352 Nadia Pinardi et al.
forcing fields should be at relatively high resolution (10 km or less) and used at relatively high frequency (only 1-2 hours instead of six or more) which adds the need
for a systematic increase of the atmospheric operational models resolution.
The ecosystem models for the coastal areas should consider quite complex food
web interactions and several species of primary producers functional groups to
cope with the relatively high biodiversity of the coastal ecosystem at the pelagic
and benthic levels. The relative availability of multidisciplinary data sets in coastal
areas with respect to the open ocean makes it possible to check and calibrate severaI parameters of complex ecosystem models. Such complex models could be able
in the future to predict harmful algae blooms and help to assess health and productivity of the ocean issues associated with changes in community structure and species.
The data assimilation component is a relatively young branch of research in
oceanography and a very new one for ecosystem models. However, in the past ten
years, the large-scale hydrodynamic community has developed robust assimilation
methods for all the observational data sets discussed above and pre-operational
ocean forecasting systems are now capable to assimilate a combination of satellite
and in situ measurements collected at the global scale. Data assimilation for coastal
hydrodynamic models is very novel and only few examples exist of successful initialization with in situ data sets (Lermusiaux, 2001). In addition to the traditional
problem of assimilating the available data, the nesting of the coastal model within
the coarser model introduces the coarser model fields as an additional type of data
to be considered for assimilation. The coarser grid information is very important
and should be melded efficiently with the observational data in order to specify an
optimal initial condition for the high-resolution coastal model. A tirst solution to
this problem has been found by using a tangent linear model to downscale the initial and boundary conditions from a coarse to an high-resolution model (Auclair,
1999). This technique is very promising but in the future it should be used in conjunction with the assimilation of observed data.
In synthesis, the coastal environmental prediction problem is dependent upon the
existence of: 1) large scale optimal estimates for initial and boundary conditions; 2)
high resolution atmospheric models; 3) complex ecosystem models which require
multidisciplinary data sets and river run-off and nutrient loading data for initialization and forcing; 4) data assimilation techniques which could consider coarse resolution fields and field observations in the right sequence.
In the following section we show the application of some of the concepts of
Fig. 17.5 in the Mediterranean Sea.
forcing fields should be at relatively high resolution (10 km or less) and used at relatively high frequency (only 1-2 hours instead of six or more) which adds the need
for a systematic increase of the atmospheric operational models resolution.
The ecosystem models for the coastal areas should consider quite complex food
web interactions and several species of primary producers functional groups to
cope with the relatively high biodiversity of the coastal ecosystem at the pelagic
and benthic levels. The relative availability of multidisciplinary data sets in coastal
areas with respect to the open ocean makes it possible to check and calibrate severaI parameters of complex ecosystem models. Such complex models could be able
in the future to predict harmful algae blooms and help to assess health and productivity of the ocean issues associated with changes in community structure and species.
The data assimilation component is a relatively young branch of research in
oceanography and a very new one for ecosystem models. However, in the past ten
years, the large-scale hydrodynamic community has developed robust assimilation
methods for all the observational data sets discussed above and pre-operational
ocean forecasting systems are now capable to assimilate a combination of satellite
and in situ measurements collected at the global scale. Data assimilation for coastal
hydrodynamic models is very novel and only few examples exist of successful initialization with in situ data sets (Lermusiaux, 2001). In addition to the traditional
problem of assimilating the available data, the nesting of the coastal model within
the coarser model introduces the coarser model fields as an additional type of data
to be considered for assimilation. The coarser grid information is very important
and should be melded efficiently with the observational data in order to specify an
optimal initial condition for the high-resolution coastal model. A tirst solution to
this problem has been found by using a tangent linear model to downscale the initial and boundary conditions from a coarse to an high-resolution model (Auclair,
1999). This technique is very promising but in the future it should be used in conjunction with the assimilation of observed data.
In synthesis, the coastal environmental prediction problem is dependent upon the
existence of: 1) large scale optimal estimates for initial and boundary conditions; 2)
high resolution atmospheric models; 3) complex ecosystem models which require
multidisciplinary data sets and river run-off and nutrient loading data for initialization and forcing; 4) data assimilation techniques which could consider coarse resolution fields and field observations in the right sequence.
In the following section we show the application of some of the concepts of
Fig. 17.5 in the Mediterranean Sea.