[ 1 ] We report on long-term surface elevation changes of the central Amery Ice Shelf (AIS) by comparing elevation records spanning 4decades . We use elevation records acquired with the following methods: optical leveling (1968)(1969); ERS radar altimetry (1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003); GPS (1995GPS ( -2006; and Ice, Cloud, and land Elevation Satellite (ICESat) laser altimetry (2003)(2004)(2005)(2006)(2007). We compute multidecadal elevation trend ( dh/ dt)v alues at crossovers between the leveling route and each of the GPS and ICESat tracks as well as shorter-period dh/ dt at ERS-ERS, GPS-GPS, and ICESat-ICESat crossovers. At GPS-leveling crossovers the mean long-term dh/ dt is 0.003 ma 1 ,and at ICESat-leveling crossovers the mean dh/ dt is +0.013 ma 1 ;n either trend is significantly different from zero. The data do, however,e xhibit variable trends: near-zero change between 1991 and mid-1996, then thickening to 2003, followed by thinning [2003][2004][2005][2006][2007], with 5y ear dh/ dt averages exceeding ±0.1 ma 1 .T he changes in dh/ dt pattern in mid-1996 and again in 2003 occur with unexpected speed. The ice shelf exhibits different dh/ dt patterns than does the surrounding grounded ice, suggesting that surface mass balance variations or longer-term variations in firn densification processes are unlikely to be major causes. We conclude that these observed multiyear elevation changes must be due to currently unexplained or presently poorly quantified phenomena involving surface or basal processes and/or ice dynamics. With the multidecadal stability of the AIS established, the short-term fluctuations that we observe suggests that for other ice shelves, observed strong dh/ dt signals over short time periods do not necessarily indicate ice shelf instability.

[1] Historic velocity measurements of Antarctica's ice sheet represent vital baseline values that allow ice flow velocity variations to be observed over multidecadal timescales, such as those due to climate change. Using velocity values derived from geodetic quality measurements made in the 1960s and from more recent GPS and remote sensing studies during the 1990s, we examine the variability of ice flow velocities of the northern Amery Ice Shelf, East Antarctica, over the 30-year period, 1968-1999. The historic ice shelf velocities are reexamined using original field notes and the previous data analyses are shown to be erroneous, yielding positional and velocity errors of up to 4 km and 150 m/yr respectively. Once corrected, these historic measurements are shown to be in close (1-5 m/yr) general agreement with GPS-derived velocities from the 1990s at similar geographical locations, providing one of the first precise constraints on multidecadal ice shelf velocity variations on a large Antarctic ice shelf. Comparison of the terrestrial and GPS velocities with spatially dense velocity measurements from remote sensing imagery reveals a systematic bias in the latter of up to ±30 m/yr, mostly due to propagation of unmodeled vertical ice shelf motion due to tides and atmospheric pressure variations. We therefore excluded the remote sensing velocities from further comparison and suggest caution in the interpretation of similarly derived ice shelf velocities. Velocity differences between the GPS and terrestrial surveys were compared at nine geographical locations suggesting a small ($2.2 m/yr or $0.6%), but statistically significant, slowdown of the ice shelf.

Comparisons between computed balance velocities, obtained from two different computing schemes, and global positioning system (GPS)-derived velocities were made in the Lambert Glacier basin region, East Antarctica. The two computing schemes used for the balance-velocity computations (a flowline (FL) scheme (Remy and Minster, 1993) and a finite-difference (BW) scheme (Budd and Warner, 1996; Fricker and others, 2000)) were first evaluated and compared. One of the key issues studied was the spatial resolution of the digital elevation model (DEM), representing the surface topography of the ice sheet, and the sensitivity of the balance velocities to the length of smoothing applied to the DEM. Comparison with the GPS velocities validated the two schemes to within 5–25% but showed the high sensitivity of the flowline method to the length scale of the smoothing. The finite-difference scheme was found to be robust to the chosen smoothing scale, but the balance-velocity values increased when a finer-resolution DEM was used. Both FL and BW computing schemes tended to overestimate the balance velocities in comparison with the GPS values; some of this discrepancy can be attributed to ice-sheet sliding.

No abstract