the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 24 Apr 2024
A climatology of cold pools distinct from background turbulence at the Eastern North Atlantic observations site
Abstract. We develop an algorithm to identify cold pools at the DOE’s Eastern North Atlantic (ENA) site on Graciosa Island and examine the statistics of retrieved cold pools for the entire observational record from late 2016 to 2023. The retrieval strategy relies on leveraging above-background bivariate deviations in near-surface temperature and water vapor mixing ratio from the ENA station time series. Cold pools at ENA tend to be weak with a prominent annual cycle peaking in the cooler months and caused by reductions in the background turbulence during those months. Often, surface rain events are not associated with cold pools due to a combination of factors including but not limited to high background turbulence, high relative humidity, and low rain rate. The retrieval correctly identifies cold pools that are not associated with observed surface rain at the met station. Understanding the factors that lead to the formation of weak cold pools will lead to a greater understanding of the dynamics of the marine boundary layer at ENA and how those dynamics feed back to the cloud morphological structures.
- Preprint
(5280 KB) - Metadata XML
-
Supplement
(741 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-1098', Anonymous Referee #1, 20 May 2024
Review of the article titled “A climatology of cold pools distinct from background turbulence at the Eastern North Atlantic observations site” by Smalley and coauthors for publication in Atmospheric Chemistry and Physics. The authors have used data collected at an ARM site to come up with a new technique for identifying cold pools. The technique applies Singular Value Decomposition (SVD) on bivariate distributions of temporal changes in the surface air temperature and moisture from their background values. The technique is then applied to data collected over several years. The main findings are that the cold pools at the site are weak, peak in the winter months, and tough to distinguish from the background turbulence. The article is overall well-written, and easy to follow. The results will be useful for scientists studying cold pools and those using data from the ARM site. However, the article falls short in many ways as mentioned below. Hence, I recommend this article for publication only after the authors have addressed the concerns listed below.
Major Comments:
The authors have criticized past studies that used a fixed delta-t and/or delta-r threshold to identify cold pools, especially calling out studies of Terai and Wood (2013), Vogel et al. (2021) Willibanks et al. (2015) and Ghate et al. (2020). Although these studies used a fixed threshold for identifying cold pools, they utilized data from several instruments including satellites, cloud radar, lidars etc. So although they might not have used a sophisticated technique to identify cold pools, by utilizing data from these other instruments they restrict their analysis to raining low cloud conditions only. The authors however have only used data from surface met station and rain gage. As cold pools travel away from the rain shaft, there should be at least some rain in the vicinity of the observed cold pools. In addition, possibly due to the imposition of positive delta-q threshold in the proposed technique, some cold pools associated with weak drizzle might have been lost. So, I suggest you either used data from variety of instruments in your technique, or rephrase the sentences in Line 49-71. Please see Zuidema et al. 2017 for further discussion on this topic.
The technique is validated using a single 7-hour case-study and only data from the scanning cloud radar is used. To further solidify the results, I suggest the authors use data from the vertically pointing instruments during few cases and try to understand how things evolved during those. I highly recommend picking up these cases based on the histograms shown in Figure-1 and Figure-4. So few cases spanning from each of the four quadrants would be ideal. Author Jeong’s past paper suggests that the team has height resolved rain properties for several years Thank you.
The main finding of the paper is that most of the cold pools are too weak to be identified from the background turbulence. However, the authors have made very limited attempts on diagnosing the origins of these weak cold pools or the high background turbulence. As it is a marine location, it is expected to have many precipitation induced cold pools. Is the higher number of cold pools during the winter months related to increased precipitation as reported by Wu et al. (2020 J. Climate), Ghate et al. (2021 JAMC), and Lamer et al. (2020 JGR) or they are associated with topography induced drainage flows? Is the increased background turbulence in summer months related to the island heating as reported by Ghate et al. (2021 JAMC). In addition, the winter months also encounter higher number of frontal passages as shown in Figure 12 that is similar to the findings of Ilotoviz et al. (2021 JGR). It seems that your technique only identifies cold pools associated with heavy rain only and hence the finding.
Lastly, I understand that the authors have proposed a new technique for identifying cold pools. But for this technique to be applicable to other studies, and used by other researchers, the authors should make an attempt to put their results in the context of previous work. So for example, if one would have used a fixed delta-T threshold for identifying cold pools, would they have also been able to produce Figures similar to 7 to 11. Or how would the Figure 7-11 would have looked if the cold pools would have been identified from a fixed delta-T threshold.
Minor Comments:
Line 13: DOE is an acronym that needs to be defined, especially as ACP is an international journal. Thank you.
Line 33: Jiang et al. 2021 is not in the references. Please define.
Line 428: probably better to use the phrase boundary layer rather than lower tropospheric.
Line 685: I wonder why the authors chose to cite the campaign report rather than the BAMS article. I suggest citing the BAMS article as it is easy to find and more relevant. Thank you.
Citation: https://doi.org/10.5194/egusphere-2024-1098-RC1 -
RC2: 'Comment on egusphere-2024-1098', Anonymous Referee #2, 27 May 2024
General Comments
The study "A climatology of cold pools distinct from background turbulence at the Eastern North Atlantic observations site" by Smalley and co-authors in consideration for publication in ACP introduces a new method to detect passages of oceanic cold pools in observations of a surface-based meteorological station. The detection algorithm is developed and validated using an 8-year measurement record of the Eastern North Atlantic site on Graciosa Island (Azores) and is based on bivariate fluctuations in temperature-moisture space above a background state that depends on the annual and diurnal cycle. The authors spent a substantial share of the manuscript on describing the concepts and derivation of the method and its distinction from a simpler threshold-based method, followed by a discussion on the properties of the detected cold pools by analyzing various meteorological parameters.
The study introduces a cold pool detection method that utilizes a (to my knowledge) novel concept that is of potential interest for the cold pool and convective organization community. However, the study has several substantial weaknesses, both with respect to the scientific quality and the presentation style, which can be summarized as follows:
- The concept of "background turbulence" is unclear. Although it is of central importance for the study, the authors do not state what spatial and temporal scales this term refers to (especially in the context of typical cold pool sizes and lifetimes) and which processes contribute to it. Furthermore, I am skeptical that the accuracy of the temperature data (0.1 K) allows to reliably distinguish between cold pool signals and background noise and that the temporal resolution of the data (1-min data smoothed over an 11-minute period) is sufficient to capture circulations that are typically relevant for boundary layer turbulence.
- The introduced detection method is insufficiently described. After reading through the methods section several times, it is still unclear to me how exactly the distinction between cold pool and background values in temperature-moisture space is done and what the role of the weight parameter is compared to the semimajor axis of the derived ellipse.
- The benefit of the detection method is unclear. Although the authors correctly acknowledge the lack of a benchmark data set for validation of the cold pool detections, any statements on the performance of the algorithm, especially compared to existing and much simpler methods, remain vague.
- The study lacks integration into the existing literature. It does not state the open scientific problem that it aims at, does not motivate why this problem is relevant, does not discuss the results in the context of previous studies, and does not explore possible implications of the presented work.
- The reader is poorly guided through the manuscript. The authors rarely motivate why a specific analysis is performed and reveal the overall concept of the detection method only at the end of the (lengthy) methodology section.
In the light of the listed deficiencies, I see the manuscript on the borderline between major revisions and rejection, leaning towards the latter. If the editor decides to not reject the manuscript, the authors need to satisfactorily address all of these critical issues in a revised version of the manuscript. More specific comments are listed below.
Specific Comments
- Line 1: I suggest to replace the term "background turbulence" with something like "background variability/fluctuations in temperature and humidity".
- L29: Please add a sentence on the findings of the study.
- L45: The authors should mention that the sign of the moisture signal is more uncertain than for the other variables (as discussed later in the study).
- L47: The "fundamental change in the character of the boundary layer clouds" need to be specified.
- L49: Please state here or elsewhere that this study focuses on oceanic/marine cold pools. There is a whole body of literature on continental cold pools that is not mentioned at all but also relevant for the study.
- L50: Since the study does not specifically exclude strong cold pools or deep convective case from the analysis, it is not fair to say that the study focuses on weak cases. The presented method is rather capable to detect a wider range of cold pool intensities. This also applies to similar statements elsewhere in the manuscript.
- L52: Please explain the term "boundary layer rain". Is it a synonym for weak convection?
- L63-71: Since the term "(background) turbulence" is a central concept for the study, it is essential to properly define it here in the context of the following analyses. Which processes contribute to the background turbulence? What spatial and temporal scales are included? How do these scales compare to typical sizes and life times of (oceanic) cold pools? Clarifying these issues is critical to assess the usability and performance of the introduced method and to discuss the retrieved findings.
- L81: Further describe the measurement location (e.g., distance to coast line) as these information can be relevant for interpreting the observations.
- L89: Based on Supplement Fig. 1a, the instrumental noise of the temperature measurements calculated over 30-min periods is close to 0.1 K/min. For shorter periods we can expect it to be even stronger and, therefore, of substantial magnitude compared to the cold pool temperature signals shown later. The authors should discuss if and to what extent the instrumental noise could impact the presented results.
- L93-96: As correctly stated by the authors, the measurement resolution of 0.1 K is relatively large compared to the studied cold pool signals. However, the applied temporal smoothing does not increase the information content of the data beyond the instrumental limits (even though the numerical values have a resolution below 0.1 K) but rather removes information on the short-term variability.
- L110: Describe how the gaps in the rain measurements are filled.
- L114: What is meant by "very small-scale turbulence"? From a meteorological perspective, I would probably think about spatial scales of 1 m or even below. With a temporal resolution of 1 min and a typical wind speed of 5 m/s, one can resolve flow features with a size of about 300 m, which would be referred to as a "large eddy" in the context of PBL turbulence. This again suggests to avoid the term "turbulence" in this study.
- L119/120: What is the original resolution of the Doppler Lidar data?
- L132: At this point, it is unclear why the following analyses are performed. In the previous sentence, the authors state that rainfall data is not required for the detection algorithm but analyze rain events in the following paragraph.
- Fig. 1: I have several issues with this figure:
- What temperature and moisture signals are shown in Fig.1? Every 1-min signal during the respective rain event or the first one, or the strongest one?
- The data presentation in polar coordinates for panels a to c is quite confusing. Do the color-coded tiles refer to the actual area they cover? I suggest to use a scatter plot or 2D density plot (in Cartesian coordinates) instead.
- It is very hard to compare the results of the panels due the different relative analysis times. It also unclear why these times where chosen.
- L136-137: Another possible explanation for negative moisture signals is the downward transport of drier free-tropospheric air into the surface layer.
- L149: Since the data is smoothed, the time scale is larger than 1 min.
- L149: How many NearRain minutes does the data set contain (compared to FarFromRain minutes)?
- L152-153: The reader should be informed first about what is done rather than what is not.
- L171: This statement is not justified so far.
- L173: The authors should mention that there are several other studies on continental cold pools, that also used threshold-based detection methods often combined with other data sources (e.g., Engerer et al, 2008, MWR; Redl et al, 2015, MWR; Provod et al, 2016, MWR; Kirsch et al, 2021, MWR; Kruse et al, 2022, QJRMS)
- L175: It is unclear how exactly the method of Vogel et al (2021) is applied to the present data set. According to the notation convention used throughout the manuscript, I assume that dT/dt denotes the perturbation of the raw temperature data (as opposed to the smoothed data T'). Moreover, does the threshold need to be continuously exceeded over 20-min periods as in Vogel et al (2021) or only for individual 1-min time steps (which would be very problematic)? The authors have to clarify this.
- L177-182: I understand the statistical argument that the authors make on the cold pool detection rate in rainy vs. rain-free periods, however, I think that it could be strengthened by directly comparing the detection rates of both methods for different periods (since there is no reference data set). Also, it is unclear how large the absolute numbers of (presumably) false detections are compared to the full data set, i.e. how strongly they actually impact the performance of the simple method.
- Fig. 2: The detection thresholds are below or around the measurement resolution of the observations. How large is the impact of instrumental noise in this analysis?
- Fig. 3: Show the time of the day in local time rather than UTC, which eases the interpretation of the results.
- L209-211: This sentence again refers to the method, which was not yet properly described. The reader can only guess how exactly the method actually works.
- L217-225: This paragraph is very hard to follow. For example, it is unclear to me what are eddy depth and eddy strength as opposed to eddy size. Since the described issue seems to be of minor importance, I suggest to delete the paragraph.
- Fig. 4: Please explain the dots shown in panels a and b. I guess they correspond to the ellipses shown in panels c to f. Why were these examples selected?
- L261: What does "greater than the line extending along the ellipse's major axis" mean? Does this mean that the point has to lie outside of the ellipse? How many of the points do actually lie outside of the ellipse (if at all)? Since this is the core of the introduced method, the authors need to be very clear on how it works!
- L270: Explain the term "semimajor axis".
- L275: The authors should explicitly state that the semimajor axis is a measure of the cold pool strength (as this is implicitly assumed later on).
- L279: How do the authors know that these are actual cold pools?
- L280: Again, the authors do not sufficiently motivate why the following analyses are performed.
- L282: How are false alarms identified? What is the reference?
- L286: I suggest to call them "detected" rather than "true" cold pools, since there is no reference data set (as also stated later in the manuscript, L533)
- L290: Is "T" temperature or time?
- L302: What is "a" as opposed to "s"? Both seem to somehow represent the semimajor axis.
- L308-312: It is unclear to me why the weight W is needed (as it appears to correlate with the semimajor axis) and how it is used to identify cold pools.
- L330: This does not become clear from the definition of W in the previous section. Is 1 the maximum value that W can take?
- L338-341: Delete this sentence as the wind analysis does not seem to be of significance for the method.
- L341-343: I don't understand this sentence.
- Fig. 6, top panels: Explain the solid and dashed rings in the plots.
- Fig. 6, bottom panels: I suggest to code the weight as color rather than as line thickness since it is very hard to compare visually.
- L346: Explain the top row first.
- L347: State the time period in the caption.
- L354: Define the period.
- L357-375: Given the unjustified and rather arbitrary choice of thresholds in different variable as well as the overall very small sample size, the presented analysis is not convincing in validating the cold pool detection algorithm.
- L379: Can the weight actually become negative?
- L379-380: Why do the weights sum up to a number of events? I am afraid I still did not get the concept of the weight.
- L385: Does "the entire population of cold pool candidates" refer to the full sample (~200.000 cold pools) or only to the retrieved ones (~8500). Either the statement in L383 is wrong or the wording is misleading.
- L388: Wind gusts are usually defined over 3-second time periods. The used measurements do not have the sufficient resolution to justify statements about gustiness.
- L389-390: Isn't this a trivial statement since a_max measures the perturbation strength in temperature-moisture space?
- L393: How are the cold pool strength categories chosen and what are the respective sample sizes?
- L397: The authors do not motivate why the cold pool size is estimated.
- L408-411: How should the reader interpret this information? How do these numbers compare to the results found by other authors? There is certainly more literature that has already studied the cold size of oceanic cold pools. Why are these number relevant?
- L410: What is the "cord length"?
- L416-475: Figs. 8 to 12 are shown without any motivation and are insufficiently described and discussed. The authors either need to expand this part (see also the following comments) or delete the entire part.
- L418: What is a "more limited time period"?
- Fig. 9: It is unclear to me if this figure shows fraction of cold pool events or cloud types or both.
- L432-433: I would think that the upward motion signature before the arrival of the downdraft is rather caused by lifting ahead of the cold pool front.
- L442: This climatology should be better placed at the beginning of section 3.
- L466-469: I don't understand this sentence.
- Fig. 12: Indicate the data source for this analysis.
- L482-521: It is unclear why all these variables are discussed and what the conclusions are. Also, the results are not discussed in the context of existing literature.
- L493: Further explain the argument about post-frontal scattered showers.
- L494-495: The passage of a cold pool usually leads to an increase in air pressure, meanings that this is not a condition for which cold pools are more likely to occur (as the wording suggests).
- L496: Does "boundary layer motions" refer to wind?
- L510: Define the boundary layer decoupling index.
- L517: The text does not provide any information about the use of balloon measurements.
- Fig. 14: Panels a, b, g, and h are not mentioned in the text.
- L526-532: Such a short summary of the method concept would be required much earlier in the manuscript.
- L531: As stated in the following sentence, there is no "truth" or reference data set.
- L545: How do the authors come to the conclusion of an "overly-strong representation of the cold pool diurnal cycle"? Is there any existing literature that justifies this statement?
- L567: The observed signals in the different variable are rather a consequence of cold pool passages than a condition for their formation.
- L579-583: As already mentioned earlier, the authors miss a discussion on the broader implications of the developed algorithm for the convection and cloud organization community.
- L584: Since the development of the introduced detection method is the key innovation of the study, I strongly recommend to make the code publicly available.
Technical Corrections
- L13: Please explain "DOE"
- L13: Add a note on where Graciosa Island is located (Azores or geographical coordinates)
- L14. Delete "entire"
- L61: Name the Caribbean Island
- L78: I suggest to create a separate section for the data description part.
- L79: Specify that "one-dimensional" refers to time
- L80: Add "United States" to DOE
- L81: Add the coordinates of the measurement location
- L87: Please shortly explain what a "serialized combination" is.
- L88: Change to "before ... and after ..."
- L109: Better call them "rain events" since the term "objects" is usually associated with spatial entities (and stick to the naming convention throughout the manuscript)
- L112: Delete "often"
- L116-117: The sentence is incomplete.
- L117: q is usually used for specific humidity. I suggest to use r for the mixing ration instead.
- L117: Add the missing slash in "dq_v'/dt"
- L121: Define the "short period"
- L122: Which study does "this study" refer to?
- L137: Delete "will"
- L143: I suggest to rename the "FarFromRain" category to "NoRain"
- L153: Change to "weak cold pools"
- L155: Change to something like "rain cells that have formed right above the station "
- L163: Include the shown temperature and moisture perturbations into the caption for Fig. 1a-c.
- L163: Change "probability" to "frequency"
- L164: Please remind the reader of the analysis period ("this period").
- L169: Introduce the "CP" acronym at first use of the term and use it consistently throughout the manuscript.
- L178: Swap "year" and "day" (so that Fig. 2a is referenced before Fig. 2b)
- Fig. 2: Change "expected" to "detected" (in y label)
- Fig. 2b: Use local time instead of UTC
- L195: Change "desire" to "aim" or "goal"
- L208: Avoid the word "clear" in this context.
- L266/267: Delete "of course"
- L273-274: Move this information to the respective figure caption.
- L276-277: Change to "sharply decreases beyond that value"
- L285: Add "(Eq. 1)"
- L353: meteorological
- L354: Add "azimuth angle"
- L406: Supplementary Figure 3
- L418: Stick to one date format.
- L425: The period stated here is different from the text (L418).
- L443: Delete "strong" and "weaker"
- L444: Replace "cooler months" with "winter months"
- L460: Replace "UTC" with "time of day"
- L529: Add "ellipse in temperature-moisture space"
- L568: strong 1-hour
- L579: change "small" to "weak"
Citation: https://doi.org/10.5194/egusphere-2024-1098-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 336 | 98 | 31 | 465 | 44 | 44 | 40 |
- HTML: 336
- PDF: 98
- XML: 31
- Total: 465
- Supplement: 44
- BibTeX: 44
- EndNote: 40
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 219 | 47 |
| China | 2 | 60 | 12 |
| Germany | 3 | 58 | 12 |
| France | 4 | 17 | 3 |
| United Kingdom | 5 | 12 | 2 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 219