358 Nadia Pinardi et al.
observations are most needed in the very shallow coastal strips, where the dynamical regimes could be local and the satellite data fails to give accurate information.
17.5 The modeling system and the data assimilation components
The ocean models for predictions started to be developed at the beginning of the
eigthies and they consisted of simplified physics equations (quasigeostrophy, Robinson and Walstad, 1987, multilayer reduced gravity, Hurlburt, 1986, shallow
water, Prandle, 2001). They were used to forecast the mesoscale variability and the
coastal sea level. At the contrary of the atmosphere, ocean forecasting started from
the mesoscale and the coastal areas, in view of the high observing system requirements and following the users needs.
Data assimilation techniques were developed first for mesoscale forecasting
experiments to control the problems of initialization that can severely limit the predictability time (Walstad and Robinson, 1990). The assimilation techniques concentrated on the usage of satellite altimetry (De Mey and Robinson, 1987), sea
surf ace temperature and temperature profiles (Derber and Rosati, 1989). They all
set up the scene for the advancements in ocean data assimilation ofthe second half
ofthe eighties and the nineties (Fukumori et al., 1993, Carton et al., 2000, Masina
et al., 2001, De Mey and Benkiran, 2001).
In the following sections we will try to overview the state of the art theoretical
framework for the hydrodynamic and ecosystem modeling. The data assimilation
component will be only mentioned and the interested reader should consult the data
assimilation chapters ofthis book.
17.5.1 The large and coastal scale hydrodynamic modeling
Hydrodynamic modeling in the ocean is based upon numerical models developed
in the seventies. Recently new models have been developed (Haidvogel and Beckman, 1998) but they have not been extensively used yet for forecasting. Here we
illustrate two of the most commonly used models, one for the large and the other
for the shelf/coastal scale. We distinguish between the two scales because historically models were developed in a different way for the two different environments,
even if they are quite similar at the level of goveming equations. The large scale
modeling community concentrated mainly in the along isopycnal and non-viscous
dynamics, thus building models with different subgrid scale and physical turbulence parameterizations from the coastal models. The latter gave attention to the
bottom boundary layer and the mixing parameterizations, the sea level variability
connected to ageostrophic effects, such as storm surge and tidal dynamics. The two
modeling communities started to merge in the nineties but several differences
remain in model formulation even today.
The large scale model described here is a version of the Modular Ocean Model
(MOM, Bryan, 1969, Pacanowsky et al., 1993) that is a rigid lid, primitive equation
model with centered finite differences in horizontal and vertical. The coastal scale
observations are most needed in the very shallow coastal strips, where the dynamical regimes could be local and the satellite data fails to give accurate information.
17.5 The modeling system and the data assimilation components
The ocean models for predictions started to be developed at the beginning of the
eigthies and they consisted of simplified physics equations (quasigeostrophy, Robinson and Walstad, 1987, multilayer reduced gravity, Hurlburt, 1986, shallow
water, Prandle, 2001). They were used to forecast the mesoscale variability and the
coastal sea level. At the contrary of the atmosphere, ocean forecasting started from
the mesoscale and the coastal areas, in view of the high observing system requirements and following the users needs.
Data assimilation techniques were developed first for mesoscale forecasting
experiments to control the problems of initialization that can severely limit the predictability time (Walstad and Robinson, 1990). The assimilation techniques concentrated on the usage of satellite altimetry (De Mey and Robinson, 1987), sea
surf ace temperature and temperature profiles (Derber and Rosati, 1989). They all
set up the scene for the advancements in ocean data assimilation ofthe second half
ofthe eighties and the nineties (Fukumori et al., 1993, Carton et al., 2000, Masina
et al., 2001, De Mey and Benkiran, 2001).
In the following sections we will try to overview the state of the art theoretical
framework for the hydrodynamic and ecosystem modeling. The data assimilation
component will be only mentioned and the interested reader should consult the data
assimilation chapters ofthis book.
17.5.1 The large and coastal scale hydrodynamic modeling
Hydrodynamic modeling in the ocean is based upon numerical models developed
in the seventies. Recently new models have been developed (Haidvogel and Beckman, 1998) but they have not been extensively used yet for forecasting. Here we
illustrate two of the most commonly used models, one for the large and the other
for the shelf/coastal scale. We distinguish between the two scales because historically models were developed in a different way for the two different environments,
even if they are quite similar at the level of goveming equations. The large scale
modeling community concentrated mainly in the along isopycnal and non-viscous
dynamics, thus building models with different subgrid scale and physical turbulence parameterizations from the coastal models. The latter gave attention to the
bottom boundary layer and the mixing parameterizations, the sea level variability
connected to ageostrophic effects, such as storm surge and tidal dynamics. The two
modeling communities started to merge in the nineties but several differences
remain in model formulation even today.
The large scale model described here is a version of the Modular Ocean Model
(MOM, Bryan, 1969, Pacanowsky et al., 1993) that is a rigid lid, primitive equation
model with centered finite differences in horizontal and vertical. The coastal scale