simply travels upward without strong mixing). Supersaturation generation in a convection-cloud chamber is the result of mixing, not of ascent. That mixing may have a link to plume propagation distance and therefore to distance from the top or bottom boundaries, but it is not the ascent itself that is the source of supersaturation.
Other important points to be addressed are:
What is the fundamental problem statement or hypothesis underlying the investigation? For example, on lines 58-60 the direction of this investigation is stated, but it is not clear why this topic would be of interest to explore.
A “free-slip” boundary condition at the top and bottom surfaces (line 106) is unphysical and may fundamentally alter the nature of the flow. Can this be justified as having negligible effect on the microphysical problem investigated in this study? If so, please provide clear evidence. If not, please clearly state this limitation in the abstract and provide a discussion of how the conclusions will change.
In multiple places in the manuscript it is pointed out that mean and fluctuations of supersaturation are reduced as cloud droplet concentration is increased. Suppression of supersaturation for shorter phase relaxation time has a clear theoretical interpretation; for example, see Equations 12 and 13 in Chandrakar et al. (Journal of the Atmospheric Sciences, 2018, “Influence of Turbulent Fluctuations on Cloud Droplet Size Dispersion and Aerosol Indirect Effects”; note that in that paper is the phase relaxation time).
There are also multiple places in the manuscript that would be more clearly interpreted in terms of the quasi-steady supersaturation. For example, on line 322, the statement that condensation causes “rapid vapor depletion… before S can develop” does not make sense. Without supersaturation, there is no vapor depletion. I suppose what is meant is the analog to quasi-steady supersaturation, which is the result of two rates: the rate at which supersaturation is produced versus the rate at which it is consumed. So the low supersaturation observed is simply a result of faster consumption by droplet growth (due to shorter phase relaxation time) for an unchanged rate of production. This is not a surprising result. Another example is lines 218-219: There the concept could be made clearer by drawing an analogy with the quasi-steady supersaturation in a convective cloud. For equivalent forcing, such as updraft strength, smaller phase relaxation time leads to lower quasi-steady supersaturation.
Figure 5 shows large oscillations in the mean supersaturation with height near the top and bottom boundaries. What is the source of this numerical instability? The concern is that the oscillations could strongly influence the correlation with vertical air speed, which is an important focus point in this investigation. This needs to be addressed and its influence on the w’-s’ correlation needs to be checked.
Specific comments
Lines 102-105: In one sentence the “sidewall temperatures” are stated, but in the next line it is stated that the model is configured with “periodic lateral boundaries”.
Lines 112-114: How significant are droplet losses to the sidewalls compared to the bottom surface?
Lines 157-158: What is the reason for the arbitrary use of 1 micrometer as a cutoff between haze and cloud droplets? For the broad aerosol size distributions used in this study, this can be very different than the true activation radii. An advantage of Lagrangian microphysics is that the aerosol properties are known for each super-droplet, so the boundary between haze and cloud can be determined for each super-droplet. Please explain whether this simplification has any significant consequences for the interpretation of results.
Line 188: It’s not incorrect, but it seems a little strange to use “altitude” instead of “height”.
Line 194: It isn’t clear to me whether these refer to averages over horizontal planes at 0.3 m and 1.7 m, or to averages over layers (e.g., from 0 to 0.3 m and from 1.7 to 2.0 m). Also, is there any particular reason for choosing the value of 30 cm?
Figure 3 caption: For the benefit of those reading the paper quickly, it would be helpful to provide a qualitative description of within the caption.
Line 229: It is not clear what is meant by the “largest aerosol mode”.
Figure 5 caption: Clarify that the vertical profiles are for the horizontally-averaged supersaturation. Also, are the histograms for supersaturations within the full volume, or a subset of the chamber volume?
Line 265: Should be “high aerosol concentration”.
Line 267: Should be “The physical explanation…”.
Lines 281-283: The “rapid equilibration” argument makes sense, if by that it means that the supersaturation is rapidly depleted. It is unclear what is meant by “droplets complete their growth before experiencing the full extent of vertical S variability”. Does it mean that the supersaturation is depleted after relatively short ascent, such that droplet growth ceases above some level? As written, this argument is too vague.
Line 314: The definition “the timescale over which S environment evolves” is ambiguous. One could also say that the phase relaxation time is the characteristic time for the supersaturation to evolve (due to droplet growth). I assume what is meant is that it is the timescale on which the supersaturation varies due to turbulent mixing.
Line 316: The units are for a velocity standard deviation, not a velocity variance. Please correct either the wording or the units.
Line 359: The statement that “sustained upward motion enhances droplet growth by exposing SDs to regions of higher s” is inconsistent with the results shown. For most of the simulated cases, a peak occurs at a negative value of normalized vertical displacement, which suggests the opposite of what is stated. Only in the two cleanest cases is the mean growth associated with positive vertical displacement.
Line 388: Should be “confinement of the S production zone”.
Lines 403-406: The two statements that “activation fraction generally increases with normalized vertical displacement” and “it peaks around a normalized displacement of approximately -0.25” are inconsistent. There is an initial increase, but over most of the range of normalized vertical displacement the activation fraction decreases or is only weakly changing.
Lines 408-414: This discussion is not clear.
Figure 14: Wouldn’t it be more insightful to show the spatial locations of where activation occurred, rather than the locations of droplets that were activated within the last 20 seconds, which is sufficient time that they can be transported over the maximum dimension of the chamber? This seems like a missed opportunity to gain deeper understanding of what processes are controlling activation.
Lines 456 and 463: After reading the paper and looking at the evidence, I disagree with the statements that “droplets experienced sustained condensational growth during ascent” and “droplet growth (dr) and activation strongly correlated with upward displacement”.