Multisensor validation of tidewater glacier flow fields derived from synthetic aperture radar (SAR) intensity tracking

Abstract

Following the general warming trend in Greenland, an increase in calving rates, retreat and ice flow has been observed at ocean-terminating outlet glaciers. These changes contribute substantially to the current mass loss of the Greenland Ice Sheet. In order to constrain models of ice dynamics as well as estimates of mass change, detailed knowledge of geometry and ice flow is needed, in particular on the rapidly changing tongues of ocean-terminating outlet glaciers. In this study, we validate velocity estimates and spatial patterns close to the calving terminus of such an outlet derived from an iterative offset-tracking method based on synthetic aperture radar (SAR) intensity data with a collection of three independent reference measurements of glacier flow. These reference datasets are comprised of measurements from differential GPS, a terrestrial radar interferometer (TRI) and repeated unmanned aerial vehicle (UAV) surveys. Our approach to SAR velocity processing aims at achieving a relatively fine grid spacing and a high temporal resolution in order to best resolve the steep velocity gradients in the terminus area and aims to exploit the 12 d repeat interval of the single-satellite Sentinel-1A sensor. Results from images of the medium-sized ocean-terminating outlet glacier Eqip Sermia acquired by Sentinel-1A and RADARSAT-2 exhibit a mean difference of 11.5 % when compared to the corresponding GPS measurements. An areal comparison of our SAR velocity fields with independently generated velocity maps from TRI and UAV surveys showed good agreement in magnitude and spatial patterns, with mean differences smaller than 0.7 m d−1. In comparison with existing operational velocity products, our SAR-derived velocities show an improved spatial velocity pattern near the margins and calving front. There 8 % to 30 % higher surface ice velocities are produced, which has implications on ice fluxes and on mass budget estimates of similarly sized outlet glaciers. Further, we show that offset tracking from SAR intensity data at relatively low spatio-temporal sampling intervals is a valid method to derive glacier flow fields for fast-flowing glacier termini of outlet glaciers and, given the repeat period of 12 d of the Sentinel-1A sensor (6 d with Sentinel-1B), has the potential to be applied operationally in a quasi-continuous mode.