Accuracy of Balloon Trajectory Forecasts in the Lower Stratosphere


Round 1
Reviewer 1 Report
Dharmalingam et al. present an analysis of the pre-Concordiasi and Concordiasi superpressure balloon trajectories compared to predicted trajectories using several ECMWF model variations. They find that analysis and even forecast products describe the Concordiasi (polar) flights well. However, tropical trajectories are very poor quality unless the balloon motion itself is ingested into the model. They provide an analysis of the reasons for this, its spatial occurrence, and its temporal span. The study's findings are discussed in light with Strateole plans to perform near-simultaneous ground radiosonde launches as the superpressure balloons fly overhead; a coordination that will require accurate trajectory forecasts several days into the future.
This is a well written paper and provides a very tangible demonstration of the difficulty in characterizing lower stratosphere winds in the tropics. I believe it will be of interest to the readership of this journal. I have only minor comments.
There are minor English usage and spelling issues throughout the paper.
Lines 145-148: I did not understand what was being described here. It seems to be a highly technical discussion of how observations are ingested into a model, and the subtleties of using dense observations where models expect much sparser ones. I would appreciate a citation here that can point the reader to a discussion (book or journal article, etc.) of how observations are ingested into models. This would assist readers, such as myself, who are not terribly familiar with how observations are incorporated.
Lines 191-198:
This paragraph is unclear to me. I think the authors mean that "Y percent chance that the simulated balloon will be within X km of the real balloon." Instead, "real balloon" is repeated. Further confusing the issue is that the polar balloons go in circles, so it could even be that the simulated balloon loops around and ends up being near the real balloon! I request the authors clarify this paragraph.
Line 194: "Significantly depart from gaussianity". The authors make repeated statements about "gaussianity" throughout the paper. It is unclear to me if they mean "it looks gaussian" or "we did statistical tests that show that the distribution is/is not gaussian." The word "significant" has a specific meaning in statistics as well. I would prefer the authors specify whether they are making a statistical claim: this distribution is/is not gaussian at X confidence, or say that the distribution "appears gaussian" if they are not intending to quantify their assertion.
Figure 5 and elsewhere: It would be easier if the authors used solely spherical distance throughout the paper. This would make it more simple to compare the Concordiasi and pre-Concordiasi model error. Also, PDFs of longitudes are unlikely to be strictly Gaussian since the absolute distance between longitude units varies poleward.
Line 236: What does "mere persistence" mean in this context? I assume it means "assume the balloon keeps going straight" but a precise definition would help.
Lines 283-290: I assume the authors are familiar with the GHOST program, a 1960s era superpressure balloon campaign that aimed to use these drifting platforms to continuously monitor UTLS winds. The authors' study shows that the need for such observations continues to the present day. It may be worth mentioning the GHOST program, see
https://www.eol.ucar.edu/field_projects/ghost-balloon
Figure 9: I don't understand the line "Horizontal lines are plotted every degree on the vertical axis". Are these latitude lines?
Lines 362-370: I agree that the errors are highly correlated. The authors should consider modeling this explicitly (a simple linear regression in log/log space should be sufficient). Then, predictions of error 24 hours in the future could be made based on observations of present error during the Strateole-2 campaign. This seems to be a useful ability given the concern about predicting balloon overflights.
Lines 429-430: This is very surprising given the balloon trajectory results. If anything, the air parcel trajectories should be worse. Are the authors confident in these "encouraging results?" I would think the results of the present paper cast doubt on the conclusions in Refs. 14 and 15 as well as the statements by Legras and Bucci.
Author Response
We wish to thank the reviewer for his or her careful reading of the manuscript and for constructive comments and suggestions. We have done our best to follow the suggestions, accordingly we modified the manuscript. It can be mentioned that some of them are retained, those are explained why we didn't change it. We believe their comments and suggestions have improved the manuscript, and once again we thank the reviewer for their careful assessment.
Author Response File: Author Response.pdf
Reviewer 2 Report
This paper compares calculated trajectories with balloon trajectories in the lower stratosphere. Since there are few such balloon flights available, this is certainly an interesting opportunity for model evaluation. The paper is well written and I think it could eventually be published. However, I have quite a number of major comments on it that highlight important weaknesses of the paper. These should be worked on to make the paper suitable for publication.
Major points:
The abstract is very qualitative (e.g. line 9, “The errors in these trajectories are large”). It should contain concrete and quantitative statements. To some extent, this is true of the entire paper, which contains relatively little quantitative analysis (see also other comment below).
This paper deals with the accuracy of trajectory calculations, yet there is hardly any reference to literature on trajectory models, trajectory accuracy, methods to determine it, etc. I think the authors are missing out on a lot of background to these topics, which could put the present study in a broader context. See, e.g., the review by Stohl (1998). Even very close to the researchers’ topic, there are lots of studies that have used balloons for trajectory model validation both in the troposphere and stratosphere, but nothing is cited, not even a paper that uses Concordiasi balloons (de la Camara et al., 2013).
The paper evaluates calculated 24-hour trajectories. However, no motivation is given why exactly 24-hour trajectories are used. The balloon flights are very long, giving the rare opportunity to evaluate long-duration trajectories, so to me it is not immediately clear why the calculated trajectories should be so short. I see that the tropical trajectories are quite inaccurate, so one may argue that longer trajectories make little sense. But this is neither explored, nor discussed.
Forecast data were also used to calculate trajectories. However, no details are given about how the forecasts were used. For example, were they initialized exactly at the trajectory starting time, or 12 hours earlier (e.g., to give time for data transfer, etc., which would be needed in an operational setting)? Were continuous forecasts used per 24-hour interval, or e.g., always a 24-hour forecast? Generally, I think it is also not mentioned at which time intervals meteorological data were used for trajectory calculations.
The trajectories are calculated at constant density levels, whereas the balloons certainly vary in height to some extent. The authors say that vertical resolution is extremely important to obtain high-quality trajectories. Now, how does this fit with not matching the exact balloon height? This certainly must be a substantial source of error (see also Baumann and Stohl, 1997).
Figures 3, 4 and 7: These figures are rather unclear. The colors for prec-01 and CTL-2016 are nearly the same on my print-out, so it is hard to see which one is which. It is also unclear where trajectories start and where they end. I would suggest to place markers at the starting locations.
Line 366: “variables are highly correlated”: Again, this is a qualitative statement, which needs to be replaced with a quantitative one. What is the correlation coefficient? There looks to be a lot of scatter, especially when considering that this is a logarithmic plot, which hides most of the scatter.
Paragraph starting with line 408: “We have shown…”: Your explanation of regional differences sounds logical, but I don’t think you have shown this. These are just a few cases and I think they are not representative enough to make firm statements about regional differences. At least you would have to make a statistical test of the differences and make sure the cases are independent. Otherwise you can suggest an explanation but you cannot claim that you have shown it. It partly comes back to the observation above that the paper lacks quantitative rigorosity.
Minor points:
Line 150: What is a “B matrix”?
Line 195: Do you mean within 160 km of the calculated balloon position?
Line 222: “Only little deviation…”: To me it looks like the trajectory length is quite different!
Line 236: “It is suggestive that mere persistence … would be more informative”. Is it more informative or not? This can be checked and a proper affirmative statement can be made.
Language, typos:
Line 61: and the and (superfluous and)
Lines 127+134: level -> levelS; on line 134, the second “levels” is superfluous.
Line 155: balloons -> balloon data
Caption to figure 3, last line: last 3 days instead of lasts 3 days (plural)
Line 260: has -> have (plural)
Line 284: defficiencies -> deficiencies
Line 318: trajectories IN the tropical UTLS (IN missing)
References:
Baumann, K., and A. Stohl (1997): Validation of a long-range trajectory model using gas balloon tracks from the Gordon Bennett Cup 95. J. Appl. Meteor. 36, 711-720.
De la Camara, A. et al. (2013): Isentropic Transport within the Antarctic Polar-Night Vortex: Rossby Wave Breaking Evidence and Lagrangian Structures, J. Atmos. Sci. 70, 2982-3001, DOI: 10.1175/JAS-D-12-0274.1.
Stohl, A. (1998): Computation, accuracy and applications of trajectories - a review and bibliography. Atmos. Environ. 32, 947-966.
Author Response
We wish to thank the reviewer for his or her careful reading of the manuscript and for constructive comments and suggestions. We have done our best to follow the suggestions, accordingly we modified the manuscript. It can be mentioned that some of them are retained, those are explained why we didn't change it. We believe their comments and suggestions have improved the manuscript, and once again we thank the reviewer for their careful assessment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Thank you for your careful responses to my comments. My concerns are satisfied and I recommend that the article be published in its present form.
Reviewer 2 Report
The authors have addressed most of my previous concerns. I would recommend publication of this revised manuscript.