tomographic processing of alptomosar airborne data for...

AlpTomoSAR: observing the internal structure of Alpine glaciers Fringes 2015 - Frascati

Tomographic processing of AlpTomoSAR airborne data for observing the internal structure of Alpine glaciers: algorithm

description, challenges, and future perspectives

Stefano Tebaldini, Thomas Nagler, Helmut Rott, Achim Heilig , Adriano Meta, Alex Coccia, Davide Giudici, Dirk Schüttemeyer, Malcolm Davidson


Introduction

Track 1

Reference Track

(Master)

Track n

height

θ

π/2

TomoSAR imaging access to the 3D structure of the illuminated media

Tomography is a fundamental tool to:

o Investigate the phenomenology of Radar scattering

o Guide physical modeling through direct observation

Mariotti d'Alessandro, M.; Tebaldini, S., GRSL, 2012


from SAOCOM CS Science Report, ESA, 2015

Introduction

Spaceborne TomoSAR is intensively being studied for forest application

Tandem-L, SAOCOM-CS

o Currently under evaluation o Possible launch in 2018 for SAOCOM-CS o L-Band o Simultaneous InSAR pairs

BIOMASS

o Selected as the next ESA Earth Explorer Core Mission o Expected launch in 2020 o P-Band o Repeat pass

Ho Tong et al., TGRS, 2012


Introduction

What about ice ?


Introduction

What about ice ?

AlpTomoSAR: First campaign entirely focused on the study of the internal structure of glacier ice using TomoSAR

o Analogy/complementarity/synergy w.r.t. Nadir looking systems

TomoSAR

o Spatial Coverage

o Sensitivity to rough internal surfaces

Sounder/GPR:

o Accurate vertical layering on transects

o Sensitivity to specular scattering (interfaces)

o Added value w.r.t. single baseline inversion

InSAR & PolInSAR single baseline surveys yield sensitivity to scattering depth, not 3D structure

o Assessment of the added value of spaceborne TomoSAR surveys over glaciers

Key questions:


Test site

Test site: Mittelbergferner, Austrian Alps

o temperate glacier at the main ridge of the Alps in Tyrol

o main test area is a flat plateau in the upper part of the glacier between 3000 and 3200 m


Field Works

Field activities

o Setting up Corner Reflectors

o Stratigraphy of winter snow pack

Density / Hardness, ice layers, grain size

o Transects of snow depth

o GPR Measurements

GPR Equipment: o IDS dual-frequency 200/600 MHz o Total length of GPR profile: 18 km


SAR Acquisitions

North-South

South-North

SAR Equipment: o FMCW SAR by MetaSensing o Transmitted bandwidth: 150 MHz o Central frequency: 1275 MHz o Fully polarimetric o Spatial resolution ≤ 2 x 2 m (ground

range, azimuth)

Flights: o Two flight directions o 20+20 passes

Aircraft o CASA C-212 operated by BLOM-CGR


Flight Trajectories

Strong trajectory perturbation caused by wind/

turbulence

o Max along track variation > 50 m

o No auto-piloting system

o Proximity to mountain peaks

-2000

-1000

0

1000

2000

-50

0

50

1004450

4500

4550

4600

azimuth [m]ground range [m]

heig

ht

[m]


2D Focusing

2D Focusing via Time Domain

Back Projection on a reference

DEM

o Optimal motion compensation

o Common target wavenumbers in

all passes

o Automatic coregistration at the

reference elevation

HH – dir 1

HH – dir 2


Phase Calibration

TomoSAR provides resolution in elevation by jointly focusing data from multiple baselines

Phase screens result in signal defocusing

o Spaceborne: tropospheric and ionospheric phase screens

o Airborne: errors of the navigational system on the order of a fraction of a wavelength

Phase calibration through Phase Center Double Localization

joint estimation of target and aircraft positions

ground range

hei

ght

Data covariance matrix (NxN)

p

ref

p

n

p

n rr ˆ4

ˆ

ground range

hei

ght

p

refr

p

nr̂

p

mr̂

(y,z)

(Y,Z )


Phase Calibration

InSAR analysis of 15 x 15 interferometric pairs


Phase Calibration


Phase Calibration Estimated trajectory errors

-1500 -1000 -500 0 500 1000 1500 -1

-0.5

0

0.5

1

azimuth [m]

Trajectory error - ground range [m]

-1500 -1000 -500 0 500 1000 1500 -1

-0.5

0

0.5

1

azimuth [m]

Trajectory error - height [m]


Phase Calibration


3D Focusing

“Extreme” acquisition geometry:

o Large baselines Large range offset variations

o Large yaw angle variations Doppler plane

variations

Range and azimuth coregistration strongly depends

on initial focusing height

z1

z2

Focusin

g

heig

ht


3D Focusing

3D Focusing via Time Domain Back Projection

o Optimal treatment of large height-dependent range and azimuth coregistration offsets

o Preservation of common horizontal wavenumbers

o Automatic 3D geocoding

o The computational burden of this step was greatly reduced by adapting the sampling frequency of

defocused data to the effective amount of azimuth displacement induced by yaw angle variations


3D Focusing

Lidar DTM

Lidar DTM


Ice Velocity Correction

Wave propagation through ice

o Wave refraction

o Changing delay-to-distance conversion law

Targets are not focused at their geometrical

position

True target position

Apparent target position


Ice Velocity Correction

Wave propagation through ice

o Wave refraction

o Changing delay-to-distance conversion law

Targets are not focused at their geometrical

position

True target position

Apparent target position

TomoSAR correction

1. Mapping between true and apparent position

based on Fermat's minimum time principle

2. Subsequent Tomogram interpolation

minimum time travel path


Physical Analysis - CRs gr

ou

nd

ran

ge

azimuth

CRs appear higher than surface scattering Surface scattering from ice/snow interface


Physical Analysis – Lower Horizon

o Clear signal from the ice/snow interface in co-polarized channels

o Clear signal from down to 60 m beneath in all polarizations


Physical Analysis – Lower Horizon

o Lower horizon retrieval through surface-picking


Comparison to 600 MHz GPR Transects

o TomoSAR vs GPR comparison by sampling TomoSAR cubes

along GPR transects

o 600 MHz GPR transects processed down to 25 m

o TomoSAR transects processed down to 60 m

600 MHz GPR - 140227 AH

distance

depth

[m

]

300 400 500 600 700 800 900

0

5

10

15

20

25

TomoSAR - Direction 1 - HV

distance

he

igh

t w

.r.t

.lid

ar

[m]

300 400 500 600 700 800 900 -60

-50

-40

-30

-20

-10

0


distance heig

ht

w.r

.t.lid

ar

[m]

300 400 500 600 700 800 900 -60

-50

-40

-30

-20

-10

0

Bedrock

Direction 1 Direction 2


600 MHz GPR - 140227 AC

distance

depth

[m

]

600 700 800

0

5

10

15

20

25

TomoSAR - Direction 2 - HH

distance

heig

ht w

.r.t.lid

ar

[m]

600 700 800 -60

-50

-40

-30

-20

-10

0


distance heig

ht w

.r.t.lid

ar

[m]

600 700 800 -60

-50

-40

-30

-20

-10

0


Firn

transitions from ice to

soaked firn (???)

Direction 2 o TomoSAR vs GPR comparison by sampling TomoSAR cubes

along GPR transects

o 600 MHz GPR transects processed down to 25 m

o TomoSAR transects processed down to 60 m



200 MHz GPR - 140227 AF

distance

depth

[m

]

0 500 1000 1500 2000 2500

0

20

40


distance

heig

ht w

.r.t.lid

ar

[m]

0 500 1000 1500 2000 2500 -60

-40

-20

0


distance

heig

ht w

.r.t.lid

ar

[m]

0 500 1000 1500 2000 2500 -60

-40

-20

0



200 MHz GPR - 140227 AF

distance

depth

[m

]

0 500 1000 1500 2000 2500

0

20

40


distance

heig

ht w

.r.t.lid

ar

[m]

0 500 1000 1500 2000 2500 -60

-40

-20

0


distance

heig

ht w

.r.t.lid

ar

[m]

0 500 1000 1500 2000 2500 -60

-40

-20

0

Firn area Crevasses Firn area

Bedrock/ground reflection


Conclusions

o Analogy/complementarity/synergy w.r.t. Nadir looking systems

Different subsurface features could be detected by both GPR and TomoSAR

Wave penetration down the bedrock ( ≈ 60 m)

o Added value w.r.t. single baseline inversion

3D imaging allowed observation of the complexity of ice internal structure (over depth-only

information)

o Assessment of the added value of spaceborne TomoSAR surveys over glaciers

Need to fully assess spaceborne performance in all environmental conditions

Potentially unprecedented mapping of ice features worldwide using the next generation of

spaceborne TomoSAR systems

Key questions:

AlpTomoSAR: First campaign entirely focused on the study of the internal structure of glacier ice using TomoSAR

tomographic processing of alptomosar airborne data for...

Documents