There are multiple methods reported in the literature to distinguish oil slicks from look-alikes. It is hard to tell which is the best method. It depends on factors including the extended operating conditions (EOCs). So, I don't think you going to find a superior method; it is all context-dependent. This is known in the machine learning/pattern classification literature as the No-Free-Lunch theorem. What works best for your problem/context may not work best for others, and vice versa.

To explicitly address your question, one method for oil slick detection is reported in Ref [1]. For further discussion on the subject, please see Page 91 (Chapter 5) in Ref [2]. DOIs are provided below for convenience.

Hope you find this helpful.

REFERENCES:

[1] Solberg, A. H S; Storvik, G.; Solberg, R.; Volden, E., "Automatic detection of oil spills in ERS SAR images," Geoscience and Remote Sensing, IEEE Transactions on , vol.37, no.4, pp.1916,1924, Jul 1999 doi: http://dx.doi.org/10.1109/36.774704

[2] Cortés, Ulises, and Manel Poch. Advanced agent-based environmental management systems. Birkhauser Basel, 2009. doi: http://dx.doi.org/10.1007/978-3-7643-8900-0_5

Hi,

Some recent studies have shown that using polarimetric SAR data is possible to improve the detection of oil slicks and to distinguish them from biogenic slicks.

You can find more details in the reference below.

References:

[1] Migliaccio, M., F. Nunziata, and A. Gambardella (2009), "On the co-polarized phase difference for oil spill observation", Int. J. Remote Sens., 30(6), 1587–1602.

[2] Maurizio Migliaccio1, Ferdinando Nunziata, Carl E. Brown, Benjamin Holt, Xiaofeng Li, William Pichel, Masanobu Shimada (2012),"Polarimetric synthetic aperture radar utilized to track oil spills",Eos Trans. AGU, 93(16), 161.

First answer: Polarimetric SAR measurements with appropriate estimator. You can read several papers on this matter that we published. As first start go to M.Migliaccio, F.Nunziata, A.Gambardella, “On the Co-Polarized Phase Difference for Oil Spill Observation”, International Journal of Remote Sensing, vol.30, no.6, pp.1587-1602, 2009.

An interesting paper about the limits of single-pol procedures is:

A.Gambardella, G.Giacinto, M.Migliaccio, A.Montali, “One-Class Classification for Oil Spill Detection”, Pattern Analysis & Applications Journal, DOI 10.1007/s10044-009-0164-z.

Victor,

I am relatively new to Research Gate and just now 'stumbled' onto this question that is a couple of years old.  Not sure if useful any longer but I'll add a comment or two:

* The airborne SLAR/SAR work done in the late 70s to mid 80s looked at several aspects of the use of radar imaging including the detection of oil slicks.  Not sure about algae related slicks (blooms?).   As you know one of the main surface characteristics that radar imaging (SAR) responses to is the surface texture / roughness and density / hardness of the feature / target.  On land it can have to do with the smoothness / roughness of the surface (soils and/or vegetation) and the same for ocean waters.  Since oil slicks tend to smooth out the ocean waters (waves etc) more of the radar enegry bounces away from the imaging sensor than it does over non-slicked areas, therefroe, they show up as 'dark radar reflectance' features.  I am not sure if the oil versus algae type of slicks would smooth the waters a different amount so that this difference could be detetected.  Of course, depending on the radar wave lenght / frequency what looks smooth vs rough will be different; it would be interesting to look at the different types of slicks with multi-frequency (C, X, and L bands) radar images.

* A 2nd possiblity would be to get radar images collected as close as possible to a Landsat TM type image.  The slicks will also look dark on the Landsat TM image, but I wonder if perhaps you would be able to detect the differences using the combination of the radar and Landsat TM spectral bands that respond to algae different than radar.  Of course the problem with Landsat TM is that it does not see thru clouds and can not image at night time.

Sorry for such a late response; fyi I was involved with airborne SLAR and the early shuttle SIR-A and images, as well as with acoustic imaging of the ocean floor (basically a brute force SLAR system).

Pat

This might be too late a reply, but I recently created an automatic rule-based classifier to detect natural slicks in SAR images. It uses both direct (geometry and texture of the segmented object) and contextual features (prevailing wind information) to distinguish between the segmented natural oil slicks and other look-alikes like natural biological blooms, oil spills etc. Have a look at my publications for more details.

Gopika