Quantifying ice-shelf basal melt and ice-stream dynamics using high-resolution DEM and GPS time series

Thesis (Ph.D.)--University of Washington, 2016-06

This dissertation describes methods to generate high-resolution digital elevation models (DEMs) of the Earth's ice sheets, and combines these observations with in situ GPS measurements to study basal melting beneath the Pine Island Glacier ice shelf, Antarctica. Pine Island Glacier (PIG) is currently losing mass at a rate of ~40 Gt/yr and contributing ~0.1 mm/yr to global sea level rise. This mass loss has been attributed to rapid retreat, speedup, thinning, and increased discharge in recent decades, due to ocean forcing and/or internal instability. The automated, open source NASA Ames Stereo Pipeline (ASP) was adapted to generate digital elevation models (DEMs) and orthoimages from very-high-resolution (VHR) commercial imagery. I outline a processing workflow for ~0.5 m ground sample distance (GSD) DigitalGlobe WorldView-1/2/3 stereo image data. Output DEM products are posted at ~2 m with direct geolocation accuracy of <5.0 m CE90/LE90. An automated co-registration workflow reduces absolute vertical and horizontal error to <0.5 m, with observed standard deviation of ~0.1-0.5 m for overlapping, co-registered DEMs. I processed all available 2010-2015 WorldView/GeoEye DEMs over the PIG ice shelf, and integrated with other available 2002-2015 DEM/altimetry data. I analyzed Eulerian elevation change (dh/dt) over grounding zones and upstream ice, and developed novel Lagrangian elevation change (Dh/Dt) methodology for elevation measurements over floating ice. I combined these results with an annual mass budget analysis to quantify the spatial and temporal evolution of ice shelf baasal melt. This analysis reveals the complex spatial/temporal evolution and interconnection of grounding zones, sub-shelf cavity geometry, basal melt rates, and upstream dynamics over grounded ice. Rapid PIG grounding line retreat ended between ~2008-2009, followed by the ephemeral regrounding of ~2-3 deep keels as a positive ice shelf thickness anomaly advected over a seabed ridge. Thinning upstream of the grounding line decreased from ~5-10 m/yr in 2008-2010 to ~0 m/yr by 2012-2014, with a small grounding line advance from 2012-2015. Mean 2008-2015 basal melt rates were ~80-90 Gt/yr for the full shelf, with ~200-250 m/yr melt rates within large channels near the grounding line, ~10-30 m/yr over the main shelf, and ~0-10 m/yr over the North and South shelves, with the notable exception of ~50-100 m/yr near the grounding line of a fast-flowing tributary on the South shelf. I processed 2008-2010 and 2012-2014 GPS records for the PIG shelf and analyzed multi-path antenna heights, horizontal velocities, strain rates, cm-accuracy surface elevation and Lagrangian Dh/Dt elevation change. These data provide validation for the corrected stereo DEMs, with sampled DEM error of ~0.7 m. The GPS antenna height records document a relative surface increase of ~0.7-1.0 m/yr, which is consistent with estimated RACMO2.3 surface mass balance (SMB) and firn compaction rates from the IMAU-FDM dynamic firn model over the PIG shelf. Observed surface Dh/Dt is highly linear for all GPS records, with trends of -1 to -4 m/yr and <0.4 m residuals. Similar Dh/Dt estimates with reduced variability are obtained after removing expected downward GPS pole base velocity from GPS antenna Dh/Dt. Basal melt rates derived from GPS Dh/Dt are ~10 to 40 m/yr for the outer PIG shelf and ~4 m/yr for the South shelf. These estimates show good agreement with contemporaneous in situ measurements and stereo DEM records. Melt rates were highest for the 2008-2010 period, with a ~20-30% decrease by 2010-2012, followed by a gradual increase from 2010-2012 to 2013-2015. Melt rates vary significantly across ~km-scale ice shelf thickness variations, with focused melting in basal channels near the grounding line and keels over the outer shelf. The DEM and GPS records also document higher melt rates within and near transverse surface depressions/rifts associated with longitudinal extension. I suggest that these ~km-scale features alter sub-shelf circulation, leading to positive feedbacks that can influence regrounding and upstream ice dynamics. A positive linear relationship between melt rate and depth is observed, with increasing melt rate magnitude and increasing variability at depth. The slope and spread of this linear relationship varies over time. Existing piecewise melt rate parameterizations in prognostic ice flow models provide reasonable approximations for this relationship, but fail to capture km-scale variability. The DEM and GPS Dh/Dt melt products do not show the ~50% decrease in melt rates between 2010 and 2012 inferred from hydrographic observations in Pine Island Bay, with no significant melt rate variability associated with observed ~2012 ocean cooling in mooring records. This suggests that PIG melt rates are not directly correlated with observed ocean heat content near the shelf front, and that during the 2008-2015 period, observed ice shelf melt and upstream dynamics were more sensitive to grounding evolution, channel-scale circulation, and internal instability than oceanographic forcing. These findings have important implications for flow modeling efforts used for projections of 21st-century sea level rise.