http://websrv.cs.umt.edu/isis/images/f/f0/File.pdf
Received 22 September 2005; revised 14 October 2005; accepted 19 October 2005; published 22 November 2005.
[1] A significant amount of the measured coastal thinning of the Greenland ice sheet may be due to recent acceleration of outlet glaciers. Using remote sensing, we measured two major periods of speedup on Helheim Glacier between 2000 and 2005 that increased peak speeds from approximately 8
to 11 km/yr. These speedups coincided with rapid retreats of the calving front, totaling over 7.5 km. The glacier also thinned by over 40 m from 2001 to 2003. Retreat of the ice front appears to decrease resistance to flow and concentrates the gravitational driving force over a smaller area. Farther
up-glacier, acceleration may be a delayed response to surface draw-down and steepening of the glacier’s main trunk. If the 2005 speedup also produces strong thinning, then much of the glacier’s main trunk may un-ground, leading to further retreat. Citation: Howat, I. M., I. Joughin, S. Tulaczyk, and S. Gogineni (2005), Rapid retreat and acceleration of Helheim Glacier, east Greenland, Geophys. Res.
Lett., 32, L22502, doi:10.1029/2005GL024737.
2. Methods
[4] Ice flow velocity at Helheim was measured from satellite image pairs once in 2000 and twice in 2003, 2004, and 2005 (Figure 1). The October 2000 velocities were determined using standard speckle tracking techniques applied to a RADARSAT image pair separated by 24-days [Joughin, 2002]. Errors in these estimates are ±3%, which are largely attributable to error in the elevation data used to correct for topographic effects.
[5] Velocities for 2003 through 2005 were obtained from automated surface feature tracking [Scambos et al., 1992], using principle component images of bands 1–3 (visible/ near infra-red) of the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensor. All image pairs were geometrically rectified using the same ground control, so the errors largely arise from uncertainty in the cross-correlation match (<10 m per image pair). To correct for additional errors induced by terrain, which are usually 5–10 m, two-dimensional displacements of known stationary features were triangulated to, and subtracted from, each on-ice measurement