the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 22 Sep 2021
Fate of sea ice in the 'New Arctic': A database of daily Lagrangian Arctic sea ice parcel drift tracks with coincident ice and atmospheric conditions
Abstract. Since the early 2000s, sea ice has experienced an increased rate of decline in thickness and extent and transitioned to a seasonal ice cover. This shift to thinner, seasonal ice in the 'New Arctic' is accompanied by a reshuffling of energy flows at the surface. Understanding the magnitude and nature of this reshuffling and the feedbacks therein remains limited. A novel database is presented that combines satellite observations, model output, and reanalysis data with daily sea ice parcel drift tracks produced in a Lagrangian framework. This dataset consists of daily time series of sea ice parcel locations, sea ice and snow conditions, and atmospheric states. Building on previous work, this dataset includes remotely sensed radiative and turbulent fluxes from which the surface energy budget can be calculated. Additionally, flags indicate when sea ice parcels travel within cyclones, recording distance and direction from the cyclone center. The database drift track was evaluated by comparison with sea ice mass balance buoys. Results show ice parcels generally remain within 100km of the corresponding buoy, with a mean distance of 82.6 km and median distance of 54 km. The sea ice mass balance buoys also provide recordings of sea ice thickness, snow depth, and air temperature and pressure which were compared to this database. Ice thickness and snow depth typically are less accurate than air temperature and pressure due to the high spatial variability of the former two quantities when compared to a point measurement. The correlations between the ice parcel and buoy data are high, which highlights the accuracy of this Lagrangian database in capturing the seasonal changes and evolution of sea ice. This database has multiple applications for the scientific community; it can be used to study the processes that influence individual sea ice parcel time series, or to explore generalized summary statistics and trends across the Arctic. Applications such as these may shed light on the atmosphere-snow-sea ice interactions in the changing Arctic environment.
This preprint has been withdrawn.
-
Withdrawal notice
This preprint has been withdrawn.
-
Preprint
(729 KB)
Interactive discussion
Status: closed
-
AC1: 'Comment on tc-2021-297, additional acknowledgements', Sean Horvath, 01 Oct 2021
The following acknowledgement will be added to Section 2.1 Lagrangian Tracked Sea Ice Parcels (near line 82):
"The Lagrangian dataset was provided by and produced by Dr. M. Tschudi and Dr. J. Scott Stewart at the University of Colorado, Boulder."
Citation: https://doi.org/10.5194/tc-2021-297-AC1 -
RC1: 'Review of "Fate of sea ice in the ‘New Arctic’: A database of daily Lagrangian Arctic sea ice parcel drift tracks with coincident ice and atmospheric conditions"', Anonymous Referee #1, 19 Oct 2021
General comments:
This paper presents a database consisting of a compilation of Lagrangian sea ice tracks combined with established satellite observations, model output, and reanalysis data. This comprehensive database spans from 2002 - 2019 and contains numerous sea ice properties and atmospheric variables. With the Lagrangian framework, i.e., by moving with the ice, the authors provide a useful addition to the more traditional, Eulerian datasets of sea ice and atmospheric properties. For example, the database could be used to study changes in the Arctic energy fluxes. The authors present two use cases for climatological and more process-orientated studies. They provide a detailed outlook of which datasets they plan to incorporate in the future.
The paper provides a detailed description of a Lagrangian database that will be useful to the sea ice community. It is well structured and easy to follow. A very positive aspect of the database is that the authors put a lot of effort into incorporating different datasets and providing a choice to the user. One central aspect I was missing was a detailed discussion of the uncertainties associated with the individual properties due to the spatial/temporal errors in the tracking. In addition, since the paper's focus is on the database, it provides limited new scientific insights about changes in the Arctic. I recommend extending the results section further with a more substantial case study and shortening the outlook.
Specific comments:
Reference to previous Lagrangian studies
The introduction would benefit from a more detailed discussion of previously conducted Lagrangian ice studies in the Arctic. So far, it briefly mentions Eulerian studies (L64-65). You may want to look at the following list (not complete):
- RGPS (RADARSAT Geophysical Processor System, e.g., https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2000JC000469)
- Lagrangian Ice Tracking System ( http://icemotion.labs.nsidc.org/SITU/, https://ui.adsabs.harvard.edu/abs/2019AGUFM.C22D..03P/abstract)
- IceTrack (https://www.nature.com/articles/s41598-019-41456-y, https://tc.copernicus.org/articles/15/3897/2021/tc-15-3897-2021.html#&gid=1&pid=1)
- neXtSIM (model): https://tc.copernicus.org/articles/10/1055/2016/
Please add a short paragraph that describes the advances of your database compared to the previous work of other Lagrangian studies.
Uncertainties of the Lagrangian drift tracks:
Sections 2.1 and 3.2.1 should be clarified by providing additional details on how you obtained the Lagrangian tracks and how the temporal and spatial uncertainty of the tracks translates into the uncertainty of the atmospheric and sea ice properties. Could you explain why you used the weekly ice motion product and interpolated it linearly to a daily resolution when a daily version of the Polar Pathfinder Sea ice Motion Vectors is available? Since sea ice motion varies substantially on short time scales, I recommend using the highest temporal resolution available if no other reasons speak against it. If this is not possible, please add a sentence why you used the weekly product in the manuscript.
Please estimate (or at least discuss) the uncertainty for the various parameters (ice thickness, air temperature, ….) introduced by a misplaced trajectory caused by the linear interpolation or errors in the Lagrangian tracking itself. For example, how much does the ice thickness / the air temperature vary if the ice parcel was located 100 km away from the trajectory? Is this spatial uncertainty the same in winter in summer? To evaluate the differences between the interpolated weekly sea ice velocities and the daily (or even sub-daily) velocities, you could use the daily PathFinder product, buoys, or SAR-derived motion field, e.g., https://resources.marine.copernicus.eu/product-detail/SEAICE_GLO_SEAICE_L4_NRT_OBSERVATIONS_011_006/INFORMATION.
Evaluation of snow and ice thickness
The authors state (Section 3.2.2) that the large spatial variability of snow and ice thickness complicates the evaluation with point measurements from ice mass balance buoys. I agree, but this database would gain weight if the authors could present the ice and snow thickness data with a more detailed uncertainty analysis. There are other datasets available that could be used to evaluate the results of your tracks, e.g., satellite-derived ice thickness (CryoSAT/IceSAT) or observations from ULS (e.g., doi: 10.1002/2015JC011102) and electromagnetic induction (https://doi.org/10.5194/tc-15-2575-2021). If this is not possible, I recommend including more plausibility checks of the ice thickness and snow results, e.g., by analyzing the ice thickness time series (like in Fig. 9). For example: why is the ice thickness decreasing already as early as April? What causes the little bumps in the ice thickness time series? What role do sea ice dynamics play in the ice thickness increase? The authors should also discuss in the manuscript that the ice mass balance buoys do not consider dynamic thickness changes.
Availability of datasets and uncertainty
Unfortunately, I could not find an example database for testing during the review process. Therefore, I cannot make any comments regarding the actual handling of the database. Please indicate in the revised version where the data will be publicly accessible after the acceptance of the manuscript. I could not check whether the individual data points come with an uncertainty estimate (from the data product and from the spatial missplacement) in the database. Where available, I would highly recommend including this information as it significantly improves the quality of climatological studies. For example, it would be interesting to see the uncertainty estimate in Figure 9, and especially in Figure 10.
Database uses:
The two examples (Section 3.3) are well-chosen to provide "proof of concepts" but contain limited new scientific insights. I recommend extending section 3.3.2 (Case Studies) with an example that provides more detailed insights into an Arctic process.
Future Additions
I found section 3.4 (Future Additions) too detailed for plans. Since changes might occur in the implementation of your plans, I would suggest cutting the subsections down to one section with a few details on the datasets you want to include and especially why you want to use them.
Technical corrections:
Title:
- The title was a bit misleading because I expected an in-depth analysis of the "fate of the New Arctic" from it. You could change the order of the words, like "A new database …. to study the fate of sea ice in the New Arctic" or remove the "fate of sea ice in the New Arctic."
Abstract:
- L11: "transitioned": I think this process is still ongoing; consider using the present tense.
- L19: "the database drift track": consider adding "the quality of the database was evaluated…"
- L23: "less accurate": please specify
Introduction:
- Consider adding some more recent literature to your introduction, for example:
- L30 (e.g. doi:10.1088/1748-9326/aae3ec, doi:10.1088/1748-9326/aade56)
- L34 (e.g. IPCC 2021)
- L38 (e.g. doi:10.1088/1748-9326/aae3ec)
- L32: "what happens in the Arctic …": Consider rephrasing this sentence to express the connection between the Arctic and the lower latitudes.
- L39: "during this time": please specify which time you mean.
- L54-L60: Please consider adding a short note on sea ice dynamics and their role for sea ice thickness and extent.
- L67: See specific comments. What about other studies that used Lagrangian tracking to study changes along the ice?
- L69: specify "characteristics."
- L70: "October 2002 and September 2019".
- L70 "starts in 2002 as this is" : is there a word missing?
- L75: your database is very rich in information and helpful but does not include any ocean information what might be relevant for mass balance studies. Maybe mention this aspect either in the introduction or write a short discussion about it in section 3.4 when you talk about the use for sea ice mass balance studies.
Section 2.1:
- L82: Which version of the PathFinder are you using? Could you please indicate this?
- L82: See specific comments. Would you mind explaining why you interpolate the weekly product?
Section 2.2:
- L112: "used for the Lagrangian tracking method described above": Do you mean that a 15% CDR ice concentration decides when to stop/start the tracking?
- L140: I suggest to remove "J." in the reference of "Stroeve et al. 2020"
Section 2.3:
- L164-166: I suggest moving this paragraph to the beginning of the section, i.e., before 2.3.1. to increase the readability.
- L172: are the errors given with +/-, or is it a bias in one direction? Would you please specify the sign?
Section 2.4:
- L215: You defined N-ICE2015 in L40.
Section 3.2
- L247: What were the criteria for a wrong location?
- L248: Consider including all tracks in Figure 3a and highlight a few to understand better where those buoys were located.
- 253: Consider including an uncertainty estimate for the trajectories based on the region (either from the PathFinder product or from your analysis). If there are such differences between the regions, it would be useful to know this as a data user.
Section 3.3:
- L286: I do not fully understand your conclusion on the increasing number of parcels. Did you remove the number of "surviving" parcels (MYI) from this number? Please specify this in the text. If not, does that mean that now a larger area of the Arctic is covered with sea ice? What else increases this number?
Section 3.4:
- I suggest shortening those subsections to one section (see specific comments) and keeping the details for a second paper when you have implemented your plans. This also applies to the connection with MOSAiC.
- L.389: "from the Multidisciplinary"
- L392: "here. MOSAiC"
Section 4:
- L411: Please specify "less accurate" and state, e.g., the mean error for those buoy subsets.
Figures:
- General comments:
- Consider adding legends to your figures. I found them hard to read with only the information given in the caption.
- Make sure that x/y labels are easy to find and close to the plots.
- Maps 3a, 4, 8 contain little information. Consider combining them into 1 or 2 figures or add more information.
Figure 1:
- I do not understand why the "sea ice parcel" box is 3 times there and what "true location" means. Could you please explain?
Figure 3:
- Would you please indicate in 3a (or any other map) the spatial zones you defined to sort your data into the subregions (Laptev, Central Arctic, …)?
- Consider adding a histogram with all data.
- Why did you choose 25 km for the red line? In the text, it appeared that 100 km is your "uncertainty estimate".
- Text in b is too small; please enlarge.
- Add the number of buoys for each subregion that you used to calculate the histograms.
- Maybe add a half-sentence about the purpose of the Freedman-Diaconis rule.
Figure 4:
- Please consider combining this figure with Figure 3a or 8.
Figure 5:
- I could not find the units for the parameters. If not done so far, please add them.
- Please consider adding the standard deviation of the distributions to get an idea of the spread.
- Would you please add the number of samples used to calculate the distributions?
- I have trouble reading the plots in panel b due to the different bin widths. Could you consider using regular bins for those plots?
Figure 7:
- Please move the x-label "Day of Year" to one of the plots
Figure 8:
- In the text, you discuss this figure regarding the drift patterns (L320-322). However, I think that showing only one trajectory is not enough to display a full drift pattern. Please consider showing more trajectories or removing this part and combining the figure with one of the other maps.
- In addition, you use Figure 8 to display the track of the time series shown in Fig. 9. Is the track of the time series in Figure 10 also displayed in Figure 8, 3, or 4? If not, please include it in one of them.
Figure 9:
- Is this the time series of one of the green trajectories seen in Figure 2 or the light green in Figure 8? Would you please clarify this in your caption?
- Add "°C" for the skin temperature.
- If possible, please display uncertainties with those variables.
- Indicate the year in the graph or in the caption
Figure 10:
- A legend for a, c, would be very helpful.
- What causes the discrepancy in snow depth for the wintertime (April-June, Nov-Jan)?
- Corresponds the data gap end of September to the restart of your tracking? Consider indicating this and explain the missing time.
- If possible, please display uncertainties with those variables.
Table 1:
- Sea ice drift/Resolution: If you used the weekly product and interpolated it to daily, please indicate this in the table somehow (instead of "daily").
Citation: https://doi.org/10.5194/tc-2021-297-RC1 -
AC2: 'Reply on RC1', Sean Horvath, 16 Dec 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2021-297/tc-2021-297-AC2-supplement.pdf
-
RC2: 'Comment on tc-2021-297', Anonymous Referee #2, 30 Oct 2021
Summary
This paper documents the assembly of a database of lagrangian drift tracks that includes many of the variables needed to study changes in the surface energy budget (SEB).
They assert that the database can be used to study a range of SEB processes, and present some results from the assembled dataset. However, the authors also show that the drift tracks have mean errors of 82.6 km, and can be as high as 500 km in certain areas, so the “tracks” are probably not actually following the same parcel of sea ice.
Detailed Major Suggestions and Comments:
- The paper needs a good scientific hypothesis or question to guide the research.
- This paper jumps into the middle of the scientific process by first producing a database, then trying to find questions that the database may help answer. A more fruitful approach would be to find a scientific hypothesis, then assemble data to test that hypothesis.
- Mean errors of 82.6 km for the lagrangian tracks may or may not be large depending on the scientific question a person is trying to answer.
- 2.1 For example, if one were trying to understand the roll of large scale cyclones, then this probably is not an issue, but if one were trying to understand the small scale changes in ice concentration, then this error is unacceptable. Looking at Figure 1, the authors mark a 25km x 25 km box. A shift of even just a few km, shows that we are looking at an area of much higher sea ice concentration than the sea ice parcel that is highlighted. The SEB in the marked pixel is much different that the SEB in the parcels surrounding this.
- 2.2 If the mean errors are this large in reproducing the tracks of a buoy that is included in the gridded ice motion database, how much larger are the errors in areas without buoys?
- 2.3 Given the errors in reproducing lagrangian tracks, why not just use the actual drift tracks of the buoys? For example, the Ice Mass Balance buoys measure many of the quantities assembled here.
- 2.4 Section 3.2.2, and Figure 5: Are the differences seen in each of the panels due to real physical changes in the parcel compared to the buoy observations, or due to errors in the lagrangian tracks?
- All the different datasets assembled here also have their own errors. As with comment 2 above, whether these errors are acceptable depends on the scientific questions we are trying to answer.
- 3.1 One thing to note is that a “lagrangian approach” may also be taken by directly using many of the disparate datasets they assembled here. For example, PIOMAS includes many of these variables as forcing or as estimates from the model. PIOMAS is well documented so the errors, biases and uncertainties are known. The model can give us a “self consistent” framework to do lagrangian studies by tracking a parcel using the ice motion provided by the model.
- 3.2 By assembling disparate datasets as is done here, we lose the “self consistency“ of each data set and quantifying the errors in our results becomes difficult. Following example, looking at figure 9, the sea ice thickness obtained from PIOMAS starts declining in May long before the onset of melt derived from AIRS skin temperatures. How can we explain this given the variables assembled?
- 3.3 Sea ice thickness also increases in PIOMAS just before the onset of melt in June (Fig. 9). What forces this change? Or is there simply a shift in the pixels that they are tracking?
- 3.4 A more thorough discussion of errors for each dataset should be included in section 2.
- Reading through their abstract and conclusions, the primary contributions of this paper to science are: 1) they produced a lagrangian data base, and 2), they find an increase in the number of sea ice parcels over time. Both these findings are moot given that they may not be tracking the same parcel of sea ice, and since they note that their lagrangian drift tracks are significantly slower near Fram Strait where most parcels of sea ice is exported from the Arctic. The increase in sea ice parcels over time can probably be attributed to more of their parcels “surviving” since less are exported through Fram Strait compared to the real world.
Minor suggestions and comments:
- Line 35: Change “known as” to “attributed to”.
- Figure 5: Add units to each row of plots.
- Figure 7a: separate FYI and MYI bars so that we may be able to see any differences or trends from year to year. Interspersing FYI and MYI as shown makes it hard to see things.
- Figure 9: Mark cyclones as in Fig. 10. It would be interesting to see if cyclones are related to the changes in in snow depth, or sea ice thickness shown here.
Citation: https://doi.org/10.5194/tc-2021-297-RC2 -
AC3: 'Reply on RC2', Sean Horvath, 16 Dec 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2021-297/tc-2021-297-AC3-supplement.pdf
- The paper needs a good scientific hypothesis or question to guide the research.
Interactive discussion
Status: closed
-
AC1: 'Comment on tc-2021-297, additional acknowledgements', Sean Horvath, 01 Oct 2021
The following acknowledgement will be added to Section 2.1 Lagrangian Tracked Sea Ice Parcels (near line 82):
"The Lagrangian dataset was provided by and produced by Dr. M. Tschudi and Dr. J. Scott Stewart at the University of Colorado, Boulder."
Citation: https://doi.org/10.5194/tc-2021-297-AC1 -
RC1: 'Review of "Fate of sea ice in the ‘New Arctic’: A database of daily Lagrangian Arctic sea ice parcel drift tracks with coincident ice and atmospheric conditions"', Anonymous Referee #1, 19 Oct 2021
General comments:
This paper presents a database consisting of a compilation of Lagrangian sea ice tracks combined with established satellite observations, model output, and reanalysis data. This comprehensive database spans from 2002 - 2019 and contains numerous sea ice properties and atmospheric variables. With the Lagrangian framework, i.e., by moving with the ice, the authors provide a useful addition to the more traditional, Eulerian datasets of sea ice and atmospheric properties. For example, the database could be used to study changes in the Arctic energy fluxes. The authors present two use cases for climatological and more process-orientated studies. They provide a detailed outlook of which datasets they plan to incorporate in the future.
The paper provides a detailed description of a Lagrangian database that will be useful to the sea ice community. It is well structured and easy to follow. A very positive aspect of the database is that the authors put a lot of effort into incorporating different datasets and providing a choice to the user. One central aspect I was missing was a detailed discussion of the uncertainties associated with the individual properties due to the spatial/temporal errors in the tracking. In addition, since the paper's focus is on the database, it provides limited new scientific insights about changes in the Arctic. I recommend extending the results section further with a more substantial case study and shortening the outlook.
Specific comments:
Reference to previous Lagrangian studies
The introduction would benefit from a more detailed discussion of previously conducted Lagrangian ice studies in the Arctic. So far, it briefly mentions Eulerian studies (L64-65). You may want to look at the following list (not complete):
- RGPS (RADARSAT Geophysical Processor System, e.g., https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2000JC000469)
- Lagrangian Ice Tracking System ( http://icemotion.labs.nsidc.org/SITU/, https://ui.adsabs.harvard.edu/abs/2019AGUFM.C22D..03P/abstract)
- IceTrack (https://www.nature.com/articles/s41598-019-41456-y, https://tc.copernicus.org/articles/15/3897/2021/tc-15-3897-2021.html#&gid=1&pid=1)
- neXtSIM (model): https://tc.copernicus.org/articles/10/1055/2016/
Please add a short paragraph that describes the advances of your database compared to the previous work of other Lagrangian studies.
Uncertainties of the Lagrangian drift tracks:
Sections 2.1 and 3.2.1 should be clarified by providing additional details on how you obtained the Lagrangian tracks and how the temporal and spatial uncertainty of the tracks translates into the uncertainty of the atmospheric and sea ice properties. Could you explain why you used the weekly ice motion product and interpolated it linearly to a daily resolution when a daily version of the Polar Pathfinder Sea ice Motion Vectors is available? Since sea ice motion varies substantially on short time scales, I recommend using the highest temporal resolution available if no other reasons speak against it. If this is not possible, please add a sentence why you used the weekly product in the manuscript.
Please estimate (or at least discuss) the uncertainty for the various parameters (ice thickness, air temperature, ….) introduced by a misplaced trajectory caused by the linear interpolation or errors in the Lagrangian tracking itself. For example, how much does the ice thickness / the air temperature vary if the ice parcel was located 100 km away from the trajectory? Is this spatial uncertainty the same in winter in summer? To evaluate the differences between the interpolated weekly sea ice velocities and the daily (or even sub-daily) velocities, you could use the daily PathFinder product, buoys, or SAR-derived motion field, e.g., https://resources.marine.copernicus.eu/product-detail/SEAICE_GLO_SEAICE_L4_NRT_OBSERVATIONS_011_006/INFORMATION.
Evaluation of snow and ice thickness
The authors state (Section 3.2.2) that the large spatial variability of snow and ice thickness complicates the evaluation with point measurements from ice mass balance buoys. I agree, but this database would gain weight if the authors could present the ice and snow thickness data with a more detailed uncertainty analysis. There are other datasets available that could be used to evaluate the results of your tracks, e.g., satellite-derived ice thickness (CryoSAT/IceSAT) or observations from ULS (e.g., doi: 10.1002/2015JC011102) and electromagnetic induction (https://doi.org/10.5194/tc-15-2575-2021). If this is not possible, I recommend including more plausibility checks of the ice thickness and snow results, e.g., by analyzing the ice thickness time series (like in Fig. 9). For example: why is the ice thickness decreasing already as early as April? What causes the little bumps in the ice thickness time series? What role do sea ice dynamics play in the ice thickness increase? The authors should also discuss in the manuscript that the ice mass balance buoys do not consider dynamic thickness changes.
Availability of datasets and uncertainty
Unfortunately, I could not find an example database for testing during the review process. Therefore, I cannot make any comments regarding the actual handling of the database. Please indicate in the revised version where the data will be publicly accessible after the acceptance of the manuscript. I could not check whether the individual data points come with an uncertainty estimate (from the data product and from the spatial missplacement) in the database. Where available, I would highly recommend including this information as it significantly improves the quality of climatological studies. For example, it would be interesting to see the uncertainty estimate in Figure 9, and especially in Figure 10.
Database uses:
The two examples (Section 3.3) are well-chosen to provide "proof of concepts" but contain limited new scientific insights. I recommend extending section 3.3.2 (Case Studies) with an example that provides more detailed insights into an Arctic process.
Future Additions
I found section 3.4 (Future Additions) too detailed for plans. Since changes might occur in the implementation of your plans, I would suggest cutting the subsections down to one section with a few details on the datasets you want to include and especially why you want to use them.
Technical corrections:
Title:
- The title was a bit misleading because I expected an in-depth analysis of the "fate of the New Arctic" from it. You could change the order of the words, like "A new database …. to study the fate of sea ice in the New Arctic" or remove the "fate of sea ice in the New Arctic."
Abstract:
- L11: "transitioned": I think this process is still ongoing; consider using the present tense.
- L19: "the database drift track": consider adding "the quality of the database was evaluated…"
- L23: "less accurate": please specify
Introduction:
- Consider adding some more recent literature to your introduction, for example:
- L30 (e.g. doi:10.1088/1748-9326/aae3ec, doi:10.1088/1748-9326/aade56)
- L34 (e.g. IPCC 2021)
- L38 (e.g. doi:10.1088/1748-9326/aae3ec)
- L32: "what happens in the Arctic …": Consider rephrasing this sentence to express the connection between the Arctic and the lower latitudes.
- L39: "during this time": please specify which time you mean.
- L54-L60: Please consider adding a short note on sea ice dynamics and their role for sea ice thickness and extent.
- L67: See specific comments. What about other studies that used Lagrangian tracking to study changes along the ice?
- L69: specify "characteristics."
- L70: "October 2002 and September 2019".
- L70 "starts in 2002 as this is" : is there a word missing?
- L75: your database is very rich in information and helpful but does not include any ocean information what might be relevant for mass balance studies. Maybe mention this aspect either in the introduction or write a short discussion about it in section 3.4 when you talk about the use for sea ice mass balance studies.
Section 2.1:
- L82: Which version of the PathFinder are you using? Could you please indicate this?
- L82: See specific comments. Would you mind explaining why you interpolate the weekly product?
Section 2.2:
- L112: "used for the Lagrangian tracking method described above": Do you mean that a 15% CDR ice concentration decides when to stop/start the tracking?
- L140: I suggest to remove "J." in the reference of "Stroeve et al. 2020"
Section 2.3:
- L164-166: I suggest moving this paragraph to the beginning of the section, i.e., before 2.3.1. to increase the readability.
- L172: are the errors given with +/-, or is it a bias in one direction? Would you please specify the sign?
Section 2.4:
- L215: You defined N-ICE2015 in L40.
Section 3.2
- L247: What were the criteria for a wrong location?
- L248: Consider including all tracks in Figure 3a and highlight a few to understand better where those buoys were located.
- 253: Consider including an uncertainty estimate for the trajectories based on the region (either from the PathFinder product or from your analysis). If there are such differences between the regions, it would be useful to know this as a data user.
Section 3.3:
- L286: I do not fully understand your conclusion on the increasing number of parcels. Did you remove the number of "surviving" parcels (MYI) from this number? Please specify this in the text. If not, does that mean that now a larger area of the Arctic is covered with sea ice? What else increases this number?
Section 3.4:
- I suggest shortening those subsections to one section (see specific comments) and keeping the details for a second paper when you have implemented your plans. This also applies to the connection with MOSAiC.
- L.389: "from the Multidisciplinary"
- L392: "here. MOSAiC"
Section 4:
- L411: Please specify "less accurate" and state, e.g., the mean error for those buoy subsets.
Figures:
- General comments:
- Consider adding legends to your figures. I found them hard to read with only the information given in the caption.
- Make sure that x/y labels are easy to find and close to the plots.
- Maps 3a, 4, 8 contain little information. Consider combining them into 1 or 2 figures or add more information.
Figure 1:
- I do not understand why the "sea ice parcel" box is 3 times there and what "true location" means. Could you please explain?
Figure 3:
- Would you please indicate in 3a (or any other map) the spatial zones you defined to sort your data into the subregions (Laptev, Central Arctic, …)?
- Consider adding a histogram with all data.
- Why did you choose 25 km for the red line? In the text, it appeared that 100 km is your "uncertainty estimate".
- Text in b is too small; please enlarge.
- Add the number of buoys for each subregion that you used to calculate the histograms.
- Maybe add a half-sentence about the purpose of the Freedman-Diaconis rule.
Figure 4:
- Please consider combining this figure with Figure 3a or 8.
Figure 5:
- I could not find the units for the parameters. If not done so far, please add them.
- Please consider adding the standard deviation of the distributions to get an idea of the spread.
- Would you please add the number of samples used to calculate the distributions?
- I have trouble reading the plots in panel b due to the different bin widths. Could you consider using regular bins for those plots?
Figure 7:
- Please move the x-label "Day of Year" to one of the plots
Figure 8:
- In the text, you discuss this figure regarding the drift patterns (L320-322). However, I think that showing only one trajectory is not enough to display a full drift pattern. Please consider showing more trajectories or removing this part and combining the figure with one of the other maps.
- In addition, you use Figure 8 to display the track of the time series shown in Fig. 9. Is the track of the time series in Figure 10 also displayed in Figure 8, 3, or 4? If not, please include it in one of them.
Figure 9:
- Is this the time series of one of the green trajectories seen in Figure 2 or the light green in Figure 8? Would you please clarify this in your caption?
- Add "°C" for the skin temperature.
- If possible, please display uncertainties with those variables.
- Indicate the year in the graph or in the caption
Figure 10:
- A legend for a, c, would be very helpful.
- What causes the discrepancy in snow depth for the wintertime (April-June, Nov-Jan)?
- Corresponds the data gap end of September to the restart of your tracking? Consider indicating this and explain the missing time.
- If possible, please display uncertainties with those variables.
Table 1:
- Sea ice drift/Resolution: If you used the weekly product and interpolated it to daily, please indicate this in the table somehow (instead of "daily").
Citation: https://doi.org/10.5194/tc-2021-297-RC1 -
AC2: 'Reply on RC1', Sean Horvath, 16 Dec 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2021-297/tc-2021-297-AC2-supplement.pdf
-
RC2: 'Comment on tc-2021-297', Anonymous Referee #2, 30 Oct 2021
Summary
This paper documents the assembly of a database of lagrangian drift tracks that includes many of the variables needed to study changes in the surface energy budget (SEB).
They assert that the database can be used to study a range of SEB processes, and present some results from the assembled dataset. However, the authors also show that the drift tracks have mean errors of 82.6 km, and can be as high as 500 km in certain areas, so the “tracks” are probably not actually following the same parcel of sea ice.
Detailed Major Suggestions and Comments:
- The paper needs a good scientific hypothesis or question to guide the research.
- This paper jumps into the middle of the scientific process by first producing a database, then trying to find questions that the database may help answer. A more fruitful approach would be to find a scientific hypothesis, then assemble data to test that hypothesis.
- Mean errors of 82.6 km for the lagrangian tracks may or may not be large depending on the scientific question a person is trying to answer.
- 2.1 For example, if one were trying to understand the roll of large scale cyclones, then this probably is not an issue, but if one were trying to understand the small scale changes in ice concentration, then this error is unacceptable. Looking at Figure 1, the authors mark a 25km x 25 km box. A shift of even just a few km, shows that we are looking at an area of much higher sea ice concentration than the sea ice parcel that is highlighted. The SEB in the marked pixel is much different that the SEB in the parcels surrounding this.
- 2.2 If the mean errors are this large in reproducing the tracks of a buoy that is included in the gridded ice motion database, how much larger are the errors in areas without buoys?
- 2.3 Given the errors in reproducing lagrangian tracks, why not just use the actual drift tracks of the buoys? For example, the Ice Mass Balance buoys measure many of the quantities assembled here.
- 2.4 Section 3.2.2, and Figure 5: Are the differences seen in each of the panels due to real physical changes in the parcel compared to the buoy observations, or due to errors in the lagrangian tracks?
- All the different datasets assembled here also have their own errors. As with comment 2 above, whether these errors are acceptable depends on the scientific questions we are trying to answer.
- 3.1 One thing to note is that a “lagrangian approach” may also be taken by directly using many of the disparate datasets they assembled here. For example, PIOMAS includes many of these variables as forcing or as estimates from the model. PIOMAS is well documented so the errors, biases and uncertainties are known. The model can give us a “self consistent” framework to do lagrangian studies by tracking a parcel using the ice motion provided by the model.
- 3.2 By assembling disparate datasets as is done here, we lose the “self consistency“ of each data set and quantifying the errors in our results becomes difficult. Following example, looking at figure 9, the sea ice thickness obtained from PIOMAS starts declining in May long before the onset of melt derived from AIRS skin temperatures. How can we explain this given the variables assembled?
- 3.3 Sea ice thickness also increases in PIOMAS just before the onset of melt in June (Fig. 9). What forces this change? Or is there simply a shift in the pixels that they are tracking?
- 3.4 A more thorough discussion of errors for each dataset should be included in section 2.
- Reading through their abstract and conclusions, the primary contributions of this paper to science are: 1) they produced a lagrangian data base, and 2), they find an increase in the number of sea ice parcels over time. Both these findings are moot given that they may not be tracking the same parcel of sea ice, and since they note that their lagrangian drift tracks are significantly slower near Fram Strait where most parcels of sea ice is exported from the Arctic. The increase in sea ice parcels over time can probably be attributed to more of their parcels “surviving” since less are exported through Fram Strait compared to the real world.
Minor suggestions and comments:
- Line 35: Change “known as” to “attributed to”.
- Figure 5: Add units to each row of plots.
- Figure 7a: separate FYI and MYI bars so that we may be able to see any differences or trends from year to year. Interspersing FYI and MYI as shown makes it hard to see things.
- Figure 9: Mark cyclones as in Fig. 10. It would be interesting to see if cyclones are related to the changes in in snow depth, or sea ice thickness shown here.
Citation: https://doi.org/10.5194/tc-2021-297-RC2 -
AC3: 'Reply on RC2', Sean Horvath, 16 Dec 2021
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2021-297/tc-2021-297-AC3-supplement.pdf
- The paper needs a good scientific hypothesis or question to guide the research.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,083 | 413 | 58 | 1,554 | 43 | 39 |
- HTML: 1,083
- PDF: 413
- XML: 58
- Total: 1,554
- BibTeX: 43
- EndNote: 39
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1