208 Alian R. Robinson and Jurgen Sellschopp
and forcing fields. Since they are only required for the compensation of developing
model deviations, they are most effective in those parts of the parameter space that
are most difficult to predict. The value of sparse observations may be extended by
applying models that describe typical ocean features.
Data managers in an REA system are responsible for time1y data transfers
between the observational network, the assimilation and model ing effort and the
users of observational and forecast products. Data formats should be pre-defined,
but in the haste of real time operations, data cannot be guaranteed to be fully quality controlled. Data users, especially modelers, will check again. A single centralized authority for REA data management could slow down the final delivery.
Collaborators should rather be connected in a network, preferably the Internet with
data access restricted to authorized sites. Data originators or sub-nodes to which
they report can keep the data on their computers and allow others to share the information. There should be a site that acts as an REA data fusion center. It has the
responsibility to keep track of all information offered by participants, to maintain
inventories and to establish links. A distributed system for data management is
more flexible and potentially faster than a centralized solution.
The number of permanent platforms such as weather ships, tide gauges or wave
poles, is very limited, and a sufficient number can hardly be expected in an area
designated for REA. Quasi-permanent platforms are available through satellites
that regularly look at the sea surface. Time intervals between repeated coverage of
the same area ranges from hours for sea color and surface temperature to weeks for
synthetic aperture radar imaging and altimetry. REA can take advantage of satellite
data, but for littoral areas substantial information must come from dedicated
smaller scale surveys.
Platforms used on demand are moored or drifting buoys, autonomous underwater
vehicles, ships and aircraft. The most important in situ sensor in the assessment of
the physical ocean measures temperature, electric al conductivity and pressure in
the water column, which through the equations of thermodynamics also results in
salinity, density and sound velocity. Conductivity-Temperature-Depth (CTD) sensors are lowered from a stationary ship, are towed in a yo-yo-ing underwater body
or in a wide aperture multi-sensor chain. As expendables (XCTD), they are
dropped from a ship or aircraft. Because of the costs, XCTDs are mostly replaced
by expendable thermometers (XBT) and the data is assimilated without salinity or
with an assumed relation between temperature and salinity, which is usually a good
assumption in the open ocean but can be misleading in a littoral environment.
11.5 Forecast System and Predictive Skill Evaluations
11.5.1 Regional System Validation, Calibration and Verification
In the development of any regional forecast system, evaluation of the integrated
system, the system components and the forecast products is of course essential.
The development of a regional system is usefully conceptualized as having three
and forcing fields. Since they are only required for the compensation of developing
model deviations, they are most effective in those parts of the parameter space that
are most difficult to predict. The value of sparse observations may be extended by
applying models that describe typical ocean features.
Data managers in an REA system are responsible for time1y data transfers
between the observational network, the assimilation and model ing effort and the
users of observational and forecast products. Data formats should be pre-defined,
but in the haste of real time operations, data cannot be guaranteed to be fully quality controlled. Data users, especially modelers, will check again. A single centralized authority for REA data management could slow down the final delivery.
Collaborators should rather be connected in a network, preferably the Internet with
data access restricted to authorized sites. Data originators or sub-nodes to which
they report can keep the data on their computers and allow others to share the information. There should be a site that acts as an REA data fusion center. It has the
responsibility to keep track of all information offered by participants, to maintain
inventories and to establish links. A distributed system for data management is
more flexible and potentially faster than a centralized solution.
The number of permanent platforms such as weather ships, tide gauges or wave
poles, is very limited, and a sufficient number can hardly be expected in an area
designated for REA. Quasi-permanent platforms are available through satellites
that regularly look at the sea surface. Time intervals between repeated coverage of
the same area ranges from hours for sea color and surface temperature to weeks for
synthetic aperture radar imaging and altimetry. REA can take advantage of satellite
data, but for littoral areas substantial information must come from dedicated
smaller scale surveys.
Platforms used on demand are moored or drifting buoys, autonomous underwater
vehicles, ships and aircraft. The most important in situ sensor in the assessment of
the physical ocean measures temperature, electric al conductivity and pressure in
the water column, which through the equations of thermodynamics also results in
salinity, density and sound velocity. Conductivity-Temperature-Depth (CTD) sensors are lowered from a stationary ship, are towed in a yo-yo-ing underwater body
or in a wide aperture multi-sensor chain. As expendables (XCTD), they are
dropped from a ship or aircraft. Because of the costs, XCTDs are mostly replaced
by expendable thermometers (XBT) and the data is assimilated without salinity or
with an assumed relation between temperature and salinity, which is usually a good
assumption in the open ocean but can be misleading in a littoral environment.
11.5 Forecast System and Predictive Skill Evaluations
11.5.1 Regional System Validation, Calibration and Verification
In the development of any regional forecast system, evaluation of the integrated
system, the system components and the forecast products is of course essential.
The development of a regional system is usefully conceptualized as having three