# Comparison of Oil Spill Classifications Using Fully and Compact Polarimetric SAR Images

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## 1. Introduction

^{3}of oil, methane or other fluids were released into the Gulf of Mexico. In 2011, approximately 700 barrels of crude oil were leaked into the Bohai Sea, and about 2500 barrels of mineral oil-based mud became deposited on the seabed. In December 2013, during an accident caused by a broken oil pipe, crude oil leaked into the coastal area of Qingdao, Shandong province, and covered approximately 1000 m

^{2}of the sea surface. In addition, a large proportion of oil spills are caused every year by deliberate discharges from tankers or cargos, for the reason that there are still vessels that secretly clean their tanks or engine before entering the harbor. These accidents and illegal acts cause damage to the coastal ecosystem, emphasizing the importance of detecting oil spills in their early stages.

## 2. Methods

#### 2.1. Quad-Polarimetric SAR Mode

_{HV}= S

_{VH}.

#### 2.2. Feature Extraction from Quad-Polarimetric SAR Data

#### 2.2.1. Single Polarimetric Intensity

#### 2.2.2. H/α Decomposition Parameters

**U**can be parameterized by Equation (4):

_{3}**T**are real numbers, arranged as ${\lambda}_{1}>{\lambda}_{2}>{\lambda}_{3}$,

**U**is the unitary matrix, whose column vectors ${\overrightarrow{u}}_{1}$, ${\overrightarrow{u}}_{2}$ and ${\overrightarrow{u}}_{3}$ are the eigenvectors of

_{3}**T**:

#### 2.2.3. Degree of Polarization

_{i}is Stokes vectors that can be used to describe both complete and partially polarized wave, and i stands for different polarization of transmission.

_{v}and E

_{h}is vertically and horizontally received backscatter, respectively, and < > also stands for multilook by using an averaging window.

#### 2.2.4. Ellipticity χ

#### 2.2.5. Pedestal Height

^{0}measures how detectable an object is per unit area on the ground. In the co-polarized signature of the scene, the σ

^{0}is a function of both the tilting angle Φ and the ellipticity χ of the polarization ellipse. The pedestal height (PH) is defined as the lowest value of all the σ

^{0}, plotted in the co-polarized signature. The PH describes the unpolarized energy of the total scattering power and behaves as a pedestal on which the co-polarized signature is set [14,33]. The normalized pedestal height (NPH) can be approximately calculated as the minimum eigenvalue divided by the maximum one:

#### 2.2.6. Co-Polarized Phase Difference

#### 2.2.7. Conformity Coefficient

#### 2.2.8. Correlation and Coherence Coefficients

**T**are elements of the coherence matrix

_{ij}**T**.

#### 2.3. Dual- and Compact Polarimetric SAR Modes

#### 2.4. Universal Feature Extraction from Dual- and Compact Polarimetric SAR Data

#### 2.4.1. Elements in Measurement Vector $\overrightarrow{K}$

#### 2.4.2. H/α Decomposition Parameters

**C**

_{CP}:

**C**

_{CP}. Entropy that is derived directly from CP SAR data has similar performance as that derived from quad-pol SAR data, in describing the complexity of the physical scattering mechanisms of targets.

_{i}can be derived from the eigenvector of the covariance of CP SAR data, similarly as in Section 2.2.

#### 2.4.3. Degree of Polarization and Ellipticity

#### 2.4.4. Pedestal Height

#### 2.4.5. Co-Polarized Phase Difference

#### 2.4.6. Conformity Coefficient

#### 2.4.7. Correlation Coefficient and Coherence Coefficient

#### 2.5. Supervised Classifications

#### 2.5.1. Support Vector Machine (SVM)

**x**

_{i}into a higher dimensional space by using kernel function Φ and, hence, finds a linear separating hyperplane with the maximal margin in this higher dimensional space. In this paper, the radial basis function is adopted as the kernel function.

#### 2.5.2. Artificial Neural Network (ANN)

#### 2.5.3. Maximum Likelihood Classification (ML)

#### 2.6. Features Selection Scheme

#### 2.7. Classification Accuracy Evaluation

#### 2.8. Dimension Reduction

## 3. Results

#### 3.1. Oil Spill Classification Based on Fully Polarimetric SAR Features

_{HHVV}, correlation coefficient, coherency coefficient, standard deviation of CPD and alpha angle. The nine features used for ANN are all of the features except ellipticity. As introduced in the previous session, all of these features have strong physical meaning, which enables them to largely contribute to the classification between mineral oil and clean sea surface/biogenic film. They are also not likely affected by the noise floor.

#### 3.2. Oil Spill Classification Based on Different Polarimetric SAR Modes

#### 3.3. Oil Spill Classification Based on Dimension Reduction of Features

## 4. Discussion

## 5. Conclusions

- Polarimetric SAR features can be input into supervised algorithms to achieve reliable oil spill classification. For this dataset, a feature set with four features is sufficient for most polarimetric features based oil spill classifications. They are: pedestal height, correlation coefficient, standard deviation of CPD and alpha angle.
- Among all of the compact polarimetric SAR modes, π/2 mode has the best performance among all of the dual- and compact polarimetric SAR modes, for its sensitivity to different scattering mechanisms caused by mineral oil and biogenic look-alikes.
- Among all of the supervised classifiers, SVM outperforms other classifiers when sufficient polarimetric information can be obtained, such as quad-pol mode. ML performs better than other supervised classifiers when only incomplete polarimetric information is available, such as traditional dual-pol and π/4 mode.

## Acknowledgments

## Author Contributions

## Conflicts of Interest

## References

- Alpers, W.; Espedal, H. Oils and surfactants. In Synthetic Aperture Radar Marine User’s Manual; US Department of Commerce: Washington, DC, USA, 2004; pp. 263–275. [Google Scholar]
- Migliaccio, M.; Gambardella, A.; Tranfaglia, M. SAR Polarimetry to Observe Oil Spills. IEEE Trans. Geosci. Remote Sens.
**2007**, 45, 506–511. [Google Scholar] [CrossRef] - Migliaccio, M.; Nunziata, F.; Gambardella, A. On the co-polarized phase difference for oil spill observation. Int. J. Remote Sens.
**2009**, 30, 1587–1602. [Google Scholar] [CrossRef] - Topouzelis, K.; Stathakis, D.; Karathanassi, V. Investigation of Genetic Algorithms Contribution to Feature Selection for Oil Spill Detection. Int. J. Remote Sens.
**2008**, 30, 611–625. [Google Scholar] [CrossRef] - Marghany, M.; Hashim, M. Discrimination between oil spill and look-alike using fractal dimension algorithm from RADARSAT-1 SAR and AIRSAR/POLSAR data. Int. J. Phys. Sci.
**2011**, 6, 1711–1719. [Google Scholar] - Zhang, B.; Perrie, W.; Li, X.; Pichel, W. Mapping sea surface oil slicks using RADARSAT-2 quad-polarization SAR image. Geophys. Res. Lett.
**2011**, 38, 415–421. [Google Scholar] [CrossRef] - Zhang, Y.; Lin, H.; Liu, Q.; Hu, J.; Li, X.; Yeung, K. Oil-spill monitoring in the coastal waters of Hong Kong and vicinity. Mar. Geod.
**2012**, 35, 93–106. [Google Scholar] [CrossRef] - Guo, J.; He, Y.; Long, X.; Hou, C.; Liu, X.; Meng, J. Repair wind field in oil contaminated areas with SAR images. Chin. J. Oceanol. Limnol.
**2015**, 33, 525–533. [Google Scholar] [CrossRef] - Gade, M.; Alpers, W. Using ERS-2 SAR for routine observation of marine pollution in European coastal waters. Sci. Total Environ.
**1999**, 237, 38441–38448. [Google Scholar] [CrossRef] - Suri, S.; Schwind, P.; Uhl, J.; Reinartz, P. Modification in the SIFT operator for effective SAR image matching. Int. J. Image Data Fusion
**2010**, 1, 243–256. [Google Scholar] [CrossRef][Green Version] - Minchew, B.; Jones, C.E.; Holt, B. Polarimetric Analysis of Backscatter from the Deepwater Horizon Oil Spill Using L-Band Synthetic Aperture Radar. IEEE Trans. Geosci. Remote Sens.
**2012**, 50, 3812–3830. [Google Scholar] [CrossRef] - Migliaccio, M.; Nunziata, F.; Buono, A. SAR polarimetry for sea oil slick observation. Int. J. Remote Sens.
**2015**, 36, 3243–4273. [Google Scholar] [CrossRef] - Li, H.; Perrie, W.; He, Y.; Wu, J.; Luo, X. Analysis of the Polarimetric SAR Scattering Properties of Oil-Covered Waters. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens.
**2015**, 8, 3751–3759. [Google Scholar] [CrossRef] - Nunziata, F.; Migliaccio, M. Gambardella, A. Pedestal height for sea oil slick observation. IET Radar Sonar Navig.
**2010**, 5, 103–110. [Google Scholar] [CrossRef] - Alpers, W. Remote sensing of oil spills. In Proceedings of the Symposium Maritime Disaster Management, King Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia, 19–23 January 2002; pp. 19–23.
- Souyris, J.C.; Imbo, P.; Fjortoft, R.; Mingot, S.; Lee, J.-S. Compact polarimetry based on symmetry properties of geophysical media: The π/4 mode. IEEE Trans. Geosci. Remote Sens.
**2005**, 43, 634–646. [Google Scholar] [CrossRef] - Sabry, R.; Vachon, P.W. A Unified Framework for General Compact and Quad Polarimetric SAR Data and Imagery Analysis. IEEE Trans. Geosci. Remote Sens.
**2014**, 52, 582–602. [Google Scholar] [CrossRef] - Chen, J.; Quegan, S. Calibration of Spaceborne CTLR Compact Polarimetric Low-Frequency SAR Using Mixed Radar Calibrators. IEEE Trans. Geosci. Remote Sens.
**2011**, 49, 2712–2723. [Google Scholar] [CrossRef] - Nord, M.E.; Ainsworth, T.L.; Lee, J.-S.; Stacy, N.J.S. Comparison of Compact Polarimetric Synthetic Aperture Radar Modes. IEEE Trans. Geosci. Remote Sens.
**2009**, 47, 174–188. [Google Scholar] [CrossRef] - Dubois-Fernandez, P.C.; Souyris, J.-C.; Angelliaume, S.; Garestier, F. The Compact Polarimetry Alternative for Spaceborne SAR at Low Frequency. IEEE Trans. Geosci. Remote Sens.
**2008**, 46, 3208–3222. [Google Scholar] [CrossRef] - Collins, M.J.; Denbina, M.; Atteia, G. On the Reconstruction of Quad-Pol SAR Data from Compact Polarimetry Data for Ocean Target Detection. IEEE Trans. Geosci. Remote Sens.
**2013**, 51, 591–600. [Google Scholar] [CrossRef] - Li, H.; Perrie, W.; He, Y.; Lehner, S.; Brusch, S. Target Detection on the Ocean with the Relative Phase of Compact Polarimetry SAR. IEEE Trans. Geosci. Remote Sens.
**2013**, 5, 3299–3305. [Google Scholar] [CrossRef] - Yin, J.; Yang, J.; Zhang, X. On the ship detection performance with compact polarimetry. In Proceedings of the 2011 IEEE Radar Conference (RADAR), Kansas City, MO, USA, 23–27 May 2011; pp. 675–680.
- Raney, R.K. Hybrid-Polarity SAR Architecture. IEEE Trans. Geosci. Remote Sens.
**2007**, 45, 3397–3404. [Google Scholar] [CrossRef] - Shirvany, R.; Chabert, M.; Tourneret, J.-Y. Ship and Oil-Spill Detection Using the Degree of Polarization in Linear and Hybrid/Compact Dual-Pol SAR. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens.
**2012**, 5, 885–892. [Google Scholar] [CrossRef][Green Version] - Cloude, S.R.; Goodenough, D.G.; Chen, H. Compact Decomposition Theory. IEEE Geosci. Remote Sens. Lett.
**2012**, 9, 28–32. [Google Scholar] [CrossRef] - Salberg, A.-B.; Rudjord, O.; Solberg, A.H.S. Oil Spill Detection in Hybrid-Polarimetric SAR Images. IEEE Trans. Geosci. Remote Sens.
**2014**, 52, 6521–6533. [Google Scholar] [CrossRef] - Nunziata, F.; Migliaccio, M.; Li, X. Sea Oil Slick Observation Using Hybrid-Polarity SAR Architecture. IEEE J. Ocean. Eng.
**2014**, 1, 426–440. [Google Scholar] [CrossRef] - Yin, J.; Yang, J.; Zhou, Z.; Song, J. The Extended Bragg Scattering Model-Based Method for Ship and Oil-Spill Observation Using Compact Polarimetric SAR. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens.
**2014**, 99, 1–13. [Google Scholar] [CrossRef] - Ulaby, F.T.; Moore, R.K.; Fung, A.K. Microwave Remote Sensing. Active and Passive; Artech House Inc.: Norwood, MA, USA, 1986; Volume 3. [Google Scholar]
- Cloude, S.R.; Pottier, E. An entropy based classification scheme for land applications of polarimetric SAR. IEEE Trans. Geosci. Remote Sens.
**1997**, 35, 68–78. [Google Scholar] [CrossRef] - Shirvany, R. Estimation of the Degree of Polarization in Polarimetric SAR Imagery: Principles & Applicaions. Ph.D. Thesis, Institut National Polytechnique de Toulouse, Toulouse, France, October 2012. [Google Scholar]
- Migliaccio, M.; Nunziata, F. On the exploitation of polarimetric SAR data to map damping properties of the Deepwater Horizon oil spill. Int. J. Remote Sens.
**2014**, 35, 3499–3519. [Google Scholar] [CrossRef] - Skrunes, S.; Brekke, C.; Eltoft, T. Characterization of Marine Surface Slicks by Radarsat-2 Multipolarization Features. IEEE Trans. Geosci. Remote Sens.
**2014**, 52, 5302–5319. [Google Scholar] [CrossRef][Green Version] - Vapnik, V. Statistical Learning Theory; Wiley: New York, NY, USA, 1998. [Google Scholar]
- Kavzoglu, T.; Mather, P.M. The use of backpropagating artificial neural networks in land cover classification. Int. J. Remote Sens.
**2003**, 24, 4907–4938. [Google Scholar] [CrossRef] - Ahmad, A.; Quegan, S. Analysis of Maximum Likelihood Classification on Multispectral Data. Appl. Math. Sci.
**2012**, 6, 6425–6436. [Google Scholar] - Roweis, S.T.; Saul, L.K. Nonlinear dimensionality reduction by locally linear embedding. Science
**2000**, 290, 2323–2326. [Google Scholar] [CrossRef] [PubMed] - Tenenbaum, J.B.; de Silva, V.; Langford, J.C. A global geometric framework for nonlinear dimensionality reduction. Science
**2000**, 290, 2319–2322. [Google Scholar] [CrossRef] [PubMed] - Alves, T.M.; Kokinou, E.; Zodiatis, G.; Radhakrishnan, H.; Panagiotakis, C.; Lardner, R. Multidisciplinary oil spill modelling to protect coastal communities and the environment of the Eastern Mediterranean Sea. Sci. Rep.
**2016**, 6, 36882. [Google Scholar] [CrossRef] [PubMed] - Alves, T.M.; Kokinou, E.; Zodiatis, G. A three-step model to assess shoreline and offshore susceptibility to oil spills: The South Aegean (Crete) as an analogue for confined marine basins. Mar. Pollut. Bull.
**2014**, 86, 443–457. [Google Scholar] [CrossRef] [PubMed] - Alves, T.M.; Kokinou, E.; Zodiatis, G.; Lardner, R. Hindcast, GIS and susceptibility modelling to assist oil spill clean-up and mitigation on the southern coast of Cyprus (Eastern Mediterranean). Deep Sea Res. II Top. Stud. Oceanogr.
**2015**, 133, 159–175. [Google Scholar] [CrossRef]

**Figure 4.**Classification accuracy achieved by three classifiers with the number of features changing from 2–10.

**Figure 5.**Classification results based on quad-pol SAR features using different classifiers. (

**a**) SVM; (

**b**) ML; (

**c**) ANN.

**Figure 7.**Classification result using SVM based on the features of: (

**a**) DP mode; (

**b**) π/4 mode; (

**c**) π/2 mode; (

**d**) ${S}_{VV}^{2}$.

**Table 1.**Behaviors of main polarimetric SAR features on different types of surfaces. DoP, degree of polarization; CPD, co-polarized phase difference.

Pol-SAR Features | Clean Sea Surface | Mineral Oil (Strong Damping) | Biogenic Slicks (Weak Damping) |
---|---|---|---|

Entropy (H) | Lower | High | Low |

Alpha (α) | Lower | High | Low |

DoP | High | Low | High |

Ellipticity | Negative | Positive | Negative |

Pedestal Height (PH) | Lower | High | Low |

Std. CPD | Lower | High | Low |

Conformity Coefficient | Positive | Negative | Positive |

Correlation Coefficient | Higher | Low | High |

Coherence Coefficient | Higher | Low | High |

${S}_{VV}^{2}$ | High | Low | Low |

Receive | H | V | H and V (Incoherently) | H and V (Coherently) | R and L (Coherently) | |
---|---|---|---|---|---|---|

Transmit | ||||||

H | Single | Single | Alternating | Dual Pol | — | |

V | Single | Single | Alternating | Dual Pol | — | |

H and V | — | — | Alternating | — | — | |

45° | — | — | — | π/4 Compact | ||

R/L | — | — | — | π/2 Compact (Hybrid) | Dual-circular (DCP) |

Number | Feature |
---|---|

1 | ${S}_{VV}^{2}$ |

2 | Pedestal Height |

3 | Entropy |

4 | DoP_{HHVV} |

5 | Correlation Coefficient |

6 | Conformity Coefficient |

7 | Coherency Coefficient |

8 | Ellipticity χ |

9 | CPD Standard Deviation |

10 | Alpha Angle |

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 5429 | 178 | 5607 |

Sea | 121 | 5357 | 5478 |

Total | 5550 | 5535 | 11,085 |

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 5411 | 256 | 5667 |

Sea | 139 | 5279 | 5418 |

Total | 5550 | 5535 | 11,085 |

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 5427 | 232 | 5659 |

Sea | 123 | 5303 | 5426 |

Total | 5550 | 5535 | 11,085 |

Number | Feature * |
---|---|

1 | ${E}_{V}^{2}$ |

2 | Pedestal Height (CP) |

3 | Entropy (CP) |

4 | DoP (CP) |

5 | Correlation Coefficient (CP) |

6 | Alpha Angle (CP) |

7 | Coherency Coefficient (CP) |

8 | Ellipticity χ (CP) |

9 | CPD Standard Deviation (CP) |

10 | Conformity Coefficient (π/2) |

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 5357 | 445 | 5802 |

Sea | 193 | 5090 | 5283 |

Total | 5550 | 5535 | 11,085 |

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 5378 | 363 | 5741 |

Sea | 172 | 5172 | 5344 |

Total | 5550 | 5535 | 11,085 |

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 5316 | 595 | 5911 |

Sea | 234 | 4940 | 5174 |

Total | 5550 | 5535 | 11,085 |

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 5438 | 4125 | 9563 |

Sea | 112 | 1410 | 1522 |

Total | 5550 | 5535 | 11,085 |

**Table 12.**Confusion matrix achieved by SVM based on four features derived from PCA on quad-pol SAR features.

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 4649 | 82 | 4731 |

Sea | 901 | 5453 | 6354 |

Total | 5550 | 5535 | 11,085 |

**Table 13.**Confusion matrix achieved by SVM based on four features derived from local linear embedding (LLE) on quad-pol SAR features.

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 4879 | 197 | 5076 |

Sea | 671 | 5338 | 6009 |

Total | 5550 | 5535 | 11,085 |

**Table 14.**Confusion matrix achieved by SVM based on four features derived from ISOMAP on quad-pol SAR features.

Class | Ground Truth (Pixels) | ||
---|---|---|---|

Oil | Sea | Total | |

Oil | 4809 | 271 | 5080 |

Sea | 741 | 5264 | 6005 |

Total | 5550 | 5535 | 11,085 |

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ( http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Zhang, Y.; Li, Y.; Liang, X.S.; Tsou, J. Comparison of Oil Spill Classifications Using Fully and Compact Polarimetric SAR Images. *Appl. Sci.* **2017**, *7*, 193.
https://doi.org/10.3390/app7020193

**AMA Style**

Zhang Y, Li Y, Liang XS, Tsou J. Comparison of Oil Spill Classifications Using Fully and Compact Polarimetric SAR Images. *Applied Sciences*. 2017; 7(2):193.
https://doi.org/10.3390/app7020193

**Chicago/Turabian Style**

Zhang, Yuanzhi, Yu Li, X. San Liang, and Jinyeu Tsou. 2017. "Comparison of Oil Spill Classifications Using Fully and Compact Polarimetric SAR Images" *Applied Sciences* 7, no. 2: 193.
https://doi.org/10.3390/app7020193