UC Irvine

Detailed maps of bed elevation and ice thickness are essential for understanding and projecting the evolution of the ice sheets. Such maps are traditionally obtained using airborne radarsounding profiler data interpolated onto regular grids using geostatistical tools such as kriging. Here we compare three mapping techniques applied to a dense radar survey of Russell Glacier, West Greenland, by NASA Operation IceBridge: (1) radar tomography (RT) processing of the radar data to map the bed elevation, (2) interpolation of radar-derived thickness by ordinary kriging (KR) and (3) reconstruction of ice thickness based on the principles of mass conservation (MC) combining radar-sounding profiler and ice motion data. RT eliminates ambiguities caused by off-nadir reflections, but is spatially limited. KR yields a standard error in bed elevation of 35 m, but large errors (>300m a-1) in flux divergence when combined with ice motion data. MC yields a comparable performance in bed elevation mapping, and errors smaller than 1m a-1 in flux divergence. When the number of radar-sounding tracks is reduced, the performance of KR decreases more rapidly than for MC. Our study site shows that MC is capable of maintaining precision levels of 60m at 400m posting with flight tracks separated by 5 km. © 2014 Publishing Technology.