Deformation composite of the RADARSAT Geophysical Processor System (RGPS) Lagrangian motion data

• Main Authors: Bouchat, Amélie, Hutter, Nils
• Format: Dataset
• Language: unknown
• Published: Zenodo 2022
• Subjects: sea ice; deformation; RGPS; satellite observations; rheology; Arctic; Tremblay; Sea ice
• Online Access: https://dx.doi.org/10.5281/zenodo.6321327; https://zenodo.org/record/6321327

Deformation composite constructed from the Lagrangian RADARSAT Geophysical Processor System (RGPS) Lagrangian motion data for January-February-March, 1997 to 2008. The nominal temporal and spatial scales for the composite data are T * = 3 days, and L * = 10 km. This data is analyzed and compared with model deformation statistics in Bouchat et al., Sea Ice Rheology Experiment (SIREx), Part I: Scaling and statistical properties of sea-ice deformation fields, Journal of Geophysical Research: Oceans (2022). The original RGPS Lagrangian motion data set consists in lists of trajectories (time and positions records) for points that are tracked in sequential synthetic aperture radar (SAR) images. The trajectories are organized in different “streams”, corresponding to different initial satellite passes over which a set of tracked points were initialized. For all streams, the trajectories are initialized on a uniform 10 km x 10 km grid at the beginning of the winter in November. Each tracked point can therefore be assigned to (i,j) indices corresponding to its initialization location on the grid. As time increases and the position records are updated, the tracked points are no longer uniformly separated, but their assigned (i,j) indices do not change. The trajectory records are updated when the tracking algorithm detects the tracked points in a new SAR image. The update interval is therefore not always the same for all points, nor is it always on the same time/day within a given stream as the tracking algorithm may be unsuccessful for certain images/points. Moreover, the multiple streams can overlap spatially, such that more than one trajectory can be assigned to the same (i,j) indices. Computing strain rates directly from the original RGPS Lagrangian motion product therefore results in deformation estimates that can span a wide range of spatio-temporal scales, that are not temporally coherent across all streams, and that can also be spatially redundant. The goal of constructing a deformation composite from the original RGPS Lagrangian motion product is to generate a coherent set of non-overlapping Lagrangian deformation estimates at fixed time intervals and with a uniform spatial scale that can be used for statistical analysis. The RGPS Lagrangian deformation composite is constructed using the weighted-average pre-processing method described in Bouchat & Tremblay (2020) and Hutter et al. (2020) and summarized here. For each stream separately, we first define quadrilateral Lagrangian cells assigned to the (i,j) indices by combining records from the (i,j), (i+1,j) , (i,j+1) , and (i+1, j+1) available Lagrangian trajectories. For each (i,j) cell, we then compute the Lagrangian strain rates if, between any two update times, the cell's records have: (i) simultaneous (plus or minus 3 hours) start and end times for all fours corners, (ii) an average time interval for all corners that corresponds to the nominal temporal resolution of T * = 3 days, and (iii) an area at the start time that corresponds to the nominal spatial resolution of L * = 10 km. The strain rates, the cell area, and the start and end times used to compute the cell's strain rates are also assigned to the (i,j) indices. Then, to create the composite deformation estimates at the same fixed start and end dates for all cells, we average the strain rate and area records at each (i,j) indices in fixed 3-day periods starting on January 1st, using the overlapping time between their start/end date interval with the fixed 3-day periods as weight. For visualization purposes only, we also average the cells' corners' starting positions from all records overlapping with the fixed 3-day interval and use these averaged positions as approximate coordinates for the composite deformation cells. Finally, all streams are spatially combined into a single strain rate composite. In the case of spatial overlap between two or more streams, we keep the cells that have the longest time coverage and discard the other ones. There is one netCDF file per year. Data are organized in matrices where the (i,j) indices are the Lagrangian cells identifier. This allows us to keep track of neighbouring cells for the scaling analysis. See below for more information on what variables are included in the files and their structure. 1. Variables included (x1,y1), (x1,y2), (x3,y3), (x4,y4) : Average positions of the composite cells' corners. Used for visualization only (deformations should not be computed using these positions) - (meters); A : Composite cells' area - (meters squared); dudx, dudy, dvdx, dvdy : Composite cell's velocity derivatives (strain rates/deformation) - (1/seconds); d_dudx, d_dudy, d_dvdx, d_dvdy : Trajectory error on the composite cells' velocity derivatives - (1/seconds); time : Day of year. *Note: The composite cells were removed if their average position was within 100 km from land. Before comparing the deformation statistics with sea-ice models, one should only keep cells available in both the model and the RGPS composite. 2. Variable structure All variables (except time ) are matrices with axes ( it, i, j ), where it is the time stamp/iteration and i,j are the cells identifiers. See below for how the cells are defined: |--------------------------------------------------------------> j-axis | | ( x1_ij,y1_ij ) o -------------------- o ( x2_ij,y2_ij ) | | | | | A_ij or dudx_ij | | | | | ( x4_ij,y4_ij ) o ------------------- o ( x3_ij,y3_ij ) | | V i-axis References: Bouchat, A., & Tremblay, B. (2020). Reassessing the Quality of Sea-Ice Deformation Estimates Derived From the RADARSAT Geophysical Processor System and Its Impact on the Spatiotemporal Scaling Statistics. Journal of Geophysical Research: Oceans, 125(8), https://doi.org/10.1029/2019JC015944 Hutter, N. and Losch, M.: Feature-based comparison of sea ice deformation in lead-permitting sea ice simulations, The Cryosphere, 14, 93–113, https://doi.org/10.5194/tc-14-93-2020, 2020. The original RGPS Lagrangian Motion data set can be accessed here: https://asf.alaska.edu/data-sets/derived-data-sets/seaice-measures/sea-ice-measures-data-products/