206 Allan R. Robinson and Jurgen Sellschopp
the migration of fish stocks over one or two weeks is of great importance to the
fishing fleet. It is the latler forecasting problem which overlaps significantly water
column REA forecasting requirements. Currents transport and entrap nutrients,
phytoplankton, zooplankton and larvae, and the concentration distribution (mass
abundance) of adult fish shifts to seek food (e.g., plankton, larvae) to remain in a
comfortable environment (e.g., temperature, salinity), and to avoid predators.
This is only one example of many uses of REA involved in management and
operations in a multi-use coastal ocean and extended EEZs. Planned operations,
e.g. dredging, and sudden crises, e.g. oiI spills, ship and aircraft crashes, represent
events requiring REA. The ocean science and technology community is becoming
increasingly aware of the advantages of dual use research and the development of
dual use methodologies (military and civilian).
11.4 The REA OOPS (Ocean Observing and Prediction System)
11.4.1 The overall system and components
Purely deterministic processes such as the orbital motion of the planets can be
predicted by integration of the governing physicallaws, starting from precise initial
conditions at a certain time. In deterministic systems, inaccuracies of the initial
conditions can in principle be decreased below any given value by repeated observations and backward modeling. Systems that are subject to stochastic influences,
either implicitly through physical processes or by random forcing, can be predicted
only with limited accuracy. In the model of a linear system, through its dependence
on initialization and parameterization of the physics, the prediction error increases
gradually with time. In a nonlinear system, predictions can degrade rapidly when
nonlinear terms, including nonlinear transfer of error scales, amplify differences
between predictions and reality.
The physical ocean is a nonlinear system with inherent stochastic processes and
stochastic forcing. Physical laws are represented by the set of hydrodynamic and
thermodynamic equations, which in general have no closed form solution. Numerical ocean models integrate simplified versions of the dynamical equations, that
neglect physical processes having minor impact on the problem under consideration. The model formulation for an ocean circulation model, for example, would
include gravity and friction forces and would omit density fluctuations that are
responsible for sound propagation and vice versa.
A valid model, initialized with reasonable climatological fields and forced with
ac curate time dependent boundary conditions, can be expected to converge towards
reality after a sufficiently extended spin-up period. Numerical calculations of this
kind have been extensively used for studies of physical processes in the ocean, for
the explanation of inter-annual changes and for climate predictions. Internal
dynamics will produce meso-scale ocean features in an eddy-resolving model, if
the resolution of the forcing fields in space and time is not too coarse. In this case,
the migration of fish stocks over one or two weeks is of great importance to the
fishing fleet. It is the latler forecasting problem which overlaps significantly water
column REA forecasting requirements. Currents transport and entrap nutrients,
phytoplankton, zooplankton and larvae, and the concentration distribution (mass
abundance) of adult fish shifts to seek food (e.g., plankton, larvae) to remain in a
comfortable environment (e.g., temperature, salinity), and to avoid predators.
This is only one example of many uses of REA involved in management and
operations in a multi-use coastal ocean and extended EEZs. Planned operations,
e.g. dredging, and sudden crises, e.g. oiI spills, ship and aircraft crashes, represent
events requiring REA. The ocean science and technology community is becoming
increasingly aware of the advantages of dual use research and the development of
dual use methodologies (military and civilian).
11.4 The REA OOPS (Ocean Observing and Prediction System)
11.4.1 The overall system and components
Purely deterministic processes such as the orbital motion of the planets can be
predicted by integration of the governing physicallaws, starting from precise initial
conditions at a certain time. In deterministic systems, inaccuracies of the initial
conditions can in principle be decreased below any given value by repeated observations and backward modeling. Systems that are subject to stochastic influences,
either implicitly through physical processes or by random forcing, can be predicted
only with limited accuracy. In the model of a linear system, through its dependence
on initialization and parameterization of the physics, the prediction error increases
gradually with time. In a nonlinear system, predictions can degrade rapidly when
nonlinear terms, including nonlinear transfer of error scales, amplify differences
between predictions and reality.
The physical ocean is a nonlinear system with inherent stochastic processes and
stochastic forcing. Physical laws are represented by the set of hydrodynamic and
thermodynamic equations, which in general have no closed form solution. Numerical ocean models integrate simplified versions of the dynamical equations, that
neglect physical processes having minor impact on the problem under consideration. The model formulation for an ocean circulation model, for example, would
include gravity and friction forces and would omit density fluctuations that are
responsible for sound propagation and vice versa.
A valid model, initialized with reasonable climatological fields and forced with
ac curate time dependent boundary conditions, can be expected to converge towards
reality after a sufficiently extended spin-up period. Numerical calculations of this
kind have been extensively used for studies of physical processes in the ocean, for
the explanation of inter-annual changes and for climate predictions. Internal
dynamics will produce meso-scale ocean features in an eddy-resolving model, if
the resolution of the forcing fields in space and time is not too coarse. In this case,