I find this to be an outstanding manuscript regarding the timescales of LWP adjustments in non-precipitating marine stratocumulus clouds. I really have no substantial suggestions for improvement. 

1. Perhaps all prime quantities could be written as dot quantities (symbol with a dot overhead) instead? This is the more typical shorthand notation for time tendencies in the physics community I believe. 

2. A reference for the -0.4 value would be appreciated.

3. Line 260 - typo: So --> To

Review of "Magnitude and timescale of liquid water path adjustments to cloud droplet number concentration perturbations for nocturnal non-precipitating marine stratocumulus" by Chen, Prabhakaran, Hoffman, Yamaguchi, Kazil and Feingold, Manuscript egusphere-2024-3891

Recommendation: Minor revisions

This nice paper revisits the earlier work of many of the coauthors in Glassmeier et al (2021, Science), which suggested a strong thinning of non-precipitating stratocumulus in response to increased aerosol concentrations for nocturnal-only simulations in the limit of long time behavior.  Here, a subset of cases from Glassmeier et al has been selected, redesigned to include a set of aerosol perturbations (seeding) for each case and other changes, re-simulated and re-analyzed.  The main result is that a shorter entrainment-related adjustment timescale seems to dominate the liquid water path evolution of the clouds up to about 12 hours after seeding.  The analysis is clear and well-presented, as suggested by the relatively few suggestions for wording changes (which is unusual for me).  My comments ask somewhat technical questions about how to think about approaching equilibria when multiple adjustment timescales exist.  I also question whether it might be possible to conclude the paper in a stronger way, as I found section 4 (while interesting) distracted a bit from the main story of the paper.

=========================

Major comments:

1. (14/289) I am somewhat uncomfortable with equation 19 as stated, because (1) (as the paper very clearly states) there are multiple timescales involved in the adjustment to equilibria and (2) that it is difficult to identify L_\infinity without actually running to equilibrium over very long timescales (days to 10s of days).  Could some language be added to make clear that the diagnosed value of \tau is only valid locally in time and may well evolve with the state of the system?  Also, is the L_\infinity here really the equilibrium value of LWP, or is it more representative of a point on the slow manifold (maybe L_qe for quasi-equilibrium) where the fast timescale has adjusted but the slower adjustment timescales are still at work?

2. The material in section 4 is very interesting for me, but I found that it distracted from the flow of the paper to its conclusion.  Might some of the sensitivity studies be moved to one or more appendices, so that the main body of the paper might be more focused?  The authors may have a different view, but I worried that the size/variety of material in the discussion might affect the impact of the paper on its readers.

==========================

Specific/minor comments (11/240 means p. 11, line 240):

14/294: I would speculate that some of the oscillations in \delta L and \delta L' after seeding result from horizontal heterogeneities of aerosol and cloud droplet number in the cloud layer.  With seeding happening below 400m, the increased aerosol would be carried into the cloud layer in updrafts and spread horizontally over time.  If frequent outputs were not saved after seeding, this may be unknowable.  If horizontal heterogeneity is driving some of the oscillations, you might consider citing Dhandapani et al (2025, https://doi.org/10.1029/2024MS004546) which talks about the preferential lofting of seeded aerosol in updrafts.  If the authors know a better/earlier reference, that might also be helpful.  The origin of this comment is partly that the timescale analysis is partly based on the assumption that the LES is behaving more or less like a mixed-layer model, and the horizontal heterogeneities would deviate from the well-mixed assumption in the same way that a decoupled boundary layer would in the vertical.

15/305: Regarding: "Considering that L for BASE is approaching its steady state, this means that L for N400 is probably approaching a different steady state."  Is it really obvious that the different simulations are approaching different steady states?  They're certainly on different slow manifolds because w_e is a function of both buoyancy flux and Nd.  However, the N perturbation is steadily decaying over time.  My guess is that all of the seeded simulations would converge to the long-term-equilibrium solution of BASE.  For this not to be true would require a solution where the higher-Nd entrainment would match the subsidence rate at a particular inversion height where the aerosol budget is also at equilibrium (with the surface source balancing changes due to entrainment and collision-coalescence).  However, this would only apply if the boundary conditions (i.e., free tropospheric temperature and moisture profiles) remain fixed, which seems to not be true in the default setup here.

While I've raised this question, I wouldn't encourage the authors to spend a lot of computer time trying to figure this out (though it could be done pretty easily in a mixed-layer model that includes prognostic aerosols, using the entrainment parameterization in the companion Hoffmann et al paper).  My suggestion would be to think about whether this would impact ideas about the value of k_\infinity.  My view is that the quasi-equilibrium value of k (maybe k_qe) computed between points on different slow manifolds (with different Nd) are probably well defined and more meaningful than the true equilibrium value since the equilibrium will take a very long time to reach.

17/345: Please write these equations down explicitly, either here or in the supplement.  It's too much to expect the reader to reconstruct them.

17/355-360:  The "intentional"/"unintentional" language seems odd to me, though maybe that reflects my background.  In my mind, "unintentional" = "perturbed boundary conditions" and "intentional" = "perturbed model state" or "perturbed initial conditions".  Hopefully, the reader should understand that perturbed BCs are likely to result in a different equilibrium solution while perturbed ICs will not, but it may be useful to remind the reader of this.

20/433: Regarding "... their L may carry the "memory" of different N."  I don't quite understand what's being discussed here.  Is it just that the state of the boundary layer (zi, Nd, etc.) is different than the BASE case because it went through a period of intensified entrainment after seeding?  Or does a quantity like entrainment depend on both the present and past values of Nd?  The latter seems unlikely to me.  If it's the former, I don't think the "memory" effect is meaningful, because L and N aren't a complete specification of the state of the system, so of course two simulations with identical L and N could vary because (even for a mixed-layer model) there are two more state variables that constrain the system.  This reliance on a limited number of state variables (or quantities derived from state variables like L) also hides some of the disequilibrium of the system and is related to my comment about L_\infinity above.

23/498-500: Are there any worries that seeding-induced decoupling might lead to cloud thinning, independent of the LWP decreases in the mostly well-mixed boundary layers seen in this work?

24/506: Please add "diurnal cycle" to the list in parenthesis.  It's arguably the most important outside timescale that's neglected in the simulations in this paper.