Quantifying Ice Sheet Dynamics and Variability with Meter-scale DEM and Velocity Products

David E. Shean1, Ian R. Joughin1, Ben E. Smith1, Zachary M. Moratto2, Claire R. Porter3, Paul Morin3 1 Applied Physics Lab, University of Washington, Seattle, WA, 98105, dshean@uw.edu

2 NASA Ames Research Center, Mountain View, CA 3 Polar Geospatial Center, University of Minnesota, Minneapolis, MN

•  Large speedup and thinning ~4-9 m/yr with >10 km of grounding line retreat since ~1996 [12,13]. •  Processed/analyzed 14 WorldView DEMs from 2010-2012.

•  Produced elevation diff. and velocity maps near grounding line on 1/13/11, 2/23/11, 12/28/10. •  Eulerian dh/dt maps capture translation of surface variations for floating ice, thickness change for

grounded ice.

•  Lagrangian dh/dt maps show advection of grounded ice, and thickness changes for floating ice due to basal melt, firn compaction, longitudinal extension.

•  Grounding zone can be inferred from transitions in eulerian and lagrangian dh/dt.

Summary/Conclusions •  Automated, open source processing pipeline for 0.5 m/px commercial imagery. •  High-res (2-4 m/px) DEM/velocity timeseries provide unprecendented observations of ice sheet

dynamics, with optimal sub-meter horizontal/vertical accuracy. •  Preliminary timeseries capture:

•  Seasonal elevation changes of +/- 50 m/yr for lower Jakobshavn between 2010-2012.

•  Apparent PIG grounding line retreat since 2009, ice shelf thinning in 2011. •  Data for complete timeseries has been ordered

We are using complementary satellite and airborne data to investigate ice-sheet/shelf dynamics and variability on seasonal to interannual timescales. High-resolution, along-track stereo satellite imagery provides unprecedented spatial (~0.5 m/px) and temporal (daily-monthly) resolution from 2009-present. We have developed an automated, open source processing pipeline to generate orthoimage, digital elevation model (DEM), velocity, and elevation change products from these data with optimal sub-meter accuracy.

We are producing time series for rapidly changing regions of the Greenland and Antarctic ice sheets. Preliminary results are presented for Jakobshavn Isbrae, Greenland and Pine Island Glacier (PIG), West Antarctica – two outlet glaciers where rapid acceleration and thinning are well-documented in recent decades. Our preliminary results show large seasonal elevation variations at Jakobshavn and grounding line retreat accompanied by ice shelf thinning at PIG.

Abstract

Introduction

Data & Methods Commercial stereo imagery •  DigitalGlobe/Geoeye: 5 polar-orbit satellites. •  0.46 m/px panchromatic, ~17 km swath width. •  Along-track stereo acquisition, ~1 min apart. •  1000s of cloud-free pairs, high-priority

locations with >10-20x repeat obs. •  Meter-scale surface texture (e.g. sastrugi)

provide near-perfect image correlation, not possible with low-res data.

•  From ~1990-2010, outlet glacier velocity increases of 50-150% [1,2] •  Thinning ~10-25 m/yr in Greenland, ~4-9 m/yr in W. Antarctica [3].

•  Increased discharge and negative mass balance [4,5]. •  ICESat altimeter failed in 2009, ICESat-2 not operational until ~2016,

IceBridge spatial/temporal coverage limited.

•  Repeat observations necessary to measure ice sheet mass loss and understand processes driving observed changes.

DEM Generation •  ~2-4 m/px output DEM resolution. •  Automated, open-source, command-line workflow. •  NASA Ames Stereo Pipeline core. •  Multithreaded, ~1-48 hours per pair on 8-core

workstation, depending on setup/inputs. •  Python utilities (GDAL/OGR, NumPy/SciPy/Matplotlib)

for pre/post-processing and analysis.

Pine Island Glacier

Velocity Map Generation •  Input aligned DEM products or orthoimages. •  Seed search window with low-res velocity products. •  Sub-pixel (~1/10) correlation provides X/Y offsets for

every input pixel.

Elevation Change Map Generation •  Eulerian elevation difference (dzE) for fixed reference frame. •  Lagrangian elevation difference (dzL) using pixel offsets

from concurrent velocity maps to track parcels. •  dzE = 0 for grounded ice in steady-state. •  dzL = 0 for floating ice with no melting/stretching.

References 1. Howat, I. M., Joughin, I. & Scambos, T. A. Rapid Changes in Ice Discharge from Greenland Outlet Glaciers. GRL 315, 1559–1561 (2007). 2. Joughin, I., Abdalati, W. & Fahnestock, M. Large fluctuations in speed on Greenland’s Jakobshavn Isbrae glacier. Nature 432, 608–610 (2004). 3. Pritchard, H. D., Arthern, R. J., Vaughan, D. G. & Edwards, L. A. Extensive dynamic thinning on the margins of the Greenland and Antarctic ice sheets. Nature 461, 971–975 (2009). 4. Rignot, E. & Kanagaratnam, P. Changes in the Velocity Structure of the Greenland Ice Sheet. Science 311, 986–990 (2006). 5. Rignot, E. et al., Acceleration of the contribution of the Greenland and Antarctic ice sheets to sea level rise. GRL 38, L05503 (2011). 6. Rignot, E., Mouginot, J. & Scheuchl, B. Ice Flow of the Antarctic Ice Sheet. Science 333, 1427–1430 (2011). 7. Moon, T., Joughin, I., Smith, B. & Howat, I. 21st-Century Evolution of Greenland Outlet Glacier Velocities. Science 336, 576–578 (2012). 8. Joughin, I. et al. Continued evolution of Jakobshavn Isbrae following its rapid speedup. JGR 113, F04006 (2008). 9. Joughin, I. et al. Seasonal to decadal scale variations in the surface velocity of Jakobshavn Isbrae, Greenland: Observation and model-based analysis. JGR 117, (2012). 10. Csatho, B., Schenk, T., Van der Veen, C. & Krabill, W. B. Intermittent thinning of Jakobshavn Isbrae, west Greenland, since the Little ice age. J. Glaciology 54, 131–144 (2008). 11. Podrasky, D. et al. Outlet glacier response to forcing over hourly to interannual timescales, Jakobshavn Isbræ, Greenland. J. Glaciology 58, 1212–1226 (2012). 12. Rignot, E. J. Fast Recession of a West Antarctic Glacier. Science 281, 549–551 (1998). 13. Joughin, I., Rignot, E., Rosanova, C. E., Lucchitta, B. K. & Bohlander, J. Timing of Recent Accelerations of Pine Island Glacier, Antarctica. GRL 30, (2003). .

Jakobshavn Isbrae

Timeline of cloud-free WorldView stereo imagery (blue) and ATM data (red). Note the availability of multiple images in 2011 and 2012, which have been ordered, but were not available for the preceding analysis.

InSAR Velocity, ~2007-2010 [6,7], ICESat surface elev. change, 2003-2007 [3].

DEM Alignment and Correction •  Uncorrected georeferencing error typically <5 m. •  Horizontally/vertically co-register “keystone” DEM to ATM/

ICESat altimetry data over control surfaces (bedrock). •  Align/mosaic additional DEMs with reference "keystone"

over control surfaces or low-velocity ice (<200 m/yr). •  Ongoing efforts to automatically correct residual along-

track CCD "striping" artifacts of <+/-0.5 m.

Pine Island Glacier DEM mosaics (A+B), dh/dt maps (C, D), velocity map derived from 4 m/px DEMs, and profiles for elevation (F) and elevation change (G). Black grounding line from 2009 TerraSAR-X tidal flexure. Note apparent thinning and grounding line retreat in circled region, consistent with expectations for basal melt.

Summer 2012

Winter 2011/2012

WorldView-2 Level-1B stereopair, 8/26/2012.

-17.3 m/yr

-15.9 m/yr

-13.4 m/yr

-11.0 m/yrWorldView DEM Mosaic (7/9/10-7/11/10)

Abstract 1469750

Detail of shaded relief and derived velocity maps for PIG south shear margin. Note rigid block rotation.

•  Largest/fastest glacier in Greenland: drains ~6.5% of GRIS, velocity ~8-16 km/yr (~20-45 m/day) [8,9]. •  ~40 km calving front retreat since ~1850 [10]; 1998 loss of ~15 km ice tongue, 2x speedup by 2003 [2].

•  Since 2004, seasonal calving front advance/retreat ~5-8 km, primary driver of seasonal velocity variation [8,11]. •  Dynamic thinning increases driving stress upstream and decreases effective pressure [9].

•  Annual springtime IceBridge ATM data shows interannual thinning rates of ~5-15 m/yr [8,9], but cannot capture seasonal elevation change – seasonal thinning signal is poorly understood.

•  >20 cloud-free WorldView stereopairs from 2010-2012.

•  Mosaics produced for DEMs with time separation ~0-4 days.

DEM alignment/correction example over exposed bedrock. Elev. diff. 7/9/10-8/12/10 before and after correction, with residual along-track "striping" artifacts.

DigitalGlobe stereo coverage as of 7/12/12. Jakobshavn (J) and Pine Island Glacier (PIG) locations noted for context.

Mosaic of three WorldView DEMs from July 2010. Centerline shown was derived from InSAR velocity grids. Black holes are from failed correlation over proglacial lakes.

Top: Elevation data from IceBridge ATM L2 products (circles, 2003-2011) and WorldView DEMs (squares, 2010-2012). See mosaic (left) for context. Trendline shows interannual thinning rate at each location. Bottom: InSAR velocity data showing rapid speedup between ~1995 and 2003, with ongoing interannual increase and large seasonal variations [9]. Note relatively low velocities between ~6/2010 and 6/2011 and substantial thickening of +30-50 m for lower trunk. Also note unprecendented >2x seasonal speedup during summer 2012, and contemporaneous elevation decrease.

Eulerian elevation change rate (m/yr) products for lower Jakobshavn showing: A) 2010/2011 winter thickening, and B) cumulative elevation change from summer 2011, winter 2011/2012 and summer 2012. Color scale is identical. Extreme values are associated with advance/retreat of the calving front.

Elevation profiles extracted from WorldView DEMs between 2010-2012. Inset shows full profile, main axes limited to lower Jakobshavn trunk. Note seasonal calving front position and substantial seasonal elevation change between spring 2012 (green) and end of summer 2012 (red).

J PIG

DigitalGlobe stereo coverage. Repeat coverage density through 1/16/12. Color scale is cloud-free image count.

Icebergs