Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

Retrieval of Forest Parameters from Multi-polarimetric and Multi-frequency

Interferometric SAR Data

Karlheinz Gutjahr1 and Mathias Schardt1,2

1Joanneum Research – Institute of Digital Image Processing, Graz, Austria2 University of Technology, Institute of Remote Sensing and Photogrammetry,

Graz, Austria

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 2

Test site “Kobernausser Wald”Forest test site:Kobernausser Wald

Geographic Coordinates:UL: 13°19’E / 48°04’NLR: 13°25’E / 48°00’N

Area Size: 3km x 9km

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 3

E-SAR data acquisition— Flight campaign on 11.05.2004

— Two corner reflectors installed

E-SAR data acquired over Kobernausser Wald test site

Dual

Dual

Single

Pass

10 mSLC & IntensityHH/HV/VH/VVL-band

20 mSLC & IntensityHH/HV/VH/VVP-band

0.5 mSLC & IntensityHH/VVX-band

BaselineData TypePolarisationFrequency

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 4

E-SAR data acquisitionP-band HH, HV, VVL-band HH, HV, VVX-band HH, HV, HH/VV

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 5

Basic methodologies— Polarimetry

» Use intensity (coherence) of backscattered signal» Empirical models

— Interferometry» Use phase of interferometric signal» Established technique

— Polarimetric SAR interferometry» Use complex coherency» Random Volume over Ground model (RVoG)

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 7

SAR interferometry— Approximation

( )sinr Bδ ζ θ= ⋅ −

( )( ) 0cos cosz h hhr r

θ θ−= = −

( ) ( )( )( )

( )

0

0

0

22 3

0

sin

cossin

cos...

2 sin

r h B

Bh

r

Bh

r

δ ζ θ

ζ θθ

ζθ

= ⋅ − +

⋅ −⋅ +

⋅

⋅⋅ +

( ) ( ) ( )( )

00

0

cos4 sin ...sinabs

Bh hr

ζ θπφ ζ θλ θ

⎡ ⎤−= ⋅ − + ⋅ +⎢ ⎥⋅⎣ ⎦

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 8

InSAR products at P-band

Wra

pped

Inte

rfero

gram

Inte

rfero

met

riche

ight

Am

plitu

de

( ) ( )2 2Im ReA I I= +

Coh

eren

ce

*, ,

1

2 2, ,

1 1

ˆ

N

m n sl nn

N N

m n sl nn n

c c

c cγ =

= =

=

⋅

∑

∑ ∑

( )( )

Imtan

ReII

φ = 2 absrkλ φπ

Δ =

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 9

Polarimetric SAR Interferometry

( )

[ ]0

11

21

32

3

expVh

Mmmm

ϕγ

σγγ

−

⎡ ⎤⎢ ⎥ℑ⎢ ⎥ ⎡ ⎤⎢ ⎥ ⎢ ⎥=⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥⎣ ⎦⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦

Random Volume over Ground model*

*K. P. Papathanassiou and S. R. Cloude: Single-baseline polarimetric SAR interferometry.In IEEE Transactions Geoscience Remote Sensing, Vol. 39, No. 11, 2352-2363, 2001.

hV = tree heightϕ0 = ground topographyσ = mean volume extinctionm = ground to volume scattering amplitudeγ = complex interferometric coherence[M] = scattering model operator

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 10

Pol-InSAR products at P-bandG

roun

dph

aseφ 0

Ext

inct

ionσ

Veg

etat

ion

laye

rhei

ghth

V

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 11

Laser scanner reference DHMsN

orth

ern

part

Sou

ther

n pa

rt

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 12

Analysis along profile

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 13

InSAR canopy height model (CHM)

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 14

Pol-InSAR canopy height model

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 15

Combined canopy height model

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 16

Correlation

Interferometry Combination

Correlation Coefficient: 0.64 Correlation Coefficient: 0.81

Workshop on 3D Remote Sensing in ForestryVienna, 14th-15th of February 2006

© Institute of Digital Image ProcessingJoanneum Research, Page - 17

Discussion and outlook— Pol-InSAR

» Only L- or P-band necessary» Preliminary results

— InSAR» Difference X-band DSM and P-band DTM» Penetration of P-band to ground not sufficient

— Combination» Best results with X-band DSM and reference DTM» Upcoming TerraSAR-X mission