210 AlIan R. Robinson and Jurgen Sellschopp
predictions in a parameter subspace have high impact on operations. Depending on
what the operations are, predictive capability would be estimated differently, obviating a general reliability assessment scheme. As a compromise, one might investigate the change of patterns and features and compare with prediction. At least the
tendency, more pleasingly also the magnitude of changes, should match (Sellschopp and Robinson, 1997).
Processes with length scales comparable to or smaller than the cells in the
numerical scheme cannot be forecasted. But also time and space resolution of the
forcing fields, which is coarser than the grid ofthe ocean model, limit predictability. The reliability of forecasted atmospheric forcing fields can be of minor importance if internal dynamics exceed external forces, otherwise it has immediate
influence on ocean prediction quality. If ocean internal dynamics is dominant, the
somewhat longer ocean predictability time is then limiting.
Validation, calibration and verification of a generic REA system rapidly deployable in any region of the global coastal ocean is a challenging and demanding task.
Curtin et al. (1993) advocate a series of Coastal Ocean Predictive Skill Experiments (CPSE), which are the essential elements ofthe predictive phase of regional
forecast system development discussed above. Since REA predictive skill experiments must be designed to determine forecast skill on the basis of minimal and
covertly attainable observations, they may be most efficiently carried out in the
context of the definitive over-sampling provided by a series of regional CPSEs
encompassing a broad range of coastal processes.
11.6 Illustrations from Rapid Response Exercises 1996, 1997,
1998
In three consecutive years from 1996 to 1998, REA exercises have taken place
in, and at the entrance to, the Mediterranean Sea. They were connected with NATO
naval exercises that sm~ceeded the REA survey in the same areas. Up to 8 ships, 6
patrol aircraft and numerous institutions have been active in REA. For support of
warfare commanders, data were collected, only part ofwhich was oceanographic in
a narrow sense. The rest was for sea bottom and beach assessment. Measurements
for real-time analysis of the physical ocean consisted of CTD stations, XBT and
XCTD casts from ships and aircraft, shipborne ADCP velocities, surface drifter
deployments and satellite images. Self-recording current meters were used for a
posteriori validation.
Nowcast and forecast results from ocean modeling were delivered in a format
suitable for the customer, who was most interested in surface currents and sound
velocity profiles. Maps and vertical sections were produced as image files for electronic distribution. Profiles were reduced to inflection points coded in a special format for automatic insertion into navy environmental systems.
predictions in a parameter subspace have high impact on operations. Depending on
what the operations are, predictive capability would be estimated differently, obviating a general reliability assessment scheme. As a compromise, one might investigate the change of patterns and features and compare with prediction. At least the
tendency, more pleasingly also the magnitude of changes, should match (Sellschopp and Robinson, 1997).
Processes with length scales comparable to or smaller than the cells in the
numerical scheme cannot be forecasted. But also time and space resolution of the
forcing fields, which is coarser than the grid ofthe ocean model, limit predictability. The reliability of forecasted atmospheric forcing fields can be of minor importance if internal dynamics exceed external forces, otherwise it has immediate
influence on ocean prediction quality. If ocean internal dynamics is dominant, the
somewhat longer ocean predictability time is then limiting.
Validation, calibration and verification of a generic REA system rapidly deployable in any region of the global coastal ocean is a challenging and demanding task.
Curtin et al. (1993) advocate a series of Coastal Ocean Predictive Skill Experiments (CPSE), which are the essential elements ofthe predictive phase of regional
forecast system development discussed above. Since REA predictive skill experiments must be designed to determine forecast skill on the basis of minimal and
covertly attainable observations, they may be most efficiently carried out in the
context of the definitive over-sampling provided by a series of regional CPSEs
encompassing a broad range of coastal processes.
11.6 Illustrations from Rapid Response Exercises 1996, 1997,
1998
In three consecutive years from 1996 to 1998, REA exercises have taken place
in, and at the entrance to, the Mediterranean Sea. They were connected with NATO
naval exercises that sm~ceeded the REA survey in the same areas. Up to 8 ships, 6
patrol aircraft and numerous institutions have been active in REA. For support of
warfare commanders, data were collected, only part ofwhich was oceanographic in
a narrow sense. The rest was for sea bottom and beach assessment. Measurements
for real-time analysis of the physical ocean consisted of CTD stations, XBT and
XCTD casts from ships and aircraft, shipborne ADCP velocities, surface drifter
deployments and satellite images. Self-recording current meters were used for a
posteriori validation.
Nowcast and forecast results from ocean modeling were delivered in a format
suitable for the customer, who was most interested in surface currents and sound
velocity profiles. Maps and vertical sections were produced as image files for electronic distribution. Profiles were reduced to inflection points coded in a special format for automatic insertion into navy environmental systems.