A sea ice deformation and rotation rates dataset (2017&ndash;2023) from the Environment and Climate Change Canada Automated Sea Ice Tracking System (ECCC-ASITS)

Plante, Mathieu; Lemieux, Jean-François; Tremblay, L. Bruno; Bouchat, Amélie; Ringeisen, Damien; Blain, Philippe; Howell, Stephen; Brady, Mike; Komarov, Alexander S.; Duval, Béatrice; Yakuden, Lekima

doi:https://doi.org/10.5194/essd-2024-227

Preprints

https://doi.org/10.5194/essd-2024-227

Preprints

12 Aug 2024

| 12 Aug 2024

Status: a revised version of this preprint was accepted for the journal ESSD and is expected to appear here in due course.

A sea ice deformation and rotation rates dataset (2017–2023) from the Environment and Climate Change Canada Automated Sea Ice Tracking System (ECCC-ASITS)

Mathieu Plante, Jean-François Lemieux, L. Bruno Tremblay, Amélie Bouchat, Damien Ringeisen, Philippe Blain, Stephen Howell, Mike Brady, Alexander S. Komarov, Béatrice Duval, and Lekima Yakuden

Abstract. Sea ice forms a thin but horizontally extensive boundary between the ocean and the atmosphere, with a complex crust-like dynamics characterized by intermittent sea ice deformations. The heterogeneity and localisation of these sea ice deformations are important characteristics of the sea ice cover that can be used to evaluate the performance of dynamical sea-ice models against observations across multiple spatial and temporal scales. Here, we present a new pan-Arctic sea-ice deformation and rotation rates (SIDRR) dataset derived from the RADARSAT Constellation Mission (RCM) and Sentinel-1 (S1) synthetic aperture radar (SAR) imagery from 01 September 2017 to 31 August 2023. The SIDRR estimates are derived from contour integrals of triangulated ice motion data, obtained from the Environment and Climate Change Canada automated sea ice tracking system (ECCC-ASITS). The SIDRR dataset is not regularised, and consist in stacked data from multiple SAR images computed on a range of temporal (0.5–6 days) and spatial (4–10 km) scales. It covers the entire Arctic Ocean and all peripheral seas except the Okhotsk Sea. Uncertainties associated with the propagation of tracking errors on the deformation values are included. We show that rectangular patterns of deformation features are visible when the sampled deformation rates are lower than the propagation error. This limits the meaningful information the can be extracted in areas with low SIDRR values, but allows for the characterisation of SIDRR in Linear Kinematic Features. The spatial coverage and range of resolutions of the SIDRR dataset provides an interesting opportunity to investigate regional and seasonal variability of sea-ice deformation statistics across scales, and can be used to determine metrics for model evaluation.

Received: 08 Jun 2024 – Discussion started: 12 Aug 2024

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Mathieu Plante, Jean-François Lemieux, L. Bruno Tremblay, Amélie Bouchat, Damien Ringeisen, Philippe Blain, Stephen Howell, Mike Brady, Alexander S. Komarov, Béatrice Duval, and Lekima Yakuden

Status: closed

RC1: 'Comment on essd-2024-227', Anton Korosov, 13 Sep 2024

The manuscript describes the methodology and the dataset of sea ice deformation and rotation derived from S1 and RCN-based sea ice drift. It is very well written and provides all the necessary information for users of the new dataset. Aspects of deformation computation and uncertainty estimates are covered in good detail. Artefacts appearing on ice deformation maps due to the coarse resolution of SAR data used for ice drift retrieval are well highlighted.
It is an excellent work, and I have only two minor comments on the manuscript's text. However, I have a major comment on the provided data. I recommend updating the dataset before the manuscript can be published, and I provide detailed instructions on how it can be done efficiently.
Minor comments on the text
Line 61. The phrase "combination of S1 and RCM SAR" reads as ice drift was derived from pairs containing S1 and RCM images. However, that was not the case. S1 was paired with S1 and RCM with RCM. Please rephrase for clarity.
Line 265: 'will BE assessed'
Major comment on the dataset
The dataset's structure is such that start/end coordinates of all three nodes of an element are provided for each deformation value. Therefore, the coordinates are duplicated for two neighbour elements. Therefore, the same coordinates are provided (duplicated) as many times as a node is shared by neighbour elements. On average, the dataset size is three -- four times larger than it should be. Moreover, this structure is not convenient to use for plotting and analysis.
For illustration purposes, let's consider two elements (triangles) with four nodes (two nodes are shared). Then, the current data structure can be presented as:
element_index = [1, 2]
longitudes_corner1 = [lon1, lon2]
longitudes_corner2 = [lon2, lon4]
longitudes_corner3 = [lon3, lon3]
Here the number with lon inside the longitude vector indicates the node index.
It is clear that values of lon2 and lon3 are stored in the dataset twice. The same happens with all start / end longitudes and latitudes and with indices of elements.
I propose to use a more efficient data structure instead:
longitudes = [lon1, lon2, lon3, lon4]
triangulation = [[1, 2, 3], [2, 3, 4]]
Longitudes is a vector with unique values of coordinates for each node.
Triangulation is a 3 x N array with indices of corners for each element.
Transformation of data from the current structure to the new structure can be efficiently performed in a few lines of code. Here I provide just a very brief snippet using Python:
# stack indices of all nodes for one image pair
ids = np.hstack([ids1, ids2, ids3])

# find unique indices and inversion rule

_, unq_idx, unq_inv = np.unique(ids, return_index=True, return_inverse=True)

# reduce the data (keep only unique values of coordinates)

start_lat = np.hstack([start_lat1, start_lat2, start_lat3])[unq_idx]

# create triangulation from the inversion rule

tri = unq_inv.reshape(3, -1).T

Detailed explanations are provided in the notebook (see also the corresponding pull request to the McGill-sea-ice/SIDRRpy repository):

https://github.com/akorosov/SIDRRpy/blob/main/optimize_data_structure.ipynb

As exemplified in the notebook, providing a ready-to-use triangulation also facilitates plotting (and further analysis of deformation fields):

plt.tripcolor(start_lon, start_lat, shr, triangles=tri)

Citation: https://doi.org/10.5194/essd-2024-227-RC1
RC2: 'Comment on essd-2024-227', Anonymous Referee #2, 16 Sep 2024

The manuscript is concerned with a dataset of sea ice and rotation rates (SIDRR) based on Sentinel-1 and RadarSAT SAR imagery is presented. The method used to build the dataset is very well described, along with a budget of the errors the estimated parameters can be affected.

This is an excellent work, the product it describes waits only to be exploited by the sea ice community.

I do not have specific comments, so I recommend publication in its present form.

Citation: https://doi.org/10.5194/essd-2024-227-RC2
RC3: 'Comment on essd-2024-227', Anonymous Referee #3, 24 Sep 2024

General comments:

This study presents a Sea Ice Deformation and Rotation Rates (SIDRR) dataset, derived from sea ice motion vectors obtained through the Environment and Climate Change Canada automated sea ice tracking system (ECCC-ASITS). The dataset spans from 2017 to 2023 and covers the pan-Arctic region, using Synthetic Aperture Radar (SAR) imagery from the Sentinel-1 (S1) and RADARSAT Constellation Mission (RCM) satellites. The paper offers valuable insights into the processing methods and analysis of sea ice deformations across multiple spatial and temporal scales, and serves as an important contribution to sea ice dynamics research.

The overall structure of the manuscript is sound, and the scientific value of the dataset is well articulated. The data processing methods are thorough, and the analysis of uncertainties is a crucial and valuable aspect of the study. That said, there are several points where the paper could be improved. Below are specific suggestions and observations.
Specific comments:

1. Line 40: The introduction of the RADARSAT Geophysical Processing System (RGPS) dataset lacks a clear transition and purpose. To improve clarity, it would be beneficial to emphasize why the RGPS dataset is insufficient for certain studies, particularly those requiring higher spatial and temporal resolution to capture fine-scale sea ice deformation. This would provide a stronger foundation for introducing the new dataset developed in the paper.

2. Line 74: You should explicitly mention the spatial resolution of both products. The temporal resolution differences between S1 and RCM are also crucial for the analysis. S1 might have less frequent passes (especially due to the polar orbit), whereas RCM can provide higher temporal resolution with more frequent observations.

3. Line 81: Explain the rationale for the 30% overlap to help the reader understand its importance. You could clarify whether this overlap increases the confidence in feature tracking or provides redundancy to account for data gaps.

4. Line 86: Define what is meant by "raw" data here. Is the goal to maintain the dataset's flexibility so that users can choose their own post-processing methods, or is there another reason to keep the data in this format?

5. Line 201: "Validation of the SIDRR features is complicated by 1. the multi-scale nature of sea ice deformations...", the first point regarding the multi-scale nature could benefit from additional explanation. What exactly about the multi-scale nature complicates validation? Adding a sentence to explain how multi-scale deformations make it difficult to compare or validate the data would be helpful.

Citation: https://doi.org/10.5194/essd-2024-227-RC3
AC1:
'Comment on essd-2024-227', Mathieu Plante, 09 Oct 2024

The comment was uploaded in the form of a supplement: https://essd.copernicus.org/preprints/essd-2024-227/essd-2024-227-AC1-supplement.pdf

Citation: https://doi.org/10.5194/essd-2024-227-AC1
- EC1: 'Editor Reply on AC1', Petra Heil, 11 Oct 2024
  
  Dear authors.
  Thank you for the clear response to the comments of the reviews.
  I am pleased for your active approach to manage your dataset (based on comment from reviewer 1).
  Based on the revised text included in your response, I suggest to be even clearer. For example:
  "the ECCC-ASITS determines the sea ice motion at a 25 and 6.25 km resolution from S1 or RCM SAR image pairs.”
  I look forward to receive your revised manuscript.
  Best wishes, Petra
  
  Citation: https://doi.org/10.5194/essd-2024-227-EC1

Status: closed

RC1: 'Comment on essd-2024-227', Anton Korosov, 13 Sep 2024

The manuscript describes the methodology and the dataset of sea ice deformation and rotation derived from S1 and RCN-based sea ice drift. It is very well written and provides all the necessary information for users of the new dataset. Aspects of deformation computation and uncertainty estimates are covered in good detail. Artefacts appearing on ice deformation maps due to the coarse resolution of SAR data used for ice drift retrieval are well highlighted.
It is an excellent work, and I have only two minor comments on the manuscript's text. However, I have a major comment on the provided data. I recommend updating the dataset before the manuscript can be published, and I provide detailed instructions on how it can be done efficiently.
Minor comments on the text
Line 61. The phrase "combination of S1 and RCM SAR" reads as ice drift was derived from pairs containing S1 and RCM images. However, that was not the case. S1 was paired with S1 and RCM with RCM. Please rephrase for clarity.
Line 265: 'will BE assessed'
Major comment on the dataset
The dataset's structure is such that start/end coordinates of all three nodes of an element are provided for each deformation value. Therefore, the coordinates are duplicated for two neighbour elements. Therefore, the same coordinates are provided (duplicated) as many times as a node is shared by neighbour elements. On average, the dataset size is three -- four times larger than it should be. Moreover, this structure is not convenient to use for plotting and analysis.
For illustration purposes, let's consider two elements (triangles) with four nodes (two nodes are shared). Then, the current data structure can be presented as:
element_index = [1, 2]
longitudes_corner1 = [lon1, lon2]
longitudes_corner2 = [lon2, lon4]
longitudes_corner3 = [lon3, lon3]
Here the number with lon inside the longitude vector indicates the node index.
It is clear that values of lon2 and lon3 are stored in the dataset twice. The same happens with all start / end longitudes and latitudes and with indices of elements.
I propose to use a more efficient data structure instead:
longitudes = [lon1, lon2, lon3, lon4]
triangulation = [[1, 2, 3], [2, 3, 4]]
Longitudes is a vector with unique values of coordinates for each node.
Triangulation is a 3 x N array with indices of corners for each element.
Transformation of data from the current structure to the new structure can be efficiently performed in a few lines of code. Here I provide just a very brief snippet using Python:
# stack indices of all nodes for one image pair
ids = np.hstack([ids1, ids2, ids3])

# find unique indices and inversion rule

_, unq_idx, unq_inv = np.unique(ids, return_index=True, return_inverse=True)

# reduce the data (keep only unique values of coordinates)

start_lat = np.hstack([start_lat1, start_lat2, start_lat3])[unq_idx]

# create triangulation from the inversion rule

tri = unq_inv.reshape(3, -1).T

Detailed explanations are provided in the notebook (see also the corresponding pull request to the McGill-sea-ice/SIDRRpy repository):

https://github.com/akorosov/SIDRRpy/blob/main/optimize_data_structure.ipynb

As exemplified in the notebook, providing a ready-to-use triangulation also facilitates plotting (and further analysis of deformation fields):

plt.tripcolor(start_lon, start_lat, shr, triangles=tri)

Citation: https://doi.org/10.5194/essd-2024-227-RC1
RC2: 'Comment on essd-2024-227', Anonymous Referee #2, 16 Sep 2024

The manuscript is concerned with a dataset of sea ice and rotation rates (SIDRR) based on Sentinel-1 and RadarSAT SAR imagery is presented. The method used to build the dataset is very well described, along with a budget of the errors the estimated parameters can be affected.

This is an excellent work, the product it describes waits only to be exploited by the sea ice community.

I do not have specific comments, so I recommend publication in its present form.

Citation: https://doi.org/10.5194/essd-2024-227-RC2
RC3: 'Comment on essd-2024-227', Anonymous Referee #3, 24 Sep 2024

General comments:

This study presents a Sea Ice Deformation and Rotation Rates (SIDRR) dataset, derived from sea ice motion vectors obtained through the Environment and Climate Change Canada automated sea ice tracking system (ECCC-ASITS). The dataset spans from 2017 to 2023 and covers the pan-Arctic region, using Synthetic Aperture Radar (SAR) imagery from the Sentinel-1 (S1) and RADARSAT Constellation Mission (RCM) satellites. The paper offers valuable insights into the processing methods and analysis of sea ice deformations across multiple spatial and temporal scales, and serves as an important contribution to sea ice dynamics research.

The overall structure of the manuscript is sound, and the scientific value of the dataset is well articulated. The data processing methods are thorough, and the analysis of uncertainties is a crucial and valuable aspect of the study. That said, there are several points where the paper could be improved. Below are specific suggestions and observations.
Specific comments:

1. Line 40: The introduction of the RADARSAT Geophysical Processing System (RGPS) dataset lacks a clear transition and purpose. To improve clarity, it would be beneficial to emphasize why the RGPS dataset is insufficient for certain studies, particularly those requiring higher spatial and temporal resolution to capture fine-scale sea ice deformation. This would provide a stronger foundation for introducing the new dataset developed in the paper.

2. Line 74: You should explicitly mention the spatial resolution of both products. The temporal resolution differences between S1 and RCM are also crucial for the analysis. S1 might have less frequent passes (especially due to the polar orbit), whereas RCM can provide higher temporal resolution with more frequent observations.

3. Line 81: Explain the rationale for the 30% overlap to help the reader understand its importance. You could clarify whether this overlap increases the confidence in feature tracking or provides redundancy to account for data gaps.

4. Line 86: Define what is meant by "raw" data here. Is the goal to maintain the dataset's flexibility so that users can choose their own post-processing methods, or is there another reason to keep the data in this format?

5. Line 201: "Validation of the SIDRR features is complicated by 1. the multi-scale nature of sea ice deformations...", the first point regarding the multi-scale nature could benefit from additional explanation. What exactly about the multi-scale nature complicates validation? Adding a sentence to explain how multi-scale deformations make it difficult to compare or validate the data would be helpful.

Citation: https://doi.org/10.5194/essd-2024-227-RC3
AC1:
'Comment on essd-2024-227', Mathieu Plante, 09 Oct 2024

The comment was uploaded in the form of a supplement: https://essd.copernicus.org/preprints/essd-2024-227/essd-2024-227-AC1-supplement.pdf

Citation: https://doi.org/10.5194/essd-2024-227-AC1
- EC1: 'Editor Reply on AC1', Petra Heil, 11 Oct 2024
  
  Dear authors.
  Thank you for the clear response to the comments of the reviews.
  I am pleased for your active approach to manage your dataset (based on comment from reviewer 1).
  Based on the revised text included in your response, I suggest to be even clearer. For example:
  "the ECCC-ASITS determines the sea ice motion at a 25 and 6.25 km resolution from S1 or RCM SAR image pairs.”
  I look forward to receive your revised manuscript.
  Best wishes, Petra
  
  Citation: https://doi.org/10.5194/essd-2024-227-EC1

Mathieu Plante, Jean-François Lemieux, L. Bruno Tremblay, Amélie Bouchat, Damien Ringeisen, Philippe Blain, Stephen Howell, Mike Brady, Alexander S. Komarov, Béatrice Duval, and Lekima Yakuden

Data sets

Sea ice deformation and rotation rates (SIDRR) from the ECCC-ASITS Mathieu Plante, Jean-François Lemieux, L. Bruno Tremblay, Amélie Bouchat, Damien Ringeisen, Philippe Blain, Stephen Howell, Mike Brady, Alexander S. Komarov, Béatrice Duval, Lekima Yakuden, and Frédérique Labelle https://doi.org/10.5281/zenodo.11520803

Mathieu Plante, Jean-François Lemieux, L. Bruno Tremblay, Amélie Bouchat, Damien Ringeisen, Philippe Blain, Stephen Howell, Mike Brady, Alexander S. Komarov, Béatrice Duval, and Lekima Yakuden

Viewed

Total article views: 819 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
470	67	282	819	14	17

HTML: 470
PDF: 67
XML: 282
Total: 819
BibTeX: 14
EndNote: 17

Views and downloads (calculated since 12 Aug 2024)

Month	HTML	PDF	XML	Total
Aug 2024	119	23	10	152
Sep 2024	151	18	28	197
Oct 2024	101	17	96	214
Nov 2024	62	6	94	162
Dec 2024	37	3	54	94

Cumulative views and downloads (calculated since 12 Aug 2024)

Month	HTML	PDF	XML	Total
Aug 2024	119	23	10	152
Sep 2024	151	18	28	197
Oct 2024	101	17	96	214
Nov 2024	62	6	94	162
Dec 2024	37	3	54	94

Viewed (geographical distribution)

Total article views: 784 (including HTML, PDF, and XML) Thereof 784 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 13 Dec 2024

Short summary

Sea ice forms a thin boundary between the ocean and the atmosphere, with a complex crust-like dynamics and ever-changing networks of sea ice leads and ridges. Statistics of these dynamical features are often used to evaluate sea ice models. Here, we present a new pan-Arctic dataset of sea ice deformations derived from satellite imagery, from 01 September 2017 to 31 August 2023. We discuss the dataset coverage and some limitations associated with uncertainties in the computed values.


Total:	0
HTML:	0
PDF:	0
XML:	0