TH4.TO4.1.ppt

IGARSS 2011, Vancouver, Canada, 24 – 29 July 2011

TERRASAR-X OBSERVATIONS OF ANTARCTIC OUTLET GLACIERS IN THE ROSS SEA SECTOR

Kenneth Jezek1, Wael Abdel Jaber2,3, Dana Floricioiu2

1 Byrd Polar Research Center, Ohio State University, Columbus, OH, USA

2 German Aerospace Center (DLR), Remote Sensing Technology Institute, Oberpfaffenhofen, Germany

3 Technical University of Munich, Remote Sensing Technology, Munich, Germany

2Remote Sensing Technology Institute

Introduction

Ice surface velocity:

Crucial parameter for mass balance estimation of glaciers and ice sheets.

Fundamental factor to establish the state of glaciers and predict sea level variations.

Interferometric SAR data requirements for ice surface velocity:

Short repeat pass time span for high frequency SAR (C-, X-band) to avoid decorrelation

Left looking capabilities for areas South of 80°S. Available for: Radarsat-1/2 (1997/2008 campaigns), COSMO-SkyMed and TerraSAR-X.

Antarctic ice velocity: ERS-1/2 (1996), RSAT-1 (1997, 2000)

[Rignot et al, 2008]. Status 2000, update ongoing!

3Remote Sensing Technology Institute

TerraSAR-X mission

TerraSAR-X mission: X-band (9.65 GHz) Synthetic Aperture Radar (SAR)

1st satellite TSX-1 operational since Jan. 2008

2nd satellite TDX-1 (launched for TanDEM-X mission) operational since Oct. 2010 (for TerraSAR-X mission)

Left and right imaging capabilities

Various imaging modes and incidence angles

Spatial resolution in stripmap mode, single look complex data: < 3 m in slant range and azimuth

Very accurate orbit estimation

11 days repeat pass cycle

During the 2007-2008 International Polar Year (IPY) DLR initiated the acquisition and processing of interferometric TerraSAR-X data over important sectors of the polar ice sheets.

TSX-1

TSX-1 & TDX-1

4Remote Sensing Technology Institute

TerraSAR-X acquisition plan over the Ross Sea Sector of Antarctica

•Left looking•Stripmap•Swath width: 30 km•Resolution: <3 m•Descending coverage•Background: RSAT-1

Nimrod

Beardmore

80°S

Byrd

Shackleton

AmundsenScott

Van der Veen and Whillans

Kamb

Bindshadler

Crary

Reedy

MacAyeal

Ross Ice Shelf

Courtesy: Katy Farness, OSU

N

5Remote Sensing Technology Institute

Nimrod Glacier (82.6°S, 160.7°E)

Two main tributaries: Nimrod and Marsh Glacier

Length: 135 km

Glacier in balanced state: mass balance 0.88 ± 0.39 Gt/a (after L. Stearns, 2010, in press)

N

Ross Ice Shelf

az

rg

Marsh Gl.

Kon-TikiNunatak

TerraSAR-X image (2009)

Nimrod Gl. flowing around the Kon-Tiki Nunatak

TerraSAR-X mosaic (2009)

http://vervoortantarctica.blogspot.com/

6Remote Sensing Technology Institute

Byrd Glacier (80.5°S, 157.5°E)

One of the largest catchment basins in Antarctica (1 101 725 km2)

Fjord length: 136 km, width: 24 km

Ice export into the Ross Ice Shelf: 22.32 ± 1.72 Gt/a

az

rg

Byrd Gl. view upstream from Mt. Quackenbush

TerraSAR-X mosaic (2010)

Ross Ice Shelf

N

7Remote Sensing Technology Institute

Ice velocity derivation by speckle tracking

Displacement for each patch is obtained from the position of the peak of the incoherent cross-correlation function (CCF).

A regular grid of patches with size of 2562 pixels and 50% overlap is used.

Input: Stripmap geocoded images: 1.25 m pixel spacing, ~1.3 looks, <3 m sp. resolution.

Correlation between repeat pass images is given by the speckle pattern (smooth, highly coherent areas) or by the image texture (e.g. crevasses).

CCF: slow and smooth area CCF: fast and crevassed area

W-E cut

N-S cut

0.156 m

1.328 m

R: master (m) G: slave (s) B: (m+s)/2

W-E cut

N-S cut

15.936 m

9.844 m

R: master (m) G: slave (s) B: (m+s)/2

11 days repeat pass 11 days repeat pass

8Remote Sensing Technology Institute

Speckle tracking: TerraSAR-X velocity map detail

0.0

2.5

[m/d

]

2 m/d

Patch size: 2562 pixel (3202 m2)Patch distance: 128 pixel (160 m)

0 500m

9Remote Sensing Technology Institute

Speckle tracking (ST) accuracy with TerraSAR-X

Absolute accuracy of velocity estimate is affected by:• Atmospheric path delay

• Solid earth tides

• Orbit estimation and timing errors

• Geocoding errors

Relative accuracy of an individual motion vector depends on:

• Correlation coefficient γ

• Patch size N

• Shape of the cross-correlation function

• Estimated empirically for highly coherent smooth areas [1]. ~10 times lower than absolute accuracy

πγγσσ

2

CRST

1

2

322

−=≅

N

[1] R. Bamler, M. Eineder, “Accuracy of differential shift estimation by correlation and split-bandwidth interferometry for wideband and delta-k SAR systems,“ IEEE Geoscience and Remote Sensing Letters, Vol. 2, No. 2, April 2005

( ) m/day 029.0111.02.02 22vel =+=σ

σrg=0.2 m, σaz=0.1 m

σCR: Cramer-Rao bound

More in presentation: „Glacier flow and topography measurements with TerraSAR-X And TanDEM-X “by Michael Eineder (session TH4.T02.5, Room 1, 4:40 PM)

10Remote Sensing Technology Institute

Geophysical glacier properties derived from velocity profiles

Longitudinal strain rate Obtained from the filtered longitudinal velocity profile applying [1]:

Tensile strength Computed for points of crevasses initiation (according to amplitude image)

Assumption: only longitudinal component of strain is present on glacier‘s center line

Application of Glen‘s flow law of ice + „von Mises“ and „Griffith‘s“ failure criteria, leads to:

d

v

∆∆=ε Δv: velocity variation between two points

Δd: distance between two points (Δd ≈ 63 m)

[1] Forster, R., Rignot, E., Isacks, B., Jezek, K., „Interferometric radar observation of Glaciers Europa and Penguin, Hielo Patagónico Sur, Chile“, Journal of Glaciology, Vol. 45, No. 150, 1999

3

1

M 3TS

=

A

ε A: flow parameter, depends on temperature, impurities and crystal orientation of ice (A = 1.61∙10-9 yr-1 kPa-3)

3

1

G 2TS

=

A

ε

11Remote Sensing Technology Institutekm0 30

N

Marsh Gl.

Kon-TikiNunatak

GL

0.0

2.5

[m/d

]

Ross Ice Shelf

(L)

(T)

Plug like shape: strong side drag

23 Stripmap pairs, coverage: ~35 000 km2

Acquired: Oct. - Nov. 2009

Nimrod Glacier: ice surface velocity from TerraSAR-X

(T)

SE NW

255.5

219.0

182.5

146.0

109.5

73.0

36.5

0.0

[m/a

]

(L)Ice flow

GL

912.5

730.0

547.5

365.0

182.5

0.0

[m/a

]

12Remote Sensing Technology Institute

Nimrod Glacier: longitudinal strain rate

km0 30

N

Marsh Gl.

Kon-TikiNunatak

0.0

2.5

[m/d

]

Ross Ice Shelf

(L)

(T)GL

(1)

(2)

(L)Ice flow

GL

(1)

(2)

(L)GL

(1)

(2)

912.5

730.0

547.5

365.0

182.5

0.0

[m/a

]

487.0 - 562.4204.1 - 235.7TSM – TSG [kPa]

0.03560.0026Strain rate [a-1]

0.720.27Velocity [m/day]

119.046.2Distance [km]

(2)(1)Point

13Remote Sensing Technology Institute

N

GL

km0 30

Ross Ice Shelf

(T2)

(T1)

(L)

0.0

2.5

[m/d

]

29 Stripmap pairs, coverage: ~44 300 km2

Acquired: Nov. 2010 – Feb. 2011

Byrd Glacier: ice surface velocity from TerraSAR-X

(L)Ice flow

GL

912.5

730.0

547.5

365.0

182.5

0.0

[m/a

]

(T1)(T2)

SE NW

912.5

730.0

547.5

365.0

182.5

0.0

[m/a

]

Plug like shape: side drag

Parabolic shape: basal drag

14Remote Sensing Technology Institute

Byrd Glacier: longitudinal strain rate

N

GL

km0 30

Ross Ice Shelf

(T2)

(T1)

(L)

0.0

2.5

[m/d

]

(1)

(2)

(3)

GL

(L)

(1)(2)

(3)

(L)Ice flow

GL

(1)(2)

(3)

912.5

730.0

547.5

365.0

182.5

0.0

[m/a

]

292.9 - 338.2

0.0077

1.48

161.0

(3)

226.68 - 261.7171.1 - 197.6TSM – TSG [kPa]

0.00360.0015Strain rate [a-1]

0.400.20Velocity [m/day]

70.03.0Distance [km]

(2)(1)Point

15Remote Sensing Technology Institute

Summary and conclusions

Two large outlet glaciers of the Transantarctic Mountains (TAM), Nimrod and Byrd, have been studied with TerraSAR-X repeat pass images.

The TAM area south of 80°S is poorly known due to the limited availability of satellite data.

The acquisition plan in left looking mode was initiated as a major contribution of DLR to the 2007-2008 International Polar Year.

Ice surface velocity maps and profiles with very high resolution were obtained by means of speckle tracking: an accurate and robust method in these highly coherent areas.

Other important geophysical parameters, such as strain rate and tensile strength were also obtained.

The two glaciers show differences in ice flow behavior, influenced by:

• Upstream processes

• The topography of the glacier channel

• (partly) the evolutionary stage of each glacier associated with the slow retreat of the Ross Ice Shelf since the last glacial maximum.

Thank your for your attention!