Utilization of the satellite radar data employing Offset Tracking algorithm for glacier velocity detection: a case study in the northern part of Antarctic Peninsula and James Ross Island

 

Kateřina Fárová

Geoscience Research Reports 53, 2020, pages 89–95

Full text (PDF, 10.92 MB)

Published online: 12 October 2020

Export to RIS

 

Abstract

The space-borne Synthetic Aperture Radar (SAR) data offer various possibilities for glacier velocity monitoring. One of the mostly used methods for determination of surface horizontal movement is the Offset Tracking (OT), also known as Intensity Tracking or Pixel Offset technique (Scambos et al. 1992). The OT is a procedure that measures features motion between two images using patch intensity cross-correlation optimization (Lu and Veci 2016). Unlike in differential interferometry, the OT exploits SAR signal amplitude only, and therefore is not sensitive to interferometric coherence, but requires distinct and stable surface features (Rott et al. 2011).
In our study, the free OT technique application Sentinel-1 (S-1) dataset was used. Part of the Antarctic Peninsula (AP) and the northern part of the Ulu Peninsula on the James Ross Island have been chosen as the area of interest (Fig. 1). The longer is the time separation between the pairs of images the more increases reasonable likelihood of temporal and geometrical decorrelation of SAR data. Therefore, the most of the 9 image pairs (Table 1) have been processed with the shortest achievable temporal resolution (namely 6 days). SNAP software provided by ESA has been used for image processing. Detailed steps of the OT algorithm are given by Lu and Veci, 2016. The attention was paid to the co-registration process, which could be the source of subsequent errors. The mean uncertainty of tracking results is estimated for C band sensors in tens of centimeters (Seehaus et al. 2018).
There are significant differences between the estimated glacier velocities in the AP and Ulu Peninsula. Whereas the values of the velocity detected in the AP reached up to 3.5 m/day (Figs 2 and 3), they hardly exceeded 0.6 m/day in Ulu Peninsula (Fig. 4). The probable reason for such a difference is that the most of the selected glaciers on Ulu are not the marine-terminating ones, and are not characterized by velocity gradient at the lowest terminating part. Such glaciers hardly move and if they do, their motion is very slow. Therefore, the OT method is more suitable for fast moving glaciers tongues. The results of OT on Ulu are at the edge of given accuracy and consequently less reliable.