20
T. Soomere
The classical problem of oil spill forecast is solved for the given location of the
accident, properties of the spill and current metocean conditions. Contemporary oil
spill models forecast the trajectory, fate and impact areas with a high accuracy. In
our work we employ several modifications of such models for modelling the fate
and drift of single pollution particles and oil spills (clusters of particles). We then
make use of them for solving a more complicated problem. The key challenge is to
establish for which release site the particular oil spill will cause the least damage
under the same metocean conditions.
Generally speaking, we are faced with an example of so-called inverse problems:
the identification of the ‘best’ location of the oil release presumes inverse tracking of
the pollution propagation. A straightforward solution to this class of problems is not
possible. 9 Moreover, no universal method exists for their analysis. This of course
does not mean that they should not be addressed at all. For the particular problem of
oil spill transport one can use, for example, a sequence of simulations with the same
release time but with slightly shifted release site. This variation of the ‘trial and
error’ method may give some hints about the fate of oil spills released in adjacent
locations. If applied systematically, one could estimate certain locations from where
the oil is less likely to be transported to some vulnerable spots, or from where the
transport will take a longer time. Furthermore, the least dangerous offshore areas
for a particular example of oil spill released at different locations can be estimated
based on a large enough number of trials.
This book describes how such a ‘trial and error’ method, to some extent resembling the Monte Carlo approach, is applied systematically to gather information
about the properties of the inverse problem of oil spill propagation. Its approximate
solution is sought by means of statistical analysis of a large number of trajectories
of single pollution particles released to the sea at different time instants and locations. Each trajectory is, in essence, an approximate solution of the associated direct
problem of propagation of a very small but persistent oil spill. The central idea of
the technology and its applications is to make use of the properties of Lagrangian
transport of such particles by surface currents.
A major technical problem is how to extract rational information from the vast
amount of numerically simulated data and how to build a reasonable implementation for the shipping industry with its specific needs and restrictions. Since the drift
of pollution or any other items in the sea is also affected by wind and waves, its
practical implementation generally presumes models integrating all potential factors. The most widespread state-of-the-art models of this type are operational oil
spill models. Their output will be used in comparisons of the results obtained from
circulation models. The importance of the local wind and wave fields and their impact to the resulting optimum fairways is established for the Gulf of Finland using
an advanced operational oil spill drift and fate model in Chap. 11.
9 Although several hydrodynamic or oil spill models are formally invertible (Ambjörn 2008), realistic 3D fluid dynamics is, in principle, non-invertible. Moreover, a unique solution to the governing
Navier–Stokes equations only exists during a finite time. Therefore, solving even the direct problem of oil spill propagation overrides this fundamental feature of fluid dynamics.
T. Soomere
The classical problem of oil spill forecast is solved for the given location of the
accident, properties of the spill and current metocean conditions. Contemporary oil
spill models forecast the trajectory, fate and impact areas with a high accuracy. In
our work we employ several modifications of such models for modelling the fate
and drift of single pollution particles and oil spills (clusters of particles). We then
make use of them for solving a more complicated problem. The key challenge is to
establish for which release site the particular oil spill will cause the least damage
under the same metocean conditions.
Generally speaking, we are faced with an example of so-called inverse problems:
the identification of the ‘best’ location of the oil release presumes inverse tracking of
the pollution propagation. A straightforward solution to this class of problems is not
possible. 9 Moreover, no universal method exists for their analysis. This of course
does not mean that they should not be addressed at all. For the particular problem of
oil spill transport one can use, for example, a sequence of simulations with the same
release time but with slightly shifted release site. This variation of the ‘trial and
error’ method may give some hints about the fate of oil spills released in adjacent
locations. If applied systematically, one could estimate certain locations from where
the oil is less likely to be transported to some vulnerable spots, or from where the
transport will take a longer time. Furthermore, the least dangerous offshore areas
for a particular example of oil spill released at different locations can be estimated
based on a large enough number of trials.
This book describes how such a ‘trial and error’ method, to some extent resembling the Monte Carlo approach, is applied systematically to gather information
about the properties of the inverse problem of oil spill propagation. Its approximate
solution is sought by means of statistical analysis of a large number of trajectories
of single pollution particles released to the sea at different time instants and locations. Each trajectory is, in essence, an approximate solution of the associated direct
problem of propagation of a very small but persistent oil spill. The central idea of
the technology and its applications is to make use of the properties of Lagrangian
transport of such particles by surface currents.
A major technical problem is how to extract rational information from the vast
amount of numerically simulated data and how to build a reasonable implementation for the shipping industry with its specific needs and restrictions. Since the drift
of pollution or any other items in the sea is also affected by wind and waves, its
practical implementation generally presumes models integrating all potential factors. The most widespread state-of-the-art models of this type are operational oil
spill models. Their output will be used in comparisons of the results obtained from
circulation models. The importance of the local wind and wave fields and their impact to the resulting optimum fairways is established for the Gulf of Finland using
an advanced operational oil spill drift and fate model in Chap. 11.
9 Although several hydrodynamic or oil spill models are formally invertible (Ambjörn 2008), realistic 3D fluid dynamics is, in principle, non-invertible. Moreover, a unique solution to the governing
Navier–Stokes equations only exists during a finite time. Therefore, solving even the direct problem of oil spill propagation overrides this fundamental feature of fluid dynamics.