A Novel Method for Creating Complete, Gapless Lines From Fragmented 2D Data of Antarctic Grounding Line Measurements

Abstract

InSAR grounding line retrieval has resulted in numerous datasets that cover most of the ice shelves that fringe Antarctica. With the wealth of data the Sentinel-1 mission collects over the Antarctic grounding zone, it has become feasible to process nearly 60 grounding lines per year for some locations. While the interferometric pipelines are well established and operational, the need for manual delineation of the landward side of the dense fringe belt detected in double difference interferograms prevented the efficient processing of dense time series. Most Antarctic-wide products contain manually traced individual grounding lines at low temporal resolution and are focused on spatial completeness. However, recent advances in automatic detection with deep neural networks now offer the possibility to create long time series [Ramanath Tarekere et al. 2024].

Neither manually nor automatically delineated grounding lines are complete, as they often contain gaps due to variations in interferometric coherence, tidal states, and the generally intricate geometry of the grounding line. Furthermore, these limitations result in a time series of fragmented 2D line segments that cannot be directly used to trace the perimeter of an entire ice shelf or even form a single circum-Antarctic grounding line. However, applications in ice shelf modelling, calculation of ice export in the mass-budget method, and retreat measurements all benefit from simple continuous grounding line geometries, which surround entire ice shelves. Currently, a widely used product that fulfills these requirements but lacks annotation and timeliness is the MEaSUREs Antarctic Boundaries dataset [Mouginot et al. 2017].

Here, we present a novel method that is able to create single gapless and complete grounding lines for individual ice shelves for targeted time periods (annual, bi-annual, monthly). The fragmented grounding lines are stacked in time to distinguish grounded from floating ice. We then apply a median threshold to detect regions grounded in at least 50% of the measurements. In order to create the gapless perimeter of an ice shelf, we prioritize grounding lines according to their acquisition time and fill the remaining gaps with measurements from other periods or other sensors. Combined with ice shelf calving front data, we can produce time-annotated ice shelf perimeter, area, and area change.

References:

Mouginot, J., Scheuchl, Bernd and Rignot, Eric (2017). MEaSUREs Antarctic Boundaries for IPY 2007-2009 from Satellite Radar. (NSIDC-0709, Version 2). Boulder, Colorado USA. NASA National Snow and Ice Data Center Distributed Active Archive Center. (Visited on 11/19/2024).

Ramanath Tarekere, S. et al. (Mar. 11, 2024). "Deep Learning Based Automatic Grounding Line Delineation in DInSAR Interferograms". In: EGUsphere, pp. 1-35. (Visited on 04/16/2024).