TU3.L09 - Critical Assessment of diverse Polarimetric SAR Systems – pros and cons

Communications, Sensing & Navigation Lab

WIDEBAND INTERFEROMETRIC SENSING AND IMAGING POLARIMETRY 1

Hawaiian Village, Honolulu, Hawaii, 2010 July 25 – 30

IGARSS10-TU3.L09-4414, Sea Pearl 1/2/3Tuesday, 2010 July 27, 13:35 – 15:15

"Critical Assessment of diverse Polarimetric SAR Systems – pros and cons”

Wolfgang-Martin Boerner1

Invited Presentation

1. UIC-ECE/CSN, 900 W Taylor St, SEL-4210, CHICAGO, IL/USA-60607-7018, + I-312-996-5480, [email protected]

IEEE International Geosciences& Remote Sensing Symposium

mailto:[email protected]


• Wolfgang-Martin Boerner1, Thomas L. Ainsworth2, Eric Pottier3, Ya-Qiu Jin 4, Shane Robert Cloude5, Yoshio Yamaguchi6, Jakob J. vanZyl7, Konstantinos (Kostas) P. Papathanassiou8, Carlos Lopez-Martinez9

• 1. UIC-ECE/CSN, 900 W Taylor St, SEL-4210, CHICAGO, IL/USA-60607-7018, + I-312-996-5480, [email protected]

• 2. NRL-RSD/ISS, Code 7263, 4555 Overlook Ave SW, Bldg-2, WASHINGTON, DC/USA-20375-5351, [email protected]

• 3. Univ-Rennes-1, IETR-SAPHIR, Beaulieu Bat 11D, 263 ave Gen. Leclerc, F-35700 RENNES, FR, +33-2-23235763, eric.pottier@univ rennes1.fr

• 4. Fudan Daxue, East Guan Hua, Floor 11, Room 1103, 220 Handan Road, Yangpu, Shanghai, PRC-200-433, [email protected]

• 5. AECL, 26 Westfield Avenue, Cupar, Fife KY15-5AA, Scotland UK, [email protected] • 6. NU-IE, Ikarashi 2 Nocho 8050, NIIGATA-Shi, 950-2128, +81-25-262-6752,

[email protected] • 7. NASA-JPL, CALTECH MS 180-804, 4800 Oak Grove Dr, PASADENA, CA/USA-

91109-8099, +1-818-354-1365, [email protected]• 8. DLR-HR, Muenchener-Str. 20, Geb-102, D-82230 OPH-WESSLING, Obb, GER,

+49-8153-28-2306, [email protected] • 9. UPC-TSC/ARS-Group, North Campus, Bldg. D3, Room 203, Jordi Girona 1 – 3,

Barcelona, Spain, ES-08034, [email protected]




mailto:eric.pottier@univ%1Frennes1.fr

mailto:eric.pottier@univ%1Frennes1.fr








ABSTRACT-1

Considerable advances have been made during the past decades in polarimetric SAR systems leading to increased ability for recovering inherent polarization information conveyed by vector-electromagnetic wave backscatter. This paper compares the benefits offered by the major types of systems in relation to their application, as a function of their polarimetric architecture. POL-SAR system characterization includes the radar sensor, processing to transform the received data to polarimetric products, and calibration. More complex scattering scenarios require more capable polarimetric data collection and analysis. The system types considered are: Mono-Pol (single polarization, amplitude-only); Dual-Pol (traditional amplitude-only configuration, including HH+VH, VV+HV); Compact-Pol (transmit one polarization in base AB, and coherently receive two orthogonal polarizations in orthogonal base CD, which retain the relative phase between the received polarizations); and Full-Pol (coherent HH, HV, VH and VV).



ABSTRACT-2Under the simplifying assumption of scattering symmetry in monostatic configurations implying HV=VH, the Full-Pol data reduce to the familiar Quad-Pol products (coherent HH, HV and VV). Compact-Pol architectures include: Coherent Linear LH-Pol (HH+HV, or VV+VH, or HH+VV, coherently in each case); Coherent Diagonal/Linear DL-Pol (transmit linear polarization at a diagonal angle of 45* with respect to horizontal, and receive coherently the conventional linear H and V, thus DH+DV); Coherent Circular LR-Pol (coherent RR+RL or LL+LR); Coherent Hybrid Circular/Linear CL-Pol (RH+RV or LH+LV). The amount of acquired data (and of all data-dependent subsystems such as storage and transfer), processing and calibration increases, for greater polarimetric sophistication; and so do mass and volume with greater polarimetric capability. The various architectures have different implications on data rate, swath-width and resolution issues that apply to any multi-channel radar whether polarimetric, interferometric or for multiple frequencies. Given this multi-dimensional possibility space, the paper attempts to identify the benefits of each type of system as a function of implementation, also addressing the forthcoming demands of space-borne POLin SAR and RP (Diff) POLinSAR deployments.



ABSTRACT-3

Since SEASAT (Mono-Pol HH, L-band), space-based SAR systems are gradually becoming polarimetrically more capable, including ENVISAT (Dual-Pol, C-band), ALOS-PALSAR (Full-Pol or Quad-Pol, depending on processing algorithm, L-band), TerraSAR-X (Full-Pol or Quad-Pol, X-band), and RADARSAT-2 (Full-Pol or Quad-Pol, C-band). Planetary examples include Magellan (Venus: HH, S-band), Cassini (Saturn: Mono-Pol – amplitude-only, Ku-band), two Mini-SARs (Moon: CL-Pol, S-band or S- and X-band), for which mass and data rate (or data volume) are critical parameters.

Polarimetric diversity implies additional costs and impacts the mission operation scenario defined by user requirements and/or technological constraints. However, the paper hazards recommendations for the polarimetric architecture of future space-based SAR Systems; and it is concluded that any so-called cost-saving measures are ill-conceived and that we need to focus fully on the advancement of fully polarimetric SAR systems technology and stop all of the regressive approaches which will only take us back to the child stage of SAR concept initiation.



Contributing remarks by Dr. Lopez-Martinez

Respect to your comments about compact-pol it is true that it is a step back concerning PolSAR. For instance, in terms of data classification, the lack of complete polarimetric information may lead to misclassification. From my experience, people are demanding operational classification. So, if we do not have fully polarimetric information, we will not be able to demonstrate this capability.

Nevertheless, there is a more "dramatic" reading of the history of compact-pol vs full-pol. Final users are not polarimetric experts, so they are not able to distinguish compact-pol from full-pol. So, if they see poor results from compact-pol data, they will assume the same type of results with full-pol. I guess, it should be necessary to show what is lost from full-pol to compact-pol.

ABSTRACT-4



Polarimetric Imaging Radar Hierarchy R.K.RaneyRadar Result

Orthogonal Tx polsCoherent Dual Rx

One Tx Pol, Coherent Dual Rx

One polarization

Processing Nomenclature

Real image

No assumptionsReciprocity &

symmetry

4x4 scattering matrix

3x3 scattering matrix

Symmetry assumptions

No symmetry assumptions

Pseudo 3x3 scattering matrix

2x2 covariance matrix

Full polarization Quadrature polarization

Compact polarization

Two Rx pols

Two Tx pols

Magnitude

2 magnitudes & co-pol

phase

2 magnitudes

2 magnitudes

Like- and Cross-pol

images

2 orthogonal Like-pol images

2 orthogonal Like-pol images & CPD

Dual polarization

Mono-polarization

No HV



HH HVVV

Compact-RH Compact-RV Touzi AnisotropyFull-Pol

Ship Detection Using the Convair-580 (30 to 60 incidence angle)Touzi et al, “Ship detection and characterization using polarimetric SAR”. CJRS, Special issue on RADARSAT-2, June 2004



Red: Sendai7602_HH polarization

Green: Sendai7603_HH polarization

Blue: Sendai7604_HH polarization

N



Square path data

Flight direction


Ascending

2006/8/17ALPSRP029970850-1.1A 2006/10/2ALPSRP036680850-1.1A

ALOS-PALSAR Polarimteric Mode

TomakomaiHokkaido

Descending

ALPSRP030192750-1.1D2006/8/19

ALPSRP091090850-1.1A2007/10/10

Yoshio Yamaguchi

©JAXA, METI



POLSAR image analysis

<Average>

Pauli Basis

Eigenvalue

Scattering Power DecompositionCovariance matrix Coherency matrix

Entropy, Alpha-angle, Anisotropy

Scattering matrix= Quad. Pol. data

Pd, Pv, Ps, Pc

VVHVHHColor-Composite

HH-VV, 2HV, HH+VV

λ1 λ2 λ3

HV Basis

HH, 2HV, VV

Ps Pv

Pd

volume scattering

double bounce

surface scattering




The four-component decomposition of scattering powers Ps, Pd, Pv, and Pc


ALPSRP072570650-1.1A

32.825N130.364E

2007/6/5

©JAXA, METI

Fugen-dakeUnzen

Google earth optical image

ALOS-PALSAR pol. image

Ps Pv

Pd




Rader line of sight

Deorientation

Rotation of imsge

4-compornent scattering power decomposition algorithm using rotated coherency matrix



Four-component decomposition New rotated decomposition

Scattering power decomposition by rotation of coherency matrix for Niigata City area in Niigata Prefecture of Japan



Deorientation

(a) Original decomposition (b) Decomposition after T33 rotation

(c) Patch A: orthogonal urban (d) Patch B: oriented urban (e) Patch C: forest

Sapporo City, Hokkaido Prefecture, Japan



Interest area

0m

3800m

Data fusion of DEM and RADARSAT SAR images By CSRSR.

Monitoring of ongoing surface deformation along Cheleng-Pu fault

Taiwan



“Natural hazards are inevitable. Natural disasters are not.”


©JAXA, METI

Scattering power decomposition

Ps Pv

Pd

Pauli-basis

HV-basis

2007/3/10

-7.942N112.870E

Indonesia

T33 Rotation

ALPSRP059887030-P1.1__A

HH, 2HV, VV　 (50 up)

HH-VV, 2HV, HH+VV　 (80 up)

Pd, Pv, Ps　 (80 up)




Ascending

Data no.ALPSRP178330260

ALOS-PALSAR Polarimetric Mode

Philippines

© METI, JAXA

Yoshio Yamaguchi

2009/5/30

13.501N123.551E



Ps Pv

Pd

Scattering power Decomposition

Google Earth optical image

Decomposed image (Ps, Pd, Pv) with rotation 2*12 window


Philippines

2009/5/30

13.501N123.551E

©METI, JAXA

N

Mt. Mayon



2009/5/30

2010/1/15

2010/4/17

Mt. MayonPhilippines

Ps Pv

Pd



FOUNDATIONS AND RELEVANCE OF MODERN EARTH REMOTE SENSING & ITS ACTIVITIES

Conclusions:

The Electromagnetic Spectrum: A Natural Global Treasure

Terrestrial Remote Sensing with PolSAR: The Diagnostics of the Health of the Earth

at all weather and volcanic conditionsand at day and night

TU3.L09 - Critical Assessment of diverse Polarimetric SAR Systems – pros and cons

Documents

Transcript of TU3.L09 - Critical Assessment of diverse Polarimetric SAR Systems – pros and cons