Sea Ice Group at the Geophysical Institute logo1
logo2
logo 3

Example 3: Abrupt deceleration of drifting ice floe

March 24-26, 2010

The two animations above are synchronous. The time between each frame is approximately 4 minutes.

The animation on the left is a timeseries of the unprocessed radar data. The animation on the right displays two layers of processed data. The first layer has been generated using a running pixel-wise mean filter over a 6-frame (~24-minute) window. Stationary ice floes will yield a stronger mean signal over time than moving ones and they appear cyan in the animation. The second layer is generated by calculating the pixel-wise standard deviation over the same window. Floes that move past a point will yield higher standard deviations and appear red in the animation. This layer also creates trails created by moving targets.

This example shows a moving floe (circled in red in the left animation) which exhibits constant motion among its neighbors before abruptly coming to a halt (at which point it is circled in cyan). We take this to be evidence of seafloor interaction. Ice floes can be seen drifting past on both its landward and seaward sides further demonstrating its grounded state. The floe becomes ungrounded during a reversal in ice motion.

The animation below shows preliminary results from a target-tracking algorithm that will be used calculate velocities of drifting ice floes and to help identify floes exhbiting anomalous behavior possibly related to seafloor gouging. Blue dots in the animation indicate "normal" free-drifting floes that could be tracked for at least 50 frames. Red dots indicate floes that are stationary at some point during this time. In this example, not all floes that come to rest indicate gouging events, but this approach will allow to us identify potential sequences of radar data for further algorithm development.

Back to examples list