Science Service System

Summary of Proposal COA2891

TitleComparison of SAR systems capabilities for the analysis of Oil Spill characteristics.
Investigator laneve, Giovanni - University of Rome , DIAEE
Team Member
Dr Marzialetti, Pablo - University of Rome , DIAEE
Dr Fusilli, Lorenzo - University of Rome , DIAEE
SummaryThis proposal aims to the analysis of data acquired from TERRASAR-X satellite, in comparison with SAR images acquired in C (RADARSAT, ASAR, ERS) and L (PALSAR) band in order to assess the capabilities of these systems tomonitor and assess the main characteristics of oil spills. Oil Spill pollution is a problem that impacts on coastal economies, ecologyand the water life cycle. At the same time, the rise in energy demand puts intoconsideration the exploitation of remote areas, many of these under severeclimate conditions, which need to be monitored. In order to guaranteesustainable management procedures, remote sensing techniques are commonly used,and in particular microwave remote sensing data, thanks to its wide areacoverage and day-and-night weather capabilities. In recent years, the DIAEE of Sapienza - Universityof Rome, has been working on algorithmsfor oil spill detection andconducting studiesonspectral characterization by remote sensing instruments,of petroleumin Venezuela (PDVSAcompany operational areas) in the Lake MaracaiboandOrinoco Belt(the most important oil reservoirsin theworld) getting a significant know-how about oil extraction process from these reservoirs. For this proposal, we aim at analyzing spills located along outerand coastal waters, as well as illegal ships discharges or offshore platformsaccidental spills for future local applications. This spills' monitoringassociated with an early detection warning system, will be a significantmanagement resource for a region which contains the most important oil reservoirsin the world. The classification process based on Neural Networks techniques (like PatternRecognition and Backpropagation Neural Network) have proved to be appropriateto reach correct classifications with a percentage of success above 85%, andtaking advantage of polarimetric information,a new feature set will be developed and tested to improve previousachievements. It is important to notice that the quantity of information analyzedwill be essential in order to guarantee better subsequent classifications.Therefore in the future, in order to reach this objective, a continuous provisionof satellite imagery will increase the neural network capabilities. At the same time, the detectability of oil spills in SAR images depends onocean surface wind speeds. Thus, in order to solve the challenges related notonly with its discrimination from contextual information but also with spillsage, and future models to predict oil spills movements, a subsystem focused onwind fields extraction from SAR imagery will be developed.

Back to list of proposals

DLR 2004-2016