On The Use Of Compact-Polarimetry SAR Architectures For Observing Sea Oil Slicks

A. Buono1, F. Nunziata1,2, X. Li3, Y. Wei2, X. Ding2

1 Università degli Studi di Napoli Parthenope, Centro Direzionale, isola C4 – 80143, Napoli, Italy

2 Shanghai Ocean University, College of Marine Science, Shanghai 201306, China

3 NOAA/NESDIS, Global Science & Technology, College Park, MD 20740, USA

Contact Author: andrea.buono@uniparthenope.it

Outline

Introduction

Polarimetric Model

Experimental Results

Conclusions

Introduction Observing oceans and coastal areas is of paramount importance for safety,

trading, tourism and environment.

Polarimetric Model

• Identification of oil slicks, i.e., to detect sea oil slick and to distinguish

them from a broad class of oil look-alikes (weak-damping surfactants); • Characterization of oil slicks, i.e., to provide at least rough information

on the spatial variability within the oil slick according to its damping properties;

Polarimetric Synthetic Aperture Radar (polSAR) measurements offer an unprecedented amount of physical information on the scattering processes that rule the observed scene. With respect to single-polarization SARs, polSARs allow significantly enhancing performance in terms of:

Polarimetric Model

• Scattering entropy; • Scene degree of polarization; • Standard deviation of the CPD; • Conformity coefficient; • Normalized pedestal height; • Etc.....

BRAGG NO BRAGG

A large set of polarimetric features can be extracted from polSAR measurements to reliably estimate the departure from slick-free sea surface Bragg scattering:

Polarimetric Model

Those issues can be partly overcome using new SAR architectures, namely Compact-Polarimetry:

• Simple oil detection; • Scattering-based discrimination; • Rough damping information;

• Limited area coverage; • Reduced AOIs; • Cross-pol signal ≈ NESZ;

ISRO RISAT-1 (C- band, 2012) JAXA ALOS-2 (L-band, 2014) CSA Radarsat Constellation Mission (C-band , 2018)

PolSAR vs Single-pol SAR: Pro & Cons

Polarimetric Model

Fully-polarimetric ALOS-PalSAR 1 vs Compact-polarimetric ALOS-PalSAR 2

Angles of incidence 8° - 30° Range resolution 26 m (at 21.5°)

Observation swath 30 km (at 21.5°) NESZ < -28 dB (at 21.5°)

Angles of incidence 8° - 70° Range resolution 6 m – 10 m

Observation swath 50 km – 70 km NESZ < -26 dB (at 37°)

Polarimetric Model

CP SAR architectures measure a projection of the scattering matrix into an orthogonal receiving basis (H-V). Hence, CP SAR data can be emulated from actual FP SAR measurement as follows:

HYBRID-POLARITY (HP):

RISAT-1 ALOS-2 (experimental)

SLANT LINEAR (π/4):

ALOS-2 (experimental)

Different CP modes can be implemented varying the polarization of the transmitted wave, given an orthogonal H-V receiving basis:

Polarimetric Model

STOKES VECTOR:

WAVE COHERENCY MATRIX:

Once a CP SAR measurement is available, second-order descriptors are needed to completely characterize the polarization state of the generic partially polarized wave backscattered off the observed scene and received at the CP SAR antenna:

Polarimetric Model Hence, several polarimetric features can be extracted to perform sea oil slick observation: Wave entropy, 0 < Hw < 1

Wave degree of polarization, 0 < Pw < 1

Ellipticity parameter, -1 < sin(2χ) < 1

Conformity coefficient, -1 < μw < 1

Polarimetric Model

LOW

HIGH Pw (≈1) μw (>0)

Hw (≈0) sin(2χ) (<0)

Hw (≈1) sin(2χ) (>0)

Pw (≈0)

μw (<0)

The CP features’ behaviour can be interpreted in terms of the polarization state of the wave backscattered off sea surface with and without oil slicks:

Experimental Results Two fully-polarimetric L-band Alos-PalSAR SAR acquisitions collected in the Gulf of Mexico are considered. The first scene is relevant to a well-known oil seep, while the second one is relevant to the Scorpion Reef area, where the 5 main vegetated small islands of the Campeche Bank Archipelago and a weak-damping surfactant likely due to the reef are present.

|VV|2 ALOS-PALSAR 2 (10000x2000 pixels)

WEAK-DAMPING LOOK-ALIKE

ISLANDS

|VV|2 ALOS-PALSAR 1 (18000x1200 pixels)

OIL SEEPS

Experimental Results: HP SAR architecture

Oil slick identification is accomplished using sin(2χ) and μw since they provide logical binary outputs without any external threshold:

HP sin(2χ), ALOS-2

(10000x2000 pixels)

WEAK-DAMPING LOOK-ALIKE OIL SEEPS HP sin(2χ),

ALOS-1 (18000x1200

pixels)

Experimental Results: HP SAR architecture

Oil slick identification is accomplished using sin(2χ) and μw since they provide logical binary outputs without any external threshold:

HP μw, ALOS-2

(10000x2000 pixels)

WEAK-DAMPING LOOK-ALIKE

OIL SEEPS HP μw, ALOS-1

(18000x1200 pixels)

Experimental Results: π/4 SAR architecture

Oil slick identification is accomplished using sin(2χ) and μw since they provide logical binary outputs without any external threshold:

π/4 sin(2χ), ALOS-2

(10000x2000 pixels)

WEAK-DAMPING LOOK-ALIKE

π/4 sin(2χ), ALOS-1

(18000x1200 pixels)

OIL SEEPS

Experimental Results: π/4 SAR architecture

Oil slick identification is accomplished using sin(2χ) and μw since they provide logical binary outputs without any external threshold:

π/4 μw, ALOS-2

(10000x2000 pixels)

WEAK-DAMPING LOOK-ALIKE π/4 μw,

ALOS-1 (18000x1200

pixels)

OIL SEEPS

Experimental Results Oil slick characterization is accomplished using Hw and Pw since they are continuous estimators of the spatial variability within the same slick:

HP Hw ALOS-2 (10000x2000pixels) HP Hw ALOS-1 (18000x1200pixels)

Experimental Results Oil slick characterization is accomplished using Hw and Pw since they are continuous estimators of the spatial variability within the same slick:

HP Pw ALOS-2 (10000x2000 pixels) HP Pw ALOS-1 (18000x1200 pixels)

Experimental Results The behaviour of Hw and Pw along with different surface slicks can be analyzed:

OIL SLICK

H > 0.8 P < 0.6

SLICK FREE SEA SURFACE

H < 0.4 P > 0.8

WEAK-DAMPING SURFACTANT

H < 0.6 P > 0.7

Conclusions

Compact-polarimetry SAR architectures can be exploited for sea oil slick observation obtaining performance that tends to the FP ones but overcoming FP drawbacks;

Both Hybrid-polarity and Slant-linear CP SAR architectures allow identifying oil slicks and characterizing them according to their damping properties;

CP SARs are able to perform sea oil slick detection without any external threshold;

Rough information on the spatial variability of different slicks can be inferred from the

polarization state of the partially polarized backscattered wave using continuous estimators.

THANK YOU ALL FOR THE ATTENTION