RESEARCH

Oil spill detection from SAR images

Intro

In the last 15 years the detection of oil spills by satellite means has been moved from experimental to operational. Actually what is really changed is the satellite image availability. From the late 1990’s, in the age of “no data” we have moved forward 15 years to the age of “Sentinels” with abundance of data. However, all these years the scientific methodology on the detection remains relatively constant. From manual analysis to fully automatic detection methodologies, no significant contribution has been published in the last years and certainly none has dramatically changed the rules of the detection. On the contrary, although the overall accuracy of experimental methodology is questioned, the four main classification steps (dark area detection, features extraction, statistic database creation and classification) are continuously improving. The present work tries to point out the drawbacks of the oil spill detection in the last years and focus on the bottlenecks of the oil spill detection methodologies.

Towards improvement on spaceborne oil spill detection

An accurate validation database is more than essential. Despite the new advanced detection methodologies applied by the scientific community validation datasets are most of the times based on experts decision or on fragmented case studies. Therefore, the different published methods cannot really be compared, and no actual improvement can be recorded. There is a need of access to a validated database created by an international organization like EMSA.

The discrimination between oils spills and lookalikes is not a correct problem description in terms of classification classes. The right procedure should be a database with the known phenomena and their feature values in different scales e.g. a local oceanic front or a large current. Further work is needed, especially on lookalike phenomena (sea truth data) verified by buoys and ground radars. These real observations will include more phenomena in the classification chain and further analysis can be achieved on the occurrence reason of each event.

Need of operational monitoring – sever based algorithm. Instead of traditional processing algorithms web based protocols should be developed. The traditional scheme should be replaced from a web processing semantic algorithms where only part of the images with sea and possible oil spills should be processed.

Recommendations

  1. Need of valid database for algorithms improvement

  Selected cases with airplane observations less than two hours of the satellite pass, both for oil spills and lookalikes.

  1. New hyper band radar sensors 

Need of multi or hyper radars, existed examples with C, X and L band.

  1. New types of algorithms with ontology

  Large data need server based algorithms based on simple ontology and further research only to interesting images, or part of them.

  1. In situ measurements

Tiny low cost buoys (biodegradable), small AUVs monitoring/sample selection, optical imagery

 

Publlications

Topouzelis K., Singha S., 2016, Oil spill detection: past and future trends, ESA Living Planet Symposium 2016, Prague, Czech Republic 9-13 May 2016

K. Topouzelis, D. Tarchi, M. Vespe, M. Posada, O. Muellenhoff and G. Ferraro, 2015, Chapter: 13, Detection, Tracking, and Remote Sensing: Satellites and Image Processing (Spaceborne Oil Spill Detection), Handbook of Oil Spill Science and Technology, Publisher: John Wiley & Sons, Editors: Merv Fingas

Topouzelis, K. and Kitsiou, 2014, D. Sea State Primitive Object Creation from SAR Data, International Journal of Geosciences (open access), 5, 1561-1570. http://dx.doi.org/10.4236/ijg.2014.513127

Topouzelis, K., Psyllos,A. 2012, Oil spill feature selection and classification using decision tree forest on SAR image data. ISPRS, J. Photogram. Remote Sensing, 68, pp. 135–143, doi:10.1016/j.isprsjprs.2012.01.005

Topouzelis, D. Stathakis, V. Karathanassi, 2009, Investigation of genetic algorithms contribution to feature selection for oil spill detection, International Journal of Remote Sensing, vol.30, no.3, pp.611-625.

G. Ferraro, B. Bulgarelli, S. Meyer-Roux, O. Muellenhoff, D. Tarchi, K. Topouzelis, 2009, Long term monitoring of oil spills in the European Seas, International Journal of Remote Sensing, vol.30, no.3, pp.627-645

 Κ. Topouzelis and V. Karathanassi, 2009, Potentiality of feed-forward neural networks for classifying dark formations to oil spills and look-alikes, Geocarto International (indexed in Thomson’s ISI Web of Knowledge, IF: 1.370), Vol. 24, No. 3, 2009, 179–191.

K. Topouzelis, 2008, “Oil spill detection by SAR images: Approaches and Algorithms”, Invited paper for the special issue: Synthetic Aperture Radar (SAR), SENSORS Journal, no. 8, pp. 6642-6659.

Ferraro, B.Bulgarelli, S. Meyer-Roux, O. Muellenhoff, D. Tarchi, K.Topouzelis, 2008, The Use of Satellite Imagery from Archives to Monitor Oil Spills in the Mediterranean Sea, Chapter on "Remote Sensing of the European Seas", Vittorio Barale & Martin Gade, editors, Springer.

K. Topouzelis, V. Karathanassi, P. Pavlakis, D. Rokos, 2008, Dark formation detection using neural networks, International Journal of Remote Sensing, vol. 29, no. 16, pp. 4705–4720.

O. Müllenhoff, B. Bulgarelli, G. Ferraro and K. Topouzelis, 2008, “The use of ancillary metocean data for the oil spill probability assessment in SAR images”, Fresenius Environmental Bulletin, Vol.17, No.9b.

K. Topouzelis, V. Karathanassi, P. Pavlakis, D. Rokos, 2007, Detection and discrimination between oil spills and look-alike phenomena through neural networks, ISPRS Journal of Photogrammetry and Remote Sensing, (62), pp. 264–270.

Ferraro, A. Bernardini, M. David, S. Meyer-Roux, O. Muellenhoff, M. Perkovic, D. Tarchi, K. Topouzelis, 2007, Towards an operational use of space imagery for oil pollution monitoring in the Mediterranean basin: A demonstration in the Adriatic Sea, Marine Pollution Bulletin, vol. 54, issue 4, pp. 403-422.

V. Karathanassi, K. Topouzelis, P. Pavlakis, D. Rokos, 2006, An object-oriented methodology to detect oil spills, International Journal of Remote Sensing, vol. 27, no. 23, pp.5235-5251.

K. Topouzelis, A. Bernardini, G. Ferraro, S. Meyer-Roux, D. Tarchi, 2006, Satellite mapping of oil spills in the Mediterranean Sea, Fresenius Environmental Bulletin, vol. 15, no. 9a, pp. 1009-1014.