The manuscript explores aerosol effects along the stratocumulus to cumulus (Sc–Cu) transition and further downwind into deep cumulus convection in the tropics. The authors combine satellite observations, reanalysis data, Lagrangian trajectories, and large-eddy simulations. The results concerning the Sc–Cu transition are generally consistent with earlier work. The main contribution of this study is its extension toward the transition to deep convection and the suggestion that aerosols may affect thermodynamic properties along this evolution.

The extension of Lagrangian analyses into the deep convective regime is a potentially valuable direction, and the manuscript addresses an important question in this context. However, several aspects of the methodology raise concerns that may affect the strength of the conclusions regarding aerosol-driven thermodynamic feedbacks, particularly in distinguishing aerosol effects from aerosol–meteorology co-variability. Overall, the results are suggestive of a possible aerosol influence on thermodynamics, but stronger evidence would be needed to robustly support this interpretation.

In addition, the manuscript would benefit from improved writing, particularly in the organization and clarity of the methodology section. Below I provide detailed comments.

Major concerns

1. Simulations

The authors note that aerosol–meteorology co-variability may precondition the Sc–Cu–DC evolution, which is clearly evident, for example, in Figure 3. Several recent studies have emphasized that polluted and clean air masses often exhibit systematically different thermodynamic and meteorological properties, which complicates causal interpretation. Relevant literature on aerosol–meteorology co-variability should be more thoroughly discussed and cited (e.g., https://acp.copernicus.org/articles/25/3413/2025/ and https://doi.org/10.5194/acp-24-7331-2024).

To isolate aerosol effects while keeping meteorology fixed, the authors employ LES simulations. Figure 6 shows differences between polluted and clean simulations, but the description of the simulation methodology is unclear and raises concerns regarding the interpretation of these results. It is not clear whether the LES are intended to closely mimic the observed meteorological evolution along the trajectories or whether they represent semi-idealized simulations. While the authors state that simulations follow the trajectories, the description suggests a more idealized setup. This ambiguity is reinforced by the statement in the conclusions that the simulations rely on idealized setups with prescribed CCN perturbations.

Several key methodological details are insufficiently described:

Are the simulations nudged, and if so, how?

How do atmospheric profiles evolve during the simulations?

How are differences in temporal resolution between the LES and reanalysis data handled?

Aerosols are not prognostic in the simulations, which is a critical limitation given that the study focuses on aerosol–cloud interactions. This point requires deeper discussion regarding its implications.

The horizontal domain size (57.6 × 57.6 km) is marginal for resolving mesoscale cloud structures during the later stages of the Sc–Cu transition, and probably not sufficient to resolve deep convection. This concern is acknowledged by the authors themselves in the conclusions, where they state that the domain is not large enough to capture convective organization. This limitation weakens the interpretation of the simulated deep convective response and raises concerns regarding the conclusions.

2. Observations

Several issues arise in the satellite-based analysis. For example, Figure 4 shows droplet effective radius along the 8-day evolution. During days 6–8, clouds appear substantially deeper and likely include glaciated cloud tops. In such cases, MODIS liquid cloud retrievals such as re and LWP may not represent the deep convective clouds.

The authors choose AOD as the primary aerosol proxy, while acknowledging its limitations later in the manuscript. It is not clear why cloud droplet number concentration (Nd), which may be more directly related to CCN, was not considered more centrally. AOD is not available under cloudy conditions, especially over stratocumulus regions where cloud fraction is high, which adds further uncertainty. To address this, the authors use reanalysis AOD, which may introduce additional uncertainty beyond that of the AOD retrieval.

Furthermore, AOD is averaged along the entire trajectory. Why was aerosol loading at the initial day not used, as in previous studies? This relates directly to the unexplained increase in AOD after day 5 in Figure 2, which requires clarification.

Regarding MODIS observations, it should be clarified whether there are time periods without observations due to satellite overpass limitations and how this affects the daily averaging.

3. Trajectories

The choice of three initial trajectory points that are geographically very close requires justification. Given the scale of spatial meteorological variability, one would not expect substantial differences between such closely spaced points. The noisy results in Figure 5 across these points may introduce additional uncertainty that should be discussed.

4. Interpretation of the hermodynamic effects

One of the most interesting aspects of the manuscript is the suggestion that aerosols may actively modify thermodynamic profiles along the cloud transition, as discussed in Section 3.3 and illustrated in Figure 9. This interpretation is intriguing and consistent with mechanisms proposed in previous modeling studies. However, given the co-variability between aerosols and large-scale meteorological and thermodynamic conditions (see major comment 1), the current analysis does not yet unambiguously isolate an aerosol-driven thermodynamic effect.

While the consistency between observations and LES results is encouraging, the observational evidence alone cannot fully rule out the influence of pre-existing environmental differences, and the idealized nature of the simulations limits the strength of causal attribution. As a result, the conclusions regarding aerosol-induced thermodynamic feedbacks may benefit from a more cautious framing or from additional sensitivity analyses that further constrain the role of co-variability.

Minor comments

The manuscript provides very detailed figure-by-figure descriptions. In many cases, the figures can speak for themselves, and the text could focus more on interpreting the key results.

Line 120: The rationale for limiting SST variability is unclear and should be better explained.

Line 124: Is a precipitation threshold defined?

Line 203: Why is this procedure applied only to the polluted group?

Line 218: Clarify what is meant by “Differences will be mentioned.”

In figures such as Figure 3, consider using δ notation in parentheses for clarity.

Line 261: Missing citation of relevant recent studies (https://acp.copernicus.org/articles/25/3413/2025/ and https://doi.org/10.5194/acp-24-7331-2024).

Figures 5 and 8 are visually noisy. Consider improved visualization.

Figure 7: Differences between polluted and clean simulations are difficult to discern. Improved visualization would help.

Line 387: Clarify what is meant by “after accounting for SST differences.”

Lines 421–423: This statement is not fully supported by the presented results.

An AI assistance statement appears after line 210 and seems misplaced.

General comments:

Aerosol impacts on cloud and precipitation has been investigated substantially in both observations and modeling studies. However, it is relatively rare for clouds along the subtropical to tropical transition, especially toward deep convection regime. In addition, aerosol effects are generally hard to be isolated because of its co-variability with meteorological conditions in observations, but this could be better resolved in model simulations by prescribing meteorological conditions. This study is unique in both regards with well-designed observational analyses and modeling approaches. The figures are well prepared with good presentation, but a better interpretation of results would strengthen the work.

Specific comments:

I have some comments and questions to be addressed before acceptance.

1. It is noted that SST is higher in polluted scenarios, especially toward later days (Fig. 3a). but why?
2. The LES simulation setup and configuration can be better described. For example, for the model setup, it is mentioned that large-scale wind forcing is prescribed using the observed zonal (u) and meridional (v) wind components. However, the moisture and temperature advection is not mentioned?
3. Although limitation of prescribed CCN in the simulation is mentioned, its potential impact on the thermodynamic conditions can be included with some references for a more complete view.
4. The difference between observations and modeling is large (Fig. 9). Although this has been briefly noted in the text, but it would be more revealing with more description and explanation. For example, it would provide more information if the difference could be described by accounting for the different regimes (Sc-Cu-DC). Another point to note is that the qv difference is not only larger in observations, but also reaching higher altitudes than modeled. This deserve more description and explanation, especially for the deep convection regime.