Automatic Recognition Of Coastal and Oceanic Environmental Events In Orbital Radars
Cristina Maria Bentz(1), Nelson F. F. Ebecken(2) and Alexandre Tadeu Politano(1)
(1) PETROBRAS Research Center, Av. Jequitibá, 950 - Ilha do Fundão, 21949-900 - Rio de Janeiro - RJ, Brazil
(2) Rio de Janeiro Federal University, Ilha do Fundão, 21949-900 – Rio de Janeiro - RJ, Brazil
The increase availability of spaceborne Synthetic Aperture Radar (SAR) is
providing opportunities for large scale oceanic monitoring and oil spill detection, compared
to scattered ship observations or aircraft surveillance in limited areas. Of primary use to
spill responders is the SAR sensors that can provide high spatial resolution images of the
sea surface delivered in near real time. The physical mechanism that allows detection of
oil and different oceanic surface phenomena is the differential modulation of wind induced
capillary waves. As a result, atmospheric processes that affect surface wind conditions or
oceanic events that directly modulate the capillary waves produce signatures readily
detected by SAR. The presence of oil dampens the capillary waves generating low
backscatter region, dark in contrast with the background radar signal. However, the
interpretation of oceanic SAR signatures is not trivial since more than one process can
operate concurrently and different phenomena produce similar backscattering signal.
This paper presents the development of an automatic classification procedure able
to identify different oceanic events, detectable in orbital radar images. The procedure was
customized to be used in the southeastern Brazilian coast, since the classification training
and test used examples extracted from 402 RADARSAT-1 images acquired in this region.
This area is presently responsible for about 89% of all the Brazilian oil and gas production.
The main oceanographic feature affecting the surface circulation is the Brazil Current (BC)
that flows predominantly from NE to SW all year round. The presence of meanders and
mesoscale vortices can, however, induce large perturbations in the prevailing flow.
Frequent and intense cold upwelling plumes are also observed.
Different sets of spectral, geometric and contextual (meteoceanographic and
location) features of selected low backscatter areas were evaluated. The examples used
to train and test the classifiers were selected using in situ information and meteooceanographic
ancillary data derived from weather satellites, scatterometers or medium
resolution sensors is being used to achieved better interpretations of the SAR images.
Satellite sensors operating in the visible part of the spectrum can be used to monitor
ocean color variations providing information on suspended sediments and biogenic
phenomena related to chlorophyll production and algae blooms. Thermal infrared
radiometers can point to location of oceanic fronts, meanders and upwelling plumes. Wind
scatterometers permit the assessment of ocean surface wind speed and direction
essential to assess the usefulness and support the SAR interpretations.
Machine learning procedures (neural networks, decision trees and support vector
machines) were used to induce classifiers to differentiate between seven classes,
belonging to two categories. The classification procedure involves two steps: first the
features area classified in one of two categories - oil pollution or meteoceanographic
event. In the second step, the identification of tree classes of oil pollution and four classes
of meteoceanographic events is done. The oil spill related classes are associated to oil
exploration and production, ship releases and others. The meteoceanographic phenomena
include biogenic slicks and /or upwellings, algae blooms, low wind areas and rain cells.
The models induced by support vector machines and neural networks achieved good
results, allowing the operational implementation of the proposed procedures.
Keywords: ESA European
Space Agency - Agence spatiale europeenne,
observation de la terre, earth observation,
satellite remote sensing,
teledetection, geophysique, altimetrie, radar,
chimique atmospherique, geophysics, altimetry, radar,