SAR Data Center Project: Arctic MeasuresMethodology

An understanding of Arctic sea ice dynamics is accomplished by monitoring sea ice movement with SAR imagery throughout the year. The convergence of satellite orbits at high latitudes means that repeat coverage tends to improve near the North and South Poles. This permits complete coverage of the entire Arctic basin every three days. These three day composite mosaics serve as a “snaphot” in time and are compared to similar snapshots in a temporal series. In the figure below, one such composite mosaic of Radarsat-1 swaths is shown.

Click to Enlarge

With the onset of each summer or winter season, a regular grid is superposed across the first snapshot of the seasonal series. (See figure below.) The grid spacing is 10 km at high latitudes and 20km near the Arctic perimeter. Monitoring sea ice movement consists of tracking the sea ice under each vertex in the regular grid. Sea ice exhibits structural characteristics based its composition of first year and multi-year ice. The key to studying ice dynamics is tracking these characteristic features as the season evolves; allowing the regular grid to warp as the ice beneath each vertex moves.

Click to enlarge

Sea ice motion consists of both translation and deformation. In the animation below, an initial 5 by 5 grid is established over a region, then followed from December 2006 through February 2007, in six day intervals. The deformation of the grid is accompanied with significant translation over the time period; indicative of the large scale forces operating in the Arctic basin.

The tracking is an automated process that relies on feature extraction to find the common points in subsequent snapshots. However, the automated process is not perfect, necessitating human intervention for quality control. Operators correct the vertices when locations are misidentified and remove points from consideration when ice begins to break-up or the grid polygons lose their physical meaning.

For the sake of product nomenclature, it should be noted that we have adopted the Lagrangian frame of reference (as compared to the Eulerian). Instead of studying sea ice motion at a particular location in the Arctic over time, we follow individual ice parcels as they move through space and time. Lagrangian tracking is better-suited for the time evolving Arctic ice motion and can be readily manipulated to generate a number of sea ice products that will be described in the Products and Tools Section.