the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 Apr 2023
On the relationship between mesoscale cellular convection and meteorological forcing: Comparing the Southern Ocean against the North Pacific
Steven T. Siems
Yi Huang
Luis Ackermann
Abstract. Marine atmospheric boundary layer (MABL) clouds cover vast areas over the ocean and have important radiative effects on the Earth’s climate system. These radiative effects are known to be sensitive to the local organization, or structure, of the mesoscale cellular convection (MCC). A convolution neural network model is used to identify the two ideal classes of MCC clouds, namely open and closed, over the Southern Ocean (SO) and Northwest Pacific (NP) from high-frequency geostationary Himawari-8 satellite observations. The results of the climatology show that MCC clouds are roughly distributed over the midlatitude storm tracks for both hemispheres, with peaks poleward of the 40° latitude. Open MCC clouds are more prevalent than closed MCC in both regions. An examination of meteorological forcing associated with open and closed MCC clouds is conducted to illustrate the influence of large-scale meteorological conditions. We establish the importance of the Kuroshio western boundary current in the spatial coverage of open and closed MCC across the NP, presumably through the supply of strong heat and moisture fluxes during marine cold air outbreaks events. For both regions, closed MCC cloud are more frequent at higher static stability than on air-sea temperature difference, opposite to the open MCC cloud behavior. The diurnal cycle reveals a pronounced daily cycle in the frequency of occurrence of closed MCC over the SO, while the NP closed MCC daily cycle is less noticeable.
- Preprint
(7488 KB) - Metadata XML
- BibTeX
- EndNote
Francisco Lang et al.
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2023-518', Anonymous Referee #1, 29 Apr 2023
Review report
The authors investigate the connection between meteorological parameters and the morphology of low-level clouds in two regions: a portion of the Southern Ocean and the North Pacific. To identify three morphological classes (i.e., “closen”, “open”, and “other”) the authors apply a neural network that was prepared in a previous publication (Lang et al., 2022) to radiance fields from the geostationary HIMAWARI satellite. The article documents the cooccurrence of class frequency with average estimated inversion strength (EIS) and a marine cold-air outbreak index, M, mainly finding that closed cells are connected to greater EIS.
The article contributes to a topic that is of great relevance to the climate community. The authors importantly leverage imagery from a geostationary platform to identify cloud morphology on a great temporal resolution – a scientific advancement that was achieved in Lang et al. (2022). With respect to the previous paper, the present article falls short of making a substantial contribution. Below, I’m outlining major issues that lead me to recommend a rejection but encourage a resubmission in improved form. In a nutshell, the authors should (1) include statistics from the class “other”, (2) quantify their results, (3) contextualize the selected meteorology parameters and possibly widen the parameter space, and (4) inspect EIS throughout the text, table, and figures.
Major issues
(1) The classification resulted in peak frequencies of ~20% for “open cells” and ~15% for “closed cells”, leaving at least ~65% of the samples classified as “other”. The authors explain that “other” constitute cases of mid- and high-level clouds, but also low-level clouds that were different enough from the first two classes (ll. 94-96). Figure 1 shows a classification example in which apparent open cellular cloud decks (e.g., ~35degN, ~160degE) were classified neither as “open” nor as “closed” (leading me to believe they were joined “other” which is not shown here). Given that “other” makes up the bulk of the samples and that the nature of these samples is unknown to the reader, the authors need to (1) provide information on them, showing, for example, statistics of cloud-top height, cloud fraction, and cloud optical depth of “other” compared to “open” and “closed” and (2) include “other” where currently only “closed” and “open” is shown. Both aspects will become highly relevant for the diurnal cycle that shows a decrease of “closed” in the afternoon but no increase of “open” at the same time, leading to believe that the class “other” increased in the afternoon and begging the question why that is (e.g., truly a transition from “closed” to “disorganized” or an increase in local cirrus cloud fraction?).
(2) The authors compare class frequency and meteorological fields and often speak of “correlation” (e.g., ll. 161-164, l. 194, ll. 254-255) or even “high correlation” (ll. 267-268) when visually comparing maps. However, there is no actual calculation of correlation. The paper would benefit from a more quantitative rather than qualitative analysis. Given the great temporal resolution, the authors could not only compare seasonally averaged maps but actually compute correlation on a pixel-by-pixel basis using the native 10-min resolution.
(3) The authors selected EIS and the M index as meteorological parameters, and it is unclear why they selected these two and why the analysis excludes other parameters. For example, near-surface wind speed was recently connected to cloud morphological transitions in Eastman et al. (2023). Also, the authors examine the diurnal cycle of class frequency but leave out diurnal cycles of meteorological parameters that were used for the seasonal analysis a few pages before.
(4) The authors equate EIS with static stability in this paper (e.g., l. 11, 164), even though it was introduced as an indicator of which (ll. 110-111). Given that sea-surface temperature can exceed lower-tropospheric air temperature (giving positive M but possibly negative LTS), EIS can become negative, as the authors show in Figure 4. The authors should explain what a negative EIS means and whether EIS still holds as an indicator of static stability in cold-air outbreak conditions. In addition to revising EIS language, there seem to be inconsistencies across Figure 4, Figure 2, and Table 1: Figure 4 shows mostly negative EIS, yet in Figure 2 and Table 1 EIS is shown to be positive. The authors need to explain why that is (e.g., do composites in Figure 2 and statistics in Table 1 cover different time frames compared to Figure 4?).
Minor issues
ll. 10-11 This sentence is hard to understand. Please rephrase.
ll. 23-26 Please also cite a recent paper by McCoy et al., (2023).
ll. 26-27 This sentence seems out of context.
ll. 95 Please clarify whether low-level clouds can be categorized as “other”.
Fig. 1 “Other” is included in the legend, but not shown in the map. Perhaps selected a color different from white.
Fig. 1 Please label the size of each box or add a reference bar to show distance.
Fig. 2 Please change label from “delta T (K)” to “EIS (K)”.
ll. 131ff. Would NP have great M values even when the Kuroshio current was absent? Also please quantify SST gradient.
ll. 141-142 To follow this sentence, please explain how cloud cover is related to both classes.
l. 145 Please substitute “total time” by “total number of observations”.
Fig. 3 Please mark maximum frequency to complement the text.
Fig. 3 Please draw Kuroshio current in here.
ll. 150-151 (and also ll. 154-155 and ll. 174-175) It is unclear which class is referred to in this sentence.
l. 157 Please substitute “to the top” with “near the top”.
l. 160 Please substitute “SST gradient” with “average SST”.
l. 163 Please finish sentence; “as highly … as” is missing an object.
l. 164 Please refer to EIS as an indicator of static stability rather than equating both (see above major points).
ll. 159-175 Please provide correlation maps (see above major points).
l. 167 Please change “44” to “4”.
ll. 171-172 Please check for redundancy is both sentences.
ll. 181-183 Does this mean the portion of “other” has increased?
ll. 196-198 Please elaborate – currently unclear how land masses explain seasonality.
ll. 196-201 Perhaps a sketch or cartoon image would be useful here to illustrate the point.
ll. 200-201 Please rephrase. Currently hard to follow. Substitute “that” with “than”.
ll. 199-200 Are lower SST connected to a greater frequency of open MCCs?
Fig. 7 Please explain how and why these regions were selected. Is the diurnal cycle of “closed” similarly and oppositely seen in “other” (given that “open” remains unchanged)?
Fig. 7 Is there a diurnal cycle in EIS or M?
ll. 231-233 Please rephrase, perhaps to “The geolocation of cloud types matches those in other studies”.
l. 233 With respect to “most prevalent”, isn’t the “other” class even greater in proportion?
l. 235 Please check for redundancy (i.e., a similar content in sentence before).
l. 239 Perhaps is it worth noting in Section 2 how single and multi-layer clouds are handled in this study.
l. 241 Unclear whether a specific type is being referred to or not.
l. 243 Please check position of parenthesis here. Perhaps rephrase to “turbulent heat and moisture (i.e., sensible plus latent heat) fluxes”.
l. 244 Perhaps better “frequency of occurrence” instead of “presence”.
l. 246 Please rephrase sentence for better understanding.
ll. 258-259 Please elaborate on “showing a strong relationship to the SST gradient”. Is SST gradient shown anywhere? How was a “strong relationship” detected?
l. 264 Please elaborate on “well-formed open cells”. Would not-so-well-formed open cells be problematic for the classification?
l. 268 Please quantify “high correlation” (see above major points).
l. 278 Please elaborate why other techniques and other classes are needed.
ll. 280-281 Please briefly explain whether HIMAWARI has a precipitation product.
References
Lang, F., Ackermann, L., Huang, Y., Truong, S. C. H., Siems, S. T., and Manton, M. J.: A climatology of open and closed mesoscale cellular convection over the Southern Ocean derived from Himawari-8 observations, Atmos. Chem. Phys., 22, 2135–2152, https://doi.org/10.5194/acp-22-2135-2022, 2022.
Eastman, R., McCoy, I. L., & Wood, R. (2022). Wind, rain, and the closed to open cell transition in subtropical marine stratocumulus. Journal of Geophysical Research: Atmospheres, 127, e2022JD036795. https://doi.org/10.1029/2022JD036795
McCoy, I. L., McCoy, D. T., Wood, R., Zuidema, P., & Bender, F. A.-M. (2023). The role of mesoscale cloud morphology in the shortwave cloud feedback. Geophysical Research Letters, 50, e2022GL101042. https://doi.org/10.1029/2022GL101042
Citation: https://doi.org/10.5194/egusphere-2023-518-RC1 -
AC1: 'Reply on RC1', Francisco Lang, 17 Aug 2023
We thank Reviewer 1 for their helpful comments, which have improved this paper. Please find our responses to your comments in “Ref1_Comments_Answered.pdf”.
In the next few days, we will submit both the original (old) manuscript and the revised (new) version, with highlighted differences.
-
AC1: 'Reply on RC1', Francisco Lang, 17 Aug 2023
-
RC2: 'Comment on egusphere-2023-518', Anonymous Referee #2, 02 May 2023
Summary:
Lang et al. present an expansion of their earlier study, Lang et al. 2022, where they used a convolutional neural network to identify open and closed mesoscale cellular convective clouds in the Southern Ocean. This study adds clouds in an additional region, the North Pacific, that have been identified with their algorithm (applied to the geostationary Himawari satellite and using brightness temperature, which enables identifications over the full diurnal cycle). Maps of cloud occurrence frequency are contrasted with maps of stability metrics in the two regions annually and seasonally. Visual relationships to regional environmental factors (e.g., Kuroshio current, oceanic polar front, storm track) are documented. The diurnal cycle is also presented for these regions annually and seasonally. Differences in behavior in the North Pacific are qualitatively documented and contrasted with the previously published Southern Ocean behaviors.
General comments:
The premise of this study is exciting, and the data developed as part of this and Lang et al. 2022 is quite valuable. It is especially noteworthy and novel to examine the diurnal cycles of MCC cloud types in these two hemispheres. The figures developed in this analysis are very well done, clear and compelling. However, the analysis is limited to qualitatively documenting the behaviors in these regions along with their visual correspondence to stability metrics and sea surface temperature. This provides some insights about regional differences but without quantitative analysis the conclusions are limited, similar to those presented in previous studies, and ultimately not as substantive as they have the potential to be. However, I think the authors can develop this analysis into a valuable contribution to the field and fully realize the potential of this work.
- My main recommendation is to add quantitative comparisons to bolster the qualitative comparisons and help with interpreting/establishing the differences between the NP and SO regions. “Correlations” are currently discussed but they are based on visual comparisons and not calculated/provided. Actual correlations, between occurrence frequency and meteorological variables, could be calculated at many scales (e.g., within spatial map grid boxes, for annual and seasonal relationships, for composite differences, for diurnal cycles, etc.). By quantifying the relationships that you suggest here, you would greatly strengthen your results and better support your conclusions.
- The diurnal cycle analysis in this and Lang et al. 2022 is novel and has a lot of potential. However, these results are currently limited to qualitatively documenting the differences between type, season, and region. There is an opportunity here to add more depth to the analysis by quantifying the connections to the meteorological environment (as you do qualitatively in the first part of the paper). This would lead to a deeper understanding of what is contributing to these diurnal cycle differences through understanding how these cloud types are responding to their environmental diurnal cycles. Being able to interpret why you see differences between regions, seasons, etc. in the MCC diurnal development cycle would be a very valuable contribution to the field.
Specific Comments:
Throughout: Please only discuss correlations or variables being “correlated” when you have computed a correlation coefficient and statistically tested whether they are correlated (e.g., p-value ≤ 0.01 for 90% confidence). Visually similar maps and cycle plots are not correlations.
Line 35: I would suggest removing “are most common in” since sub-tropical decks also have a lot of MCC. Agreed that these MCC types dominate the storm tracks (also see Agee et al. 1973, McCoy et al. 2023 for climatology).
Line 37: Fletcher et al. 2016 is for clouds in general, not MCC. Atkinson and Zhang 1997 and Wood 2012 review this MCC-CAO relationship in detail and McCoy et al. 2017 quantified it more recently.
Section 2.2: I see the rational of only identifying open and closed MCC and throwing everything else into a catch all since it gives you a high quality MCC dataset. However, I do think it would be valuable to at least subdivide the “other” into low, middle, and high clouds and clear sky so that you have an idea of what is happening with the other low clouds (besides the MCC types) in these regions. From the small MCC absolute frequencies that you are working with, there is clearly a lot of the low-cloud behavior that you are missing throughout your analysis and analyzing that could give you valuable context for whether the MCC behaviors are unique and whether their differences are statistically significant from the base behavior.
Section 2.3: It might be beneficial to expand beyond stability metrics (EIS and M) and SST. Clouds in this region respond to a variety of factors in opposing ways (e.g., Scott et al. 2020) and MCC are thought to be sensitive to more than just stability and SST in their development (e.g., Eastman et al. 2021, 2022). Temperature advection might be especially useful as it would more accurately characterize the surface forcing contribution in these regions. Expanding your meteorological variable space has the potential to quantify novel relationships between MCC (this is a strong dataset for doing this) and the characteristics of these regional environments and could help to better distinguish your analysis from previous work on MCC behavior in these regions (e.g., Muhlbauer et al. 2014, McCoy et al. 2017, Lang et al. 2022).
Line 134, 158 and Figure 2: Since you use the oceanic polar front as a reference for discussing cloud behavior, it would be useful to have the corresponding annual/seasonal climatological location of the oceanic polar front plotted on these maps. Otherwise, consider citing literature to support these statements and your conclusions.
Line 164: It would be very valuable to spatially correlate these figures, as you imply here, and show the results (e.g., as a map of correlation coefficients with significance indicated). As mentioned above, please do not refer to something as “correlated” unless you are providing a correlation coefficient.
Figure 4 and 5: It might be clearer to see how the regions differ by looking at a difference plot between the NP and SO cases (since you have the data composited to the same space already). You could also consider looking at how the “other-low cloud” category behaves in this space and how it differs from open and closed MCC (e.g., how anomalous the MCC types are from all low clouds).
Line 175: How is the relationship “better”? Please quantify these statements with correlations or other statistics.
Line 188-189: Like with the oceanic polar front, it would be helpful to include a corresponding annual/seasonal climatological storm track to show the relationship with the storm track you suggest here. You could also correlate the location with the occurrence frequency and quantify this. Otherwise, please reference literature supporting this statement about the storm track shift and your conclusions.
Figure 6: Worth noting somewhere in the text what Figure 6 is showing and adding to the story (it is only mentioned in passing, not explained).
Line 192-193: It would help for these types of comparisons if they were quantified. How do you know this is “more relevant”, hard to know that from visually comparing the plots. Consider checking the regressions of frequency on these variables (e.g., in a multiple linear regression that accounts for correlations between predictor variables) and looking at spatial correlation maps.
Line 194 (and throughout): You can say it “corresponds” instead of “correlated”, but it would be much better to calculate the coefficients and quantify this (see above comments).
Line 199-200: Why does the cooler SST explain the higher open MCC frequency? Please explain and support statement.
Line 200-201: What do you mean here? Please explain and support statement.
Line 206, 215-216, 217: These are very exciting results. Would you be able to extend your meteorological factor analysis to this diurnal cycle analysis as well (something like Vial et al. 2021)? This could really help you to begin interpreting what factors might be driving these cycles (and why they are different between regions).
Line 232-233: Also MODIS (Muhlbauer et al. 2014, McCoy et al. 2017, McCoy et al. 2023).
Line 249-251, 263-264, abstract: It seems that you are referring to M as if it is surface forcing or an air-sea temperature difference and contrasting it against static stability as measured by EIS. This is confusing since M is also a stability estimate (essentially a modified form of LTS). This is also inconsistent with your earlier discussions. M can be written as a function of EIS and the air-sea temperature difference (McCoy et al. 2017), is that what you are referencing? Please clarify your meaning and be more accurate in your language.
Line 269: Please explain how you come to this conclusion. Hard to tell from comparing Figure 2 and 5, is this based on a different analysis?
Line 272-276: Great to include the discussion on lines 272-274 of why you have these diurnal cycle differences. It would be very valuable to extend this further to the intriguing regional differences you document (i.e. to add interpretation to Lines 274-276).
Line 277: How are you quantifying “good performance” here? You previously tested and trained on the SO data (and that is presented well in Lang et al. 2022), are you able to similarly check the accuracy for the NP? Or is this from visual inspection?
Line 280-281: Extending this diurnal analysis, either here or in a future paper, would be fantastic. Your results raise so many interesting questions: why are closed MCC NP cycles smaller than SO? Why are they especially small in the summertime? Why are the closed MCC cycles peaking in magnitude in different seasons in the two regions? Why are open MCC cycles much smaller and peaking at different times? What is happening with the remaining contribution of clouds (your “other” type)? You could make a good start at answering these by quantifying the cycle relationships to the meteorological variables you discussed in the first part of the paper.
Technical Corrections:
Line 158: “but is relatively”
Line 167: “Figure 4”
Line 185: “considerably”
Line 246: “current in the Kuroshio region”
Line 248: “MCCs than in the SO”
References:
Agee, E.M., Chen, T.S., Dowell, K.E., 1973. Review of Mesoscale Cellular Convection. Bull. Amer. Meteorol. Soc. 54, 1004–1012. https://doi.org/10.1175/1520-0477(1973)054<1004:aromcc>2.0.co;2
Atkinson, B.W., Zhang, J.W., 1996. Mesoscale shallow convection in the atmosphere. Reviews of Geophysics 34, 403–431. https://doi.org/10.1029/96rg02623
Eastman, R., McCoy, I.L., Wood, R., 2022. Wind, Rain, and the Closed to Open Cell Transition in Subtropical Marine Stratocumulus. JGR Atmospheres 127. https://doi.org/10.1029/2022JD036795
Eastman, R., McCoy, I.L., Wood, R., 2021. Environmental and Internal Controls on Lagrangian Transitions from Closed Cell Mesoscale Cellular Convection over Subtropical Oceans. Journal of the Atmospheric Sciences 78, 2367–2383. https://doi.org/10.1175/Jas-D-20-0277.1
McCoy, I.L., McCoy, D.T., Wood, R., Zuidema, P., Bender, F.A. ‐M., 2023. The Role of Mesoscale Cloud Morphology in the Shortwave Cloud Feedback. Geophysical Research Letters 50. https://doi.org/10.1029/2022GL101042
Scott, R.C., Myers, T.A., Norris, J.R., Zelinka, M.D., Klein, S.A., Sun, M., Doelling, D.R., 2020. Observed Sensitivity of Low-Cloud Radiative Effects to Meteorological Perturbations over the Global Oceans. Journal of Climate 33, 7717–7734. https://doi.org/10.1175/jcli-d-19-1028.1
Vial, J., Vogel, R., Schulz, H., 2021. On the daily cycle of mesoscale cloud organization in the winter trades. Quarterly Journal of the Royal Meteorological Society. https://doi.org/10.1002/qj.4103
Wood, R., 2012. Stratocumulus Clouds. Mon. Weather Rev. 140, 2373–2423. https://doi.org/10.1175/mwr-d-11-00121.1
Citation: https://doi.org/10.5194/egusphere-2023-518-RC2 -
AC2: 'Reply on RC2', Francisco Lang, 17 Aug 2023
We thank Reviewer 2 for their helpful comments, which have improved this paper. Please find our responses to your comments in “Ref2_Comments_Answered.pdf”.
In the next few days, we will submit both the original (old) manuscript and the revised (new) version, with highlighted differences.
Francisco Lang et al.
Francisco Lang et al.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 298 | 105 | 16 | 419 | 12 | 11 |
- HTML: 298
- PDF: 105
- XML: 16
- Total: 419
- BibTeX: 12
- EndNote: 11
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1