the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Diurnal evolution of non-precipitating marine stratocumuli in an LES ensemble
Abstract. We explore the impacts of the diurnal cycle, free-tropospheric (FT) humidity values, and interactive surface fluxes on the cloud system evolution of non-precipitating marine stratocumuli based on a large ensemble of large-eddy simulations. Cases are separated into three categories based on their degree of decoupling and cloud liquid water path (LWPc). A new budget analysis method is proposed to analyze the evolution of LWPc under both coupled and decoupled conditions. More coupled clouds start with relatively low LWPc and cloud fraction (fc) but experience the least decrease in LWPc and fc during the daytime. More decoupled clouds undergo greater daytime reduction in LWPc and fc, especially those with higher LWPc at sunrise because they suffer from faster weakening of a net radiative cooling. During the nighttime, a positive correlation between FT humidity and LWPc emerges, consistent with higher FT humidity reducing both radiative cooling and the humidity jump, both of which reduce entrainment and increase LWPc. The time rate of change in the LWPc is more likely to be negative for higher LWPc and greater inversion base height (zi), conditions under which entrainment dominates as turbulence develops. In the morning, the rate of the LWPc reduction depends on the LWPc at sunrise, zi, and the degree of decoupling, with distinct contributions from subsidence and radiation. Under well-mixed conditions, it takes about 10 h for the surface fluxes to offset 15 % of the changes in entrainment warming and drying, assuming no changes in transfer coefficients or surface wind speed.
- Preprint
(4642 KB) - Metadata XML
-
Supplement
(136 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2024-1033', Anonymous Referee #1, 04 Jun 2024
Review of "Diurnal evolution of non-precipitating marine stratocumuli in an
LES ensemble" by Chen, Zhang, Hoffman, Yamaguchi, Glassmeier, Zhou and Feingold, Manuscript egusphere-2024-1033Summary:
An ensemble of large eddy simulations of marine stratocumulus clouds is run and analyzed for a variety of idealized summertime conditions over the Northeast Pacific Ocean. Building on earlier work by led by co-authors Hoffman and Glassmeier, the new simulations use a more realistic radiation parameterization, including a diurnal cycle, as well as interactive surface fluxes. The analysis seeks to understand how the diurnal cycle of cloud cover and LWPc (in-cloud liquid water path) are determined by the base state (sorting into bins with high/low decoupling and LWPc) and by different processes in understanding how different processes fix LWPc. Highlights include: daytime reductions in cloud fraction and LWPc are larger in more decoupled boundary layers. Larger free tropospheric humidity tends to support higher LWPc during nighttime.
Assessment:
The paper is well-written and nicely tells a story about how stratocumulus-capped boundary layers vary across the diurnal cycle. The manuscript is interesting and compelling, but I was struck that the paper focuses much more strongly on LWPc than on cloud fraction, which is presumably a stronger control on shortwave cloud radiative effects during the diurnal cycle in many of these simulations. However, I expect that another paper based on these simulations will tell that story later. I think the manuscript is a good fit for ACP and should be published, though I would ask the authors to consider the suggestions I make below along with those of the other reviewers.
Recommendation: Minor Revisions
===========================
Major comment:
1. Regarding the LWPc budget:
1a. For the domain-mean budget, it is natural for the subsidence term to be based on the prescribed, large-scale subsidence. However, for the cloud-volume (CV) budget, I find it puzzling that the mean vertical velocity in cloudy columns was not used. I did not ponder this until the end of section 6.1 (on the uncertainty in the entrainment term). If the mean velocity profile in the cloudy columns was not saved, then keep the analysis as it is, but if not, wouldn't the analysis be more robust if the mean cloudy-column subsidence was used for the SUBS_CV term? Then, the ENTR_CV term would be based on the actual entrainment rate in the cloudy-columns, which might be more closely tied to the radiative cooling there. The authors are free to argue that the current setup is more compelling and/or that the differences are only important in the afternoon when the cloud fraction falls precipitously in some of the simulations, but I wanted to suggest this.
1b. The LWPc budget involves large terms with different signs (e.g., RAD, ENTR, BASE-n-LAT), with the important d(LWPc)/dt trend being the small residual after these large terms almost --- but not exactly --- offset each other. Since it's hard to visualize the result of such cancellation, I found myself seeking some way to understand the budget better. From this, I had two (very optional) suggestions:
- Instead of decomposing the dz_i/dt term into separate w_s and w_e as specified on line 272-273, include the whole d(z_i)/dt term in the budget. This is easier for the reader to understand and avoids another pair of opposing terms in the budget. Helpfully, this term switches sign from night to day in some of the simulations, so that changes in dz_i/dt --- driven here by changes in entrainment --- may play a role in the diurnal cycle of LWPc. Doing this would modify the ENTR term in the LWPc budget, since it would only represent the entrainment impacts on cloud base height through d<qt>/dt_CV and d<thetal>/dt_CV. Only make such a change if the authors find it clarifies the story about the budgets.
- (MORE SPECULATIVE) Re-group the terms, combining exchange within the boundary layer (BASE-n-LAT) with radiation (RAD) since both drive increases in LWPc. Both can also be seen as driving entrainment, both the familiar cloud top cooling through (RAD) and fluxes through cloud base (BASE). Write ENTR = - C_ENTR * (RAD + BASE-n-LAT), where C_ENTR is something like the entrainment drying efficiency of the (RAD + BASE-n-LAT) terms. Then, the budget for LWPc is dominated by SUBS and (1-C_ENTR)*(RAD + BASE-n-LAT), with variations in C_ENTR controlling whether LWPc grows or decays. Looking at the budgets in figure 8, C_ENTR seems roughly constant during the nighttime but varies across the different classes of simulations. I would hypothesize that the differences between C_ENTR across the simulations depends on things like the jumps across the inversion, free tropospheric humidity, decoupling and perhaps other quantities as well, so the differences in the outcomes, e.g., positive/negative d(LWPc)/dt, might be able to explained in terms of those quantities. As noted above, this is pretty speculative, but (if it works) could potentially make the storytelling a bit cleaner.
===========================
Specific/minor comments (10/277 means p. 10, line 277):
1/3: My understanding is that the "cloud liquid water path (LWPc)" refers to the mean cloud liquid water path in cloudy columns. If so, could this be stated explicitly somewhere, even if it is broadly understood by others in the field. Also, please explain somewhere how a "cloudy column" is defined.
4/94: The broad range of initial boundary layer temperatures (10K) suggests that the initial boundary layer is relaxing towards quasi-equilibrium with the SST throughout the day, as suggested by the surface temperature jump in figure 4c. It's great to have such a variety of cases here, but I found myself wondering how such a transients might impact the results. Note also, that the mesoscale organization will also be developing through the nighttime and that this transient might have some impact on the results (e.g., the timing of precipitation development overnight).
10/277-278: The last subsection seemed to be foreshadowing a cloud-volume budget, so it's worth expanding BL and CV here for clarity.
Fig 6/bottom of page 10 (OPTIONAL): Given the degree of decoupling discussed previously, it's not surprising that the BL-budget predictions of LWPc are poor. If the authors agree, perhaps figure 6 could be removed from the paper, but a dashed or dash-dotted line could be added to the right column of figure 8 showing the d(LWPc)/dt prediction based on the BL-mean budget. That would make clear that the prediction is especially poor during the daytime and could be emphasized around the discussion of figure 8. However, if the authors believe that making this point clearly is important for the reader, please show the prediction of d(LWPc)/dt explicitly in the left and right panels of figure 6. Showing the actual and residual does not have the same impact as showing the actual prediction (for me at least).
Fig 8: My understanding is that the budgets in figures five and seven are closed by definition so that the sum of the process tendencies and the actual tendencies are identical. Since the LWPc budgets are not necessarily closed, it would be useful to plot the predicted tendency based on the sum of the individual processes instead of (or alongside) the actual tendency. If figure 6 is removed, both the BL-budget and CV-budget predictions of d(LWPc)/dt could be included in the right column of figure 8.
11/304: I don't understand what is meant by "The warming strengthens the stratification of the sub cloud layer". Is it a change in the surface temperature jump? Is the base of the cloud volume high enough that "sub cloud" includes the transition layer and part of the cloud layer? Does the sub cloud layer actually depart from being well-mixed?
14/401-407: Does decoupling explain part of the correlation of LWPc velocity and z_i, since deeper boundary layers are likely more decoupled?
14/z_i scaling: It's interesting that the z_i scaling works so well, but I found myself wishing it had been more clearly motivated. Why do we have so much faith in the z_i scaling of these budget terms when the BL-budgets did so poorly in predicting d(LWPc)/dt? As an aside, the suggestion above for computing C_ENTR resulted from efforts to understand the relationships between these budget terms better and seeking an alternative to the z_i scaling.
16/section 6.1: See major comment 1a above.
18/538-539: Glassmeier et al (2021, https://doi.org/10.1126/science.abd3980) seem to argue that long timescales are important for LWP adjustments. In such circumstances, the surface flux adjustment might play a role in modifying the steady-state LWP relative to one computed using fixed/prescribed surface fluxes. If it's feasible, the authors could make an estimate of how the inclusion of interactive surface fluxes might have impacted the slope of the predicted d(log LWP)/d(log N) in Glassmeier et al.
===========================
Typographical/rephrasing suggestions (OPTIONAL):
14/404: "... sufficiently high to suppress _cloud base_ precipitation ..."
Citation: https://doi.org/10.5194/egusphere-2024-1033-RC1 -
RC2: 'Comment on egusphere-2024-1033', Anonymous Referee #2, 03 Jul 2024
General comment:
This paper investigates the diurnal behavior of an ensemble of stratocumulus-topped-boundary-layers produced by LES. While the topic is interesting, I find this paper somewhat confusing. In particular, I think it would be beneficial if the authors clarified the main message of the paper, as it currently seems too broad, and if they limited their numerous analyses to the most essential ones.
As this type of investigations definitely helps in broadening our knowledge on the topic, it should be explained early in the paper that the simulated cases are not likely to be observed in nature, especially due the assumption of zero wind, additional non-physical modifications of the surface fluxes calculation, and the assumption of fixed subsidence with varying inversion strength. In my opinion, the most valuable part is the LWP budget, although there seems to be an error in its derivation in addition to some confusing statements. My recommendation is major revision.
My general suggestions are:
- Simplify the message
- Remove the hysteresis plots and discussion
- Review equations for correctness
- Clarify the setup and simplifying assumptions
- Remove surface fluxes analysis
Specific comments:
L2: Clarify how you construct your ensemble (what do you vary?). Provide a reason why it is large. You mention 3 categories first, but then only 2 important categories are discussed: coupled and decoupled layers, which is confusing. It would be helpful to mention that you only look at the statistics.
L5, L6: What do you mean by more coupled and decoupled clouds? Is coupling / decoupling a feature of clouds here?
It would be important to show, at least in one figure, the most representative examples of the ‘coupled’ and ‘decoupled’ STBLs in terms of their vertical structure (mean profiles Thetal and qt profiles, qc, turbulence). It would also help link your analysis with observations more closely.
L10: “The time rate of change in the LWPc is more likely to be negative for higher LWPc and greater inversion base height (zi).” Unclear. Does it suggest that LWP more likely decreases at night for larger LWP?
L12: The sentence about 10h and 15% offset is difficult to understand and seems insignificant. Consider removing it.
L32: Your ‘early works’ include papers that are more recent than ‘recent works’. Please correct.
L37: Explain that these are highly idealized simulations (e.g., with fixed boundary conditions like surface fluxes, for quasi-steady states, etc.). Is it only for different initial conditions? We need to keep a clear distinction between the real world and idealized and simplified simulations.
L43: More important than the number of simulations is that they covered a multi-dimensional space of idealized atmospheric conditions, aiming to represent – but only to some extent - the observed variability of STBLs. Please clarify if that’s the case.
Please explain if this dataset has been validated for the realism of the states it produces.L53: When citing Hoffmann et al. (2023), please specify how idealized their simulations were. This is essential for understanding the strengths and limitations of these analyses. Additionally, please mention Chung et al. (2012), where they derived and proved via simulations the asymptotic values of cloud fraction for a range of steady states driven by various environmental conditions for stratocumulus-to-cumulus transitions.
L74: If it is only similar to Deardorff, what modifications does it include? In particular, how does it calculate horizontal vs vertical turbulent mixing coefficients?
L79: Does this sentence cast doubt on the results from the two papers? Can you show the differences between the two approaches using a simple example (e.g., a 4-hour DYCOMS-II simulation)?
L84: What ‘details’ do you mean here? Do they refer to the differences between the two approaches, or the way the hydrometeors are calculated?
L92: If the initial wind speed is 0 m/s, how realistic are these simulations? It should be highlighted in the abstract that this is an ensemble of no-mean-flow simulations, differentiating them from what we typically observe in the subtropics.
L102: I find this paragraph a bit confusing. The mean flow is 0 m/s, but you add 7 m/s to calculate surface fluxes. Why 7 m/s and not another value? While it is encouraging that you have put a lot of effort into making your environmental conditions consistent with observations, this surface condition, combined with the no-wind condition, appears highly unrealistic. In reality, a non-zero wind is needed to form a logarithmic wind profile, which is a significant source of TKE. You seem to focus on the limit of free convection where the mean flow approaches zero but still need to mimic reasonable surface fluxes. This is a very specific idealization of your setup that needs to be explained in the abstract and introduction, as such cases can never be observed.
L108: You seem to use only one value of the large-scale divergence for the entire ensemble, and it matches that from DYCOMS-II RF01 (Stevens et al. 2005), which needs to be explained. That case was actually characterized by strong subsidence producing large T and q gradients at the top of STBL. Typically, weaker subsidence is associated with weaker inversion. Have you looked into the relationship between inversion strength and subsidence (cf. Wood and Bretherton 2006) to clarify it?
Your ERA5 climatology for SST is 292.4K, whereas Stevens et al. 2005 uses 292.5K. I can understand that difference, although I first thought you followed their study. However, your choice for the 7 m/s wind speed near the surface explained as ERA5 climatology matches one of the geostrophic wind components from Stevens et al. 2005. That is difficult to understand because its magnitude near the surface would likely be a small fraction of the geostrophic wind (Fig. C1 therein). ERA5 seems to be too coarse to provide a credible information about wind magnitude just above the surface.
L109: Explain the reason for using such non-uniform grids (dz/dx=1:20). A 200 m grid length is even more than what is typically used for LES of deep convection. I am quite surprised by this choice because Sc circulations are generally weak and very local. Assuming that effective resolution is typically around 4-6dx (Skamarock et al. 2014), you may not be able to represent the relevant small-scale dynamics here. What is the partitioning between explicit and subgrid TKE for this setup in the cloud and subcloud layers? Is it still LES or more of a gray-zone approach? In your free-convection limit it may give you ~1km size of the smallest eddies. Did you look into that? I would be curious to see some examples of the flow patterns.
L110: If your initial well mixed layer can be as high as 1300m (L95), and your inversion can be very weak or even non-existent (L95) then one can expect the convective layer to get deeper than 2km during daytime. Is it justified to apply a damping layer above 2km? Can you please demonstrate that all the simulated cloud layers have their tops much below 2km? If some of them reach 1.9-2km then I think they should be excluded from the analysis.
L111: What day of the year is it? Are you following Stevens et al. 2005 for the length of day here?
L113: Do your cases precipitate in general? What controls the autoconversion rate? If you only focus on non-precipitating cases, is it because you suppress autoconversion or you only select cases with shallow cloud layers?
L115: I now understand that you actually exclude the cases where convection reaches 2 km. However, if your damping layer starts at 2 km, there is typically an entrainment interfacial layer right above the cloud layer (Haman et al. 2007) that is part of the circulation and needs to be accounted for. My suggestion would be to exclude the cases with cloud top height above 1.9 km or so.
L130: van der Dussen et al. (2013) used a simple flux-based measure of decoupling useful in finding temporary decoupling conditions between the subcloud and cloud layers. How does your index compare to theirs?
Please put your results in the context of Nowak et al. (2021) on coupled and decoupled STBL.
L138: This is a typical diurnal evolution of Sc; see van der Dussen et al. (2013) or Smalley et al. (2024).
L140 and Fig 2c: what is your LWP criterion for loDloL? This category significantly overlaps with loDhiL, which makes it unclear why they are considered as two different categories. My understanding is that analyzing your data in terms of coupled vs decoupled STBLs may help you reach a broader audience.
L163: I don’t think this is a hysteresis, it looks more like a fraction of an open loop. The size of the loop is (not surprisingly) the largest for the largest LWP values, as we typically observe. You seem to prefer analyzing the data in various parameter spaces, which sometimes poses questions on what is the purpose of it. It may be worth considering to just focus on several main pieces of the analysis and reduce complexity of your analysis. I suggest to remove this panel.
In fig 3, what is the envelope of the results for each of the categories? Please add some shading.
Fig 3 c shows that all of the LWP curves basically collapse to similar values at the end of the day (see my comment to L138).
Both Fig. 1 and Fig 3 look at the evolution of LWP but in different parameter spaces. Fig 1. Clearly shows that N is not a control parameter as practically all the trajectories are vertical. Please consider simplifying the message (merging Figs 1 and 3?).
Fig 4. – I suggest to remove it or move it to supplemental material. This may help clarify the message you want to deliver, which I think is on the evolution of the cloud layer (LWP, cf).
L194: Please clarify why you need to use LWPc in your detailed budget analysis? For example, van der Dussen focuses on the LWP tendency for adiabatic Sc layers.
L201-205: Please clarify how your CV approach is “also based” on this observation. Is CV a cloudy fraction of the cloud layer?
Eq.5: Since you already introduced fc, I suggest to use one symbol for area fraction (fc vs f). Consider using z_b rather than z0 to distinguish from the surface. I think it should be Psi(t), M(t), etc.
Eq. 6: Since zi and z0 can evolve, <rho0> doesn't seem to be constant.
L217: Is psi(z,t) the mean profile or is it the mean over CV only?
L220: What derivation did you mean here? Do you follow the derivation for the budget from their paper? If you mean the decomposition, then citing Kazil et al. should take place around L.222.
Eq. 8: This equation doesn’t seem correct. I think the 2nd term should read as -/M^2 dM/dt. Besides, since Psi=Psi(t,z), and M=M(t,z), I think these should be partial derivatives.
Eq. 9 and next equations: they all carry the same mistake as Eq. 8. This puts in question all the analysis presented in this paper.
Also, what do you mean by dM/dt|P? Why would different processes such as lateral entrainment or radiation change the mass of the volume that is fixed for given zo and zi (because rho0 is constant)? Wouldn’t that imply dM/dt=0? Please clarify.For your mixed layer analysis, do you assume anything about your qt and thetal profiles in the cloud layer? Please list your assumptions for clarity.
Fig5-Fig8: Please add the envelopes of model spread to understand how different those tendencies are for your different subsets of cases.
L286: What do you mean by actual? Is it the real tendency from the model? Or is it the sum of all tendencies?
Fig. 11: This caption is confusing. “a few extra terms” doesn’t sound precise.
Section 6.2: You’ve already shown a lot in this paper and to me this part is not necessary.
REFERENCES
Chung, D., G. Matheou, and J. Teixeira, 2012: Steady-State Large-Eddy Simulations to Study the Stratocumulus to Shallow Cumulus Cloud Transition. J. Atmos. Sci., 69, 3264–3276.
Haman, K.E., Malinowski, S.P., Kurowski, M.J., Gerber, H. and Brenguier, J.-L. (2007), Small scale mixing processes at the top of a marine stratocumulus—a case study. Q.J.R. Meteorol. Soc., 133: 213-226.
Nowak, J. L., Siebert, H., Szodry, K.-E., and Malinowski, S. P.: Coupled and decoupled stratocumulus-topped boundary layers: turbulence properties, Atmos. Chem. Phys., 21, 10965–10991, https://doi.org/10.5194/acp-21-10965-2021, 2021.
Skamarock, W. C., S. Park, J. B. Klemp, and C. Snyder, 2014: Atmospheric Kinetic Energy Spectra from Global High-Resolution Nonhydrostatic Simulations. J. Atmos. Sci., 71, 4369–4381.
Smalley, K. M., Lebsock, M. D., & Eastman, R. (2024). Diurnal patterns in the observed cloud liquid water path response to droplet number perturbations. Geophysical Research Letters, 51, e2023GL107323.
Stevens, B., and Coauthors, 2005: Evaluation of Large-Eddy Simulations via Observations of Nocturnal Marine Stratocumulus. Mon. Wea. Rev., 133, 1443–1462.
Wood, R., and C. S. Bretherton, 2006: On the Relationship between Stratiform Low Cloud Cover and Lower-Tropospheric Stability. J. Climate, 19, 6425–6432.
van der Dussen, J. J., S. R. de Roode, A. S. Ackerman, P. N. Blossey, C. S. Bretherton, M. J. Kurowski, A. P. Lock, R. A. J. Neggers, I. Sandu, and A. P. Siebesma (2013), The GASS/EUCLIPSE model intercomparison of the stratocumulus transition as observed during ASTEX: LES results, J. Adv. Model. Earth Syst., 5, 483–499.
Citation: https://doi.org/10.5194/egusphere-2024-1033-RC2 - AC1: 'Comment on egusphere-2024-1033', Yaosheng Chen, 25 Aug 2024
Status: closed
-
RC1: 'Comment on egusphere-2024-1033', Anonymous Referee #1, 04 Jun 2024
Review of "Diurnal evolution of non-precipitating marine stratocumuli in an
LES ensemble" by Chen, Zhang, Hoffman, Yamaguchi, Glassmeier, Zhou and Feingold, Manuscript egusphere-2024-1033Summary:
An ensemble of large eddy simulations of marine stratocumulus clouds is run and analyzed for a variety of idealized summertime conditions over the Northeast Pacific Ocean. Building on earlier work by led by co-authors Hoffman and Glassmeier, the new simulations use a more realistic radiation parameterization, including a diurnal cycle, as well as interactive surface fluxes. The analysis seeks to understand how the diurnal cycle of cloud cover and LWPc (in-cloud liquid water path) are determined by the base state (sorting into bins with high/low decoupling and LWPc) and by different processes in understanding how different processes fix LWPc. Highlights include: daytime reductions in cloud fraction and LWPc are larger in more decoupled boundary layers. Larger free tropospheric humidity tends to support higher LWPc during nighttime.
Assessment:
The paper is well-written and nicely tells a story about how stratocumulus-capped boundary layers vary across the diurnal cycle. The manuscript is interesting and compelling, but I was struck that the paper focuses much more strongly on LWPc than on cloud fraction, which is presumably a stronger control on shortwave cloud radiative effects during the diurnal cycle in many of these simulations. However, I expect that another paper based on these simulations will tell that story later. I think the manuscript is a good fit for ACP and should be published, though I would ask the authors to consider the suggestions I make below along with those of the other reviewers.
Recommendation: Minor Revisions
===========================
Major comment:
1. Regarding the LWPc budget:
1a. For the domain-mean budget, it is natural for the subsidence term to be based on the prescribed, large-scale subsidence. However, for the cloud-volume (CV) budget, I find it puzzling that the mean vertical velocity in cloudy columns was not used. I did not ponder this until the end of section 6.1 (on the uncertainty in the entrainment term). If the mean velocity profile in the cloudy columns was not saved, then keep the analysis as it is, but if not, wouldn't the analysis be more robust if the mean cloudy-column subsidence was used for the SUBS_CV term? Then, the ENTR_CV term would be based on the actual entrainment rate in the cloudy-columns, which might be more closely tied to the radiative cooling there. The authors are free to argue that the current setup is more compelling and/or that the differences are only important in the afternoon when the cloud fraction falls precipitously in some of the simulations, but I wanted to suggest this.
1b. The LWPc budget involves large terms with different signs (e.g., RAD, ENTR, BASE-n-LAT), with the important d(LWPc)/dt trend being the small residual after these large terms almost --- but not exactly --- offset each other. Since it's hard to visualize the result of such cancellation, I found myself seeking some way to understand the budget better. From this, I had two (very optional) suggestions:
- Instead of decomposing the dz_i/dt term into separate w_s and w_e as specified on line 272-273, include the whole d(z_i)/dt term in the budget. This is easier for the reader to understand and avoids another pair of opposing terms in the budget. Helpfully, this term switches sign from night to day in some of the simulations, so that changes in dz_i/dt --- driven here by changes in entrainment --- may play a role in the diurnal cycle of LWPc. Doing this would modify the ENTR term in the LWPc budget, since it would only represent the entrainment impacts on cloud base height through d<qt>/dt_CV and d<thetal>/dt_CV. Only make such a change if the authors find it clarifies the story about the budgets.
- (MORE SPECULATIVE) Re-group the terms, combining exchange within the boundary layer (BASE-n-LAT) with radiation (RAD) since both drive increases in LWPc. Both can also be seen as driving entrainment, both the familiar cloud top cooling through (RAD) and fluxes through cloud base (BASE). Write ENTR = - C_ENTR * (RAD + BASE-n-LAT), where C_ENTR is something like the entrainment drying efficiency of the (RAD + BASE-n-LAT) terms. Then, the budget for LWPc is dominated by SUBS and (1-C_ENTR)*(RAD + BASE-n-LAT), with variations in C_ENTR controlling whether LWPc grows or decays. Looking at the budgets in figure 8, C_ENTR seems roughly constant during the nighttime but varies across the different classes of simulations. I would hypothesize that the differences between C_ENTR across the simulations depends on things like the jumps across the inversion, free tropospheric humidity, decoupling and perhaps other quantities as well, so the differences in the outcomes, e.g., positive/negative d(LWPc)/dt, might be able to explained in terms of those quantities. As noted above, this is pretty speculative, but (if it works) could potentially make the storytelling a bit cleaner.
===========================
Specific/minor comments (10/277 means p. 10, line 277):
1/3: My understanding is that the "cloud liquid water path (LWPc)" refers to the mean cloud liquid water path in cloudy columns. If so, could this be stated explicitly somewhere, even if it is broadly understood by others in the field. Also, please explain somewhere how a "cloudy column" is defined.
4/94: The broad range of initial boundary layer temperatures (10K) suggests that the initial boundary layer is relaxing towards quasi-equilibrium with the SST throughout the day, as suggested by the surface temperature jump in figure 4c. It's great to have such a variety of cases here, but I found myself wondering how such a transients might impact the results. Note also, that the mesoscale organization will also be developing through the nighttime and that this transient might have some impact on the results (e.g., the timing of precipitation development overnight).
10/277-278: The last subsection seemed to be foreshadowing a cloud-volume budget, so it's worth expanding BL and CV here for clarity.
Fig 6/bottom of page 10 (OPTIONAL): Given the degree of decoupling discussed previously, it's not surprising that the BL-budget predictions of LWPc are poor. If the authors agree, perhaps figure 6 could be removed from the paper, but a dashed or dash-dotted line could be added to the right column of figure 8 showing the d(LWPc)/dt prediction based on the BL-mean budget. That would make clear that the prediction is especially poor during the daytime and could be emphasized around the discussion of figure 8. However, if the authors believe that making this point clearly is important for the reader, please show the prediction of d(LWPc)/dt explicitly in the left and right panels of figure 6. Showing the actual and residual does not have the same impact as showing the actual prediction (for me at least).
Fig 8: My understanding is that the budgets in figures five and seven are closed by definition so that the sum of the process tendencies and the actual tendencies are identical. Since the LWPc budgets are not necessarily closed, it would be useful to plot the predicted tendency based on the sum of the individual processes instead of (or alongside) the actual tendency. If figure 6 is removed, both the BL-budget and CV-budget predictions of d(LWPc)/dt could be included in the right column of figure 8.
11/304: I don't understand what is meant by "The warming strengthens the stratification of the sub cloud layer". Is it a change in the surface temperature jump? Is the base of the cloud volume high enough that "sub cloud" includes the transition layer and part of the cloud layer? Does the sub cloud layer actually depart from being well-mixed?
14/401-407: Does decoupling explain part of the correlation of LWPc velocity and z_i, since deeper boundary layers are likely more decoupled?
14/z_i scaling: It's interesting that the z_i scaling works so well, but I found myself wishing it had been more clearly motivated. Why do we have so much faith in the z_i scaling of these budget terms when the BL-budgets did so poorly in predicting d(LWPc)/dt? As an aside, the suggestion above for computing C_ENTR resulted from efforts to understand the relationships between these budget terms better and seeking an alternative to the z_i scaling.
16/section 6.1: See major comment 1a above.
18/538-539: Glassmeier et al (2021, https://doi.org/10.1126/science.abd3980) seem to argue that long timescales are important for LWP adjustments. In such circumstances, the surface flux adjustment might play a role in modifying the steady-state LWP relative to one computed using fixed/prescribed surface fluxes. If it's feasible, the authors could make an estimate of how the inclusion of interactive surface fluxes might have impacted the slope of the predicted d(log LWP)/d(log N) in Glassmeier et al.
===========================
Typographical/rephrasing suggestions (OPTIONAL):
14/404: "... sufficiently high to suppress _cloud base_ precipitation ..."
Citation: https://doi.org/10.5194/egusphere-2024-1033-RC1 -
RC2: 'Comment on egusphere-2024-1033', Anonymous Referee #2, 03 Jul 2024
General comment:
This paper investigates the diurnal behavior of an ensemble of stratocumulus-topped-boundary-layers produced by LES. While the topic is interesting, I find this paper somewhat confusing. In particular, I think it would be beneficial if the authors clarified the main message of the paper, as it currently seems too broad, and if they limited their numerous analyses to the most essential ones.
As this type of investigations definitely helps in broadening our knowledge on the topic, it should be explained early in the paper that the simulated cases are not likely to be observed in nature, especially due the assumption of zero wind, additional non-physical modifications of the surface fluxes calculation, and the assumption of fixed subsidence with varying inversion strength. In my opinion, the most valuable part is the LWP budget, although there seems to be an error in its derivation in addition to some confusing statements. My recommendation is major revision.
My general suggestions are:
- Simplify the message
- Remove the hysteresis plots and discussion
- Review equations for correctness
- Clarify the setup and simplifying assumptions
- Remove surface fluxes analysis
Specific comments:
L2: Clarify how you construct your ensemble (what do you vary?). Provide a reason why it is large. You mention 3 categories first, but then only 2 important categories are discussed: coupled and decoupled layers, which is confusing. It would be helpful to mention that you only look at the statistics.
L5, L6: What do you mean by more coupled and decoupled clouds? Is coupling / decoupling a feature of clouds here?
It would be important to show, at least in one figure, the most representative examples of the ‘coupled’ and ‘decoupled’ STBLs in terms of their vertical structure (mean profiles Thetal and qt profiles, qc, turbulence). It would also help link your analysis with observations more closely.
L10: “The time rate of change in the LWPc is more likely to be negative for higher LWPc and greater inversion base height (zi).” Unclear. Does it suggest that LWP more likely decreases at night for larger LWP?
L12: The sentence about 10h and 15% offset is difficult to understand and seems insignificant. Consider removing it.
L32: Your ‘early works’ include papers that are more recent than ‘recent works’. Please correct.
L37: Explain that these are highly idealized simulations (e.g., with fixed boundary conditions like surface fluxes, for quasi-steady states, etc.). Is it only for different initial conditions? We need to keep a clear distinction between the real world and idealized and simplified simulations.
L43: More important than the number of simulations is that they covered a multi-dimensional space of idealized atmospheric conditions, aiming to represent – but only to some extent - the observed variability of STBLs. Please clarify if that’s the case.
Please explain if this dataset has been validated for the realism of the states it produces.L53: When citing Hoffmann et al. (2023), please specify how idealized their simulations were. This is essential for understanding the strengths and limitations of these analyses. Additionally, please mention Chung et al. (2012), where they derived and proved via simulations the asymptotic values of cloud fraction for a range of steady states driven by various environmental conditions for stratocumulus-to-cumulus transitions.
L74: If it is only similar to Deardorff, what modifications does it include? In particular, how does it calculate horizontal vs vertical turbulent mixing coefficients?
L79: Does this sentence cast doubt on the results from the two papers? Can you show the differences between the two approaches using a simple example (e.g., a 4-hour DYCOMS-II simulation)?
L84: What ‘details’ do you mean here? Do they refer to the differences between the two approaches, or the way the hydrometeors are calculated?
L92: If the initial wind speed is 0 m/s, how realistic are these simulations? It should be highlighted in the abstract that this is an ensemble of no-mean-flow simulations, differentiating them from what we typically observe in the subtropics.
L102: I find this paragraph a bit confusing. The mean flow is 0 m/s, but you add 7 m/s to calculate surface fluxes. Why 7 m/s and not another value? While it is encouraging that you have put a lot of effort into making your environmental conditions consistent with observations, this surface condition, combined with the no-wind condition, appears highly unrealistic. In reality, a non-zero wind is needed to form a logarithmic wind profile, which is a significant source of TKE. You seem to focus on the limit of free convection where the mean flow approaches zero but still need to mimic reasonable surface fluxes. This is a very specific idealization of your setup that needs to be explained in the abstract and introduction, as such cases can never be observed.
L108: You seem to use only one value of the large-scale divergence for the entire ensemble, and it matches that from DYCOMS-II RF01 (Stevens et al. 2005), which needs to be explained. That case was actually characterized by strong subsidence producing large T and q gradients at the top of STBL. Typically, weaker subsidence is associated with weaker inversion. Have you looked into the relationship between inversion strength and subsidence (cf. Wood and Bretherton 2006) to clarify it?
Your ERA5 climatology for SST is 292.4K, whereas Stevens et al. 2005 uses 292.5K. I can understand that difference, although I first thought you followed their study. However, your choice for the 7 m/s wind speed near the surface explained as ERA5 climatology matches one of the geostrophic wind components from Stevens et al. 2005. That is difficult to understand because its magnitude near the surface would likely be a small fraction of the geostrophic wind (Fig. C1 therein). ERA5 seems to be too coarse to provide a credible information about wind magnitude just above the surface.
L109: Explain the reason for using such non-uniform grids (dz/dx=1:20). A 200 m grid length is even more than what is typically used for LES of deep convection. I am quite surprised by this choice because Sc circulations are generally weak and very local. Assuming that effective resolution is typically around 4-6dx (Skamarock et al. 2014), you may not be able to represent the relevant small-scale dynamics here. What is the partitioning between explicit and subgrid TKE for this setup in the cloud and subcloud layers? Is it still LES or more of a gray-zone approach? In your free-convection limit it may give you ~1km size of the smallest eddies. Did you look into that? I would be curious to see some examples of the flow patterns.
L110: If your initial well mixed layer can be as high as 1300m (L95), and your inversion can be very weak or even non-existent (L95) then one can expect the convective layer to get deeper than 2km during daytime. Is it justified to apply a damping layer above 2km? Can you please demonstrate that all the simulated cloud layers have their tops much below 2km? If some of them reach 1.9-2km then I think they should be excluded from the analysis.
L111: What day of the year is it? Are you following Stevens et al. 2005 for the length of day here?
L113: Do your cases precipitate in general? What controls the autoconversion rate? If you only focus on non-precipitating cases, is it because you suppress autoconversion or you only select cases with shallow cloud layers?
L115: I now understand that you actually exclude the cases where convection reaches 2 km. However, if your damping layer starts at 2 km, there is typically an entrainment interfacial layer right above the cloud layer (Haman et al. 2007) that is part of the circulation and needs to be accounted for. My suggestion would be to exclude the cases with cloud top height above 1.9 km or so.
L130: van der Dussen et al. (2013) used a simple flux-based measure of decoupling useful in finding temporary decoupling conditions between the subcloud and cloud layers. How does your index compare to theirs?
Please put your results in the context of Nowak et al. (2021) on coupled and decoupled STBL.
L138: This is a typical diurnal evolution of Sc; see van der Dussen et al. (2013) or Smalley et al. (2024).
L140 and Fig 2c: what is your LWP criterion for loDloL? This category significantly overlaps with loDhiL, which makes it unclear why they are considered as two different categories. My understanding is that analyzing your data in terms of coupled vs decoupled STBLs may help you reach a broader audience.
L163: I don’t think this is a hysteresis, it looks more like a fraction of an open loop. The size of the loop is (not surprisingly) the largest for the largest LWP values, as we typically observe. You seem to prefer analyzing the data in various parameter spaces, which sometimes poses questions on what is the purpose of it. It may be worth considering to just focus on several main pieces of the analysis and reduce complexity of your analysis. I suggest to remove this panel.
In fig 3, what is the envelope of the results for each of the categories? Please add some shading.
Fig 3 c shows that all of the LWP curves basically collapse to similar values at the end of the day (see my comment to L138).
Both Fig. 1 and Fig 3 look at the evolution of LWP but in different parameter spaces. Fig 1. Clearly shows that N is not a control parameter as practically all the trajectories are vertical. Please consider simplifying the message (merging Figs 1 and 3?).
Fig 4. – I suggest to remove it or move it to supplemental material. This may help clarify the message you want to deliver, which I think is on the evolution of the cloud layer (LWP, cf).
L194: Please clarify why you need to use LWPc in your detailed budget analysis? For example, van der Dussen focuses on the LWP tendency for adiabatic Sc layers.
L201-205: Please clarify how your CV approach is “also based” on this observation. Is CV a cloudy fraction of the cloud layer?
Eq.5: Since you already introduced fc, I suggest to use one symbol for area fraction (fc vs f). Consider using z_b rather than z0 to distinguish from the surface. I think it should be Psi(t), M(t), etc.
Eq. 6: Since zi and z0 can evolve, <rho0> doesn't seem to be constant.
L217: Is psi(z,t) the mean profile or is it the mean over CV only?
L220: What derivation did you mean here? Do you follow the derivation for the budget from their paper? If you mean the decomposition, then citing Kazil et al. should take place around L.222.
Eq. 8: This equation doesn’t seem correct. I think the 2nd term should read as -/M^2 dM/dt. Besides, since Psi=Psi(t,z), and M=M(t,z), I think these should be partial derivatives.
Eq. 9 and next equations: they all carry the same mistake as Eq. 8. This puts in question all the analysis presented in this paper.
Also, what do you mean by dM/dt|P? Why would different processes such as lateral entrainment or radiation change the mass of the volume that is fixed for given zo and zi (because rho0 is constant)? Wouldn’t that imply dM/dt=0? Please clarify.For your mixed layer analysis, do you assume anything about your qt and thetal profiles in the cloud layer? Please list your assumptions for clarity.
Fig5-Fig8: Please add the envelopes of model spread to understand how different those tendencies are for your different subsets of cases.
L286: What do you mean by actual? Is it the real tendency from the model? Or is it the sum of all tendencies?
Fig. 11: This caption is confusing. “a few extra terms” doesn’t sound precise.
Section 6.2: You’ve already shown a lot in this paper and to me this part is not necessary.
REFERENCES
Chung, D., G. Matheou, and J. Teixeira, 2012: Steady-State Large-Eddy Simulations to Study the Stratocumulus to Shallow Cumulus Cloud Transition. J. Atmos. Sci., 69, 3264–3276.
Haman, K.E., Malinowski, S.P., Kurowski, M.J., Gerber, H. and Brenguier, J.-L. (2007), Small scale mixing processes at the top of a marine stratocumulus—a case study. Q.J.R. Meteorol. Soc., 133: 213-226.
Nowak, J. L., Siebert, H., Szodry, K.-E., and Malinowski, S. P.: Coupled and decoupled stratocumulus-topped boundary layers: turbulence properties, Atmos. Chem. Phys., 21, 10965–10991, https://doi.org/10.5194/acp-21-10965-2021, 2021.
Skamarock, W. C., S. Park, J. B. Klemp, and C. Snyder, 2014: Atmospheric Kinetic Energy Spectra from Global High-Resolution Nonhydrostatic Simulations. J. Atmos. Sci., 71, 4369–4381.
Smalley, K. M., Lebsock, M. D., & Eastman, R. (2024). Diurnal patterns in the observed cloud liquid water path response to droplet number perturbations. Geophysical Research Letters, 51, e2023GL107323.
Stevens, B., and Coauthors, 2005: Evaluation of Large-Eddy Simulations via Observations of Nocturnal Marine Stratocumulus. Mon. Wea. Rev., 133, 1443–1462.
Wood, R., and C. S. Bretherton, 2006: On the Relationship between Stratiform Low Cloud Cover and Lower-Tropospheric Stability. J. Climate, 19, 6425–6432.
van der Dussen, J. J., S. R. de Roode, A. S. Ackerman, P. N. Blossey, C. S. Bretherton, M. J. Kurowski, A. P. Lock, R. A. J. Neggers, I. Sandu, and A. P. Siebesma (2013), The GASS/EUCLIPSE model intercomparison of the stratocumulus transition as observed during ASTEX: LES results, J. Adv. Model. Earth Syst., 5, 483–499.
Citation: https://doi.org/10.5194/egusphere-2024-1033-RC2 - AC1: 'Comment on egusphere-2024-1033', Yaosheng Chen, 25 Aug 2024
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
360 | 107 | 30 | 497 | 36 | 23 | 17 |
- HTML: 360
- PDF: 107
- XML: 30
- Total: 497
- Supplement: 36
- BibTeX: 23
- EndNote: 17
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1