Oil Spill Detection and Prediction in the Northwest Mediterranean Sea
Jose Redondo^{(1)} and Alexei Platonov^{(1)}
^{(1)}
Universitat Politecnica de Catalunya,
Calle Jordi Girona 13,
E08034 Barcelona,
Spain
Abstract
In recent years the marine pollution has been highlighted thanks to the
advances in detection techniques. There is also more public awareness to
both the large nautical catastrophes (e.g. oil tankers Amoco Cadiz, Exxon
Valdez and recently Erika and Prestige) and the habitual smaller oil
spills from the ships. The range of marine pollution events should even
consider, due to their overall importance,
the very much smaller oil spills of a few square meters caused by small boats.
The middle size oil spills often originate due to coastal sources and from
small accidents or habitual cleaning of ballast water in ships. The larger
oil spills are caused by crude/oil tankers catastrophic accidents of
varied consequences. From the analysis of SAR observations and new
satellite based sensors new methods of oil spill detection in the Ocean,
coupled with selfsimilar statistical techniques allows to determine with
precision the size statisticas of the pollution events and their
topological structure.
The oil pollution of Gulf of Lion in the NW
Mediterranean has been studied further than from the twoyear period from
December 1996 until November 1998 within the framework of the ECfunded
project "Clean Seas". New images during the period 1999 2005 have been
analysed, together with a more comprehensive identification of other
patterns and eddies detected in the ocean surface. More
than 1000 synthetic aperture radar (SAR), some ASAR and many other types
of images have been compared over the test sites used by the Clean Seas
proyect. We have analyzed these SAR images with respect to radar
signatures of (natural and manmade) oil pollution and other surface
features. Other phenomena causing similar signatures are considered and
evaluated, such as wind, river plumes, eddies and convergence areas.
The results of our statistical analysis are presented. The additional
SAR images reveal that the NW Mediterranean is most polluted along the
main ship traffic routes, but comparatively less that near other routes in
the Indic and the Pacific. The oils spil index, defined as the number of
detected spills per 10^4 Km^2 is higher than one. The sizes of the
detected oil spills vary over a large range, and if the statistics of the
largests accidents are also considered on a longer timescale, we show that
the Zipf's Law, relating the frequency and the size of the spill in a
hyperbolic fashion is applicable. Moreover, the higher amount of oil
spills
on SAR images acquired during summer (April  September) than on SAR images
acquired during winter (October  March). When evaluated statistically
together with the local wind velocity data show a chi distribution for the
probability of detecting oil spils in high winds.
Advanced image analysis techniques, such as the calculation of the
multifractal dimensions of the observed SAR signatures, have been applied
to distinguish between natural slicks and antropogenic spills. Fractal
dimensions can also be used to predict the time of release of the spill
with an appropiate non dimensionalization of the time based on the
turbulent dissipation. The multiscale appearence, the topological
structure and the fractality of the slicks and spills may also be used as
a measure of the diffusiveness of the observed signatures, they yield
additional information on the signatures' origin, on eddy and jet type of
structures which in turn may improve automated detection algorithms and be
used in numerical models.
Fractal analysis was used to identify different dynamic processes that
influence the radar backscattering from the ocean surface. We used a
boxcounting algorithm that is able to detect the selfsimilar
characteristics for different SARimage intensity levels.
It is very interesting to relate D to the frequency spectrum or to the
spatial spectra obtained from the Fourier transform of the time or spatial
correlation functions, usual in studies of turbulence.
The reason is that from such frequency spectrum the corresponding fractal
dimension may be derived, if the tracer scalar is passively advected by a
turbulent flow. Then the fractal dimension is related to the energy of the
turbulence with a certain spatial or temporal dependence,
then the frequency spectrum exponent, provided an inertial subrange
exists, is a function of the boxcounting fractal dimension as
demonstrated by Derbyshire and Redondo (1990) and Redondo (1990).
The interactions between the selfsimilar ocean turbulent flow, where the
Rossby deformation Radius plays an important role and the oil spills is
used to model numerically the dispersion. (Gade and Redondo 1999 and
Redondo and Platonov 2001)
Traditionally in environmental studies of diffusion, oil patches have
been numerically predicted and computed with homogeneous environmental
forcing and random free
paths, which gives Brownian behavior. These stochastic methods have the
objection that they do not take into account the topology of the flow.
On the other hand, there are many ways to simulate a fluid flow, but when
this is turbulent, these simulations become complicated, expensive and
inaccurate. Together with examples we present the theoretical and
experimental bases needed to simulate acurately the behaviour of oil
spills (or tracer particles) in a turbulent flow, in a simple and
efficient way that may be updated in an emergency with the latest output
from dedicated environmental Atmosphere (wind) and ocean currents and wave
nested models. This is acomplished with a Kinematic Simulation (KS) model
and in this work we compare some predictive results with detected oil
spills and field measurements.
There is a strong dependence of horizontal eddy diffusivities with the
Wave Reynolds number as well as with the wind stress measured as the
friction velocity from wind profiles measured at the coastline. Some of
these results have been published in Bezerra et al. (1998). Both effects
are important and give several decades of variation of eddy diffusivities
measured near the coastline (between 0.0001 and 2 m2s1).
A good estimate of the eddy diffusivity comes from a scaling that includes
the thickness of the surf zone as well as the depth and the wave period.
Measurements in the Mediterranean are almost two orders of magnitude
smaller than in the Pacific coast. On a larger scale, and further away
from the coast the relevant eddy diffusivities are much larger, because
large eddies, that often scale on the Rossby deformation radius disperse
further the spils. Examples of SAR image signatures of different oceanic
and atmospheric origin are compared, In the Figure a distribution of the
detected vortical structures and the size distribution of oil Spils are
presented with a set of eddy difusivity values measured near the Ebro
Delta.
Acknowledgements
This work was supported by the Ministerio de Educacion y
Ciencia of Spain and the Universitat Politecnica de Catalunya
(RYC2003005700) and from the European Space Agency, Proyect AOID
C1P.2240
References
Redondo, J.M. (1990) The structure of density interfaces. DAMTP. Cambridge
Univ. PhD Thesis
Derbyshire, S.H. and Redondo, J.M. (1990) Fractals and Waves, some
Geometrical Approaches to StablyStratified Turbulence . Anales de Fisica,
Serie A, 86.
Gade, M. y Redondo J.M., (1999). Marine pollution in European coastal
waters monitored by the ERS2 SAR: a comprehensive statistical analysis.
IGARSS 99, Hamburg. Vol. III, 1637 1639, 308312.
Redondo, J.M., Platonov, A.K. (2001) Aplication of the SAR images to
dynamic and pollution study of the Mediterranean Sea area near Barcelona.
Ingenieria del Agua, 8, 1523.
Redondo J.M., M.A. Sanchez, J.J. Martinez Benjamin and G. Jolly
(1998)Spectral study of the Ocean surface with SAR Earth Surface Remote
Sensing II ISBN0277786X , Eds. G. Cecchi and E. Zilioli, Vol
3496,217222.
Bezerra M.O., Diez M., Medeiros C., Rodriguez A., Bahia E. SanchezArcilla
A. y Redondo J.M. (1998) Study on the influence of waves on coastal
diffusion using image analysis. Jour. Flow Turbulence and Combustion. 59,
191204.
Figure: Oil Spill Analysis
Full paper
