Resumen de Estimates of glacier ice discharge to the ocean combining synthetic-aperture radar-derived velocities and ground-penetrating radar-derived ice thickness. Applications to arctic glaciers
Pablo Sánchez Gámez
Ice discharge to the ocean is an important component of the mass balance of tidewater glaciers and marine-terminating ice caps. In this thesis we develop methodologies to improve the ice discharge calculations from remotely-sensed glacier velocities and radar-retrieved ice-thickness data, and apply them to provide updated estimates of ice discharge for Canadian High Arctic glaciers and the Academy of Sciences Ice Cap in Severnaya Zemlya, Russian Arctic.
Following an overview of the state-of-art methodologies to retrieve glacier surface velocity fields from remotely-sensed data, and presenting the basics of Synthetic Aperture Radar (SAR) data processing in chapters 1 and 2, the core results of this thesis are presented in chapters 3, 4 and 5, and finally the conclusions and outlook are summarized in Chapter 6.
Focusing on the core of the thesis, we firstly provide an improvement of the intensity offset tracking methodology for estimating glacier surface velocities. We optimise the offset tracking technique by omitting the azimuth offsets, using instead only range offsets from ascending and descending passes. By doing so, we are able to improve the final resolution of the velocity product, as the Terrain Observation by Progressive Scans (TOPS) acquisition mode of the Sentinel-1 mission provides resolutions of 5 m in range and 20 m in azimuth.
Simultaneously, we avoid the undesired ionospheric effect manifested in the data as azimuth streaks. We apply the developed technique to retrieve glacier-surface velocities from the southern Ellesmere Island ice caps, Canadian High Arctic. We additionally use Differential Interferometric Synthetic Aperture Radar (D-InSAR) techniques, and show that the latter shows its merits when applied to slow-moving areas, while offset tracking is more suitable for fast-moving areas. Both methods are thus complementary, and the use of both to determine glacier velocities is better than only using one or the other. We observe glacier surface velocities of up to 1200 m/year for the fastest tidewater glaciers. The land-terminating glaciers show typical velocities between 12 and 33 m/year , though with peaks up to 150 m/year in narrowing zones of the confining valleys.
Secondly, we analyse the various error sources in the estimation of ice discharge through flux gates, distinguishing the cases with ice-thickness data available for glacier cross-sections or only along the centreline. For the latter, we analyse the performance of three different U-shaped cross-sectional approaches. We apply this methodology to glaciers of the Canadian High Arctic. The velocity field is the main error source for small and medium-size glaciers (discharge <100 Mt/a ) with low velocities (<100 m/a ), while for large glaciers (dis- charge >100 Mt/a ) with high velocities (>100 m/a ) the error in cross-sectional area dominates. Thinning/thickening between ice thickness and velocity measurements should be considered, as it implies systematic errors up to 8% in our study. The U-shaped parabolic approach, which allows for an adjusted estimation when the ice-thickness measurement point is displaced from the glacier centreline, performs best, with small bias and admissible standard error. We observe an increase of ice discharge from the main glaciers (Trinity and Wykeham) of the Prince of Wales Icefield from 2015 to 2016, by 5% and 20%, respectively, followed by a decrease in 2017, by 10% and 15% respectively. Belcher Glacier, of the Devon Ice Cap, maintains similar discharges during 2015-2017.
Thirdly, we apply the developed methodologies, together with other state-of-art tech- niques, to the investigation of the dynamics and mass balance of the Academy of Sciences Ice Cap in the Russian Arctic, analysing its seasonal and intra-annual, as well as inter-annual, variations of velocity and ice discharge. We also analyse the contributions to the total mass balance of the ice cap of surface mass balance and frontal ablation (approximated here by the calving determined as ice discharge though flux gates close to the calving fronts), and the partitioning of total ablation into surface ablation and frontal ablation. With these aims, we process, using feature tracking, 54 pairs of Sentinel-1 synthetic-aperture radar images of the Academy of Sciences Ice Cap, acquired from November 2016 to November 2017.
Seasonal velocity variations up to 10% (20% peak-to-peak) of the yearly-averaged velocity are observed. Shorter-term intra-annual velocity variations have average deviations up to 16% and maximum up to 32% (32% and 64% peak-to-peak). This gives an indication of the errors incurred when extrapolating to the whole year discharge values determined from a single pair of SAR images. Average ice discharge for 2016-2017 was 1.93 ± 0.12 Gt/a .
The difference from an estimate of ∼ 1.4 Gt a −1 for 2003-2009 is attributed to the initiation of ice stream flow in a southern basin. The total geodetic mass balance for the ice cap over 2012-2016 is −1.72 ± 0.67 Gt/a (−0.31 ± 0.12 m w.e./a ). The climatic mass balance is not significantly different from zero, at 0.21 ± 0.68 Gt/a (0.04 ± 0.12 m w.e./a ), and has remained at this level for the last four decades, so the total mass balance is governed by the variations in ice discharge, whose long-term changes do not appear to respond to environmental changes but to the intrinsic characteristics of the ice cap.
