340 Nadia Pinardi et al.
such systems should be part ofthe information database that could support the handling of marine disasters and emergencies.
The scientific strategy pursued in this paper is the one that considers monitoring
networks producing the bulk of the observations capable of sustaining the modeling and prediction system. The modeling and prediction system are also permanent, meaning that a careful calibration of model parameters and validation of
model simulation results is done before the observing and the model system is
changed, based upon the experience with the previous permanent system components. A coastal environmental prediction system is, in our mind, based upon permanent observing systems, efficiently and scientifically designed to fit the
modeling needs and continuously re-evaluated by producing forecasts which will
enable to check the quality of both the observing and modeling components. This
permanent system will support, in the future, the climate observing system in the
coastal marine areas, producing long time series of data for trend analysis and global change signals detection.
In addition to the choice of a sustained monitoring and modeling system as a
base to understand and predict the marine ecosystem, we concentrate our attention
in the marine coastal areas, where hydrodynamic and biogeochemical cycles have
short and small time and space scales. The observing system should resolve a relevant part of the dynamical field variability so that optimal estimates of the ecosystem state variables can be done in order to initialize forecasts. Both the
measurement technologies and the mode ling tools are barely adequate for coastal
marine predictions, since there is a need to measure simultaneously physical and
biogeochemical parameters at the synoptic time scales and the hydrodynamic models need high resolution.
Our study area is the Mediterranean Sea where a Mediterranean Forecasting System approach has started to be developed, which aims at the 'understanding and
prediction of coastal primary producers biomass variability from the time scales of
days to months' (Pinardi and Flemming, 1998). Brink and Robinson (1998) have
recently discussed the coastal areas ofthe World Ocean in a global overview. In our
view, the different coastal area ecosystems could be classified on the basis of the
different values taken by a ratio between river and large scale flow field forcing in
the shelf/coastal areas. This ratio in fact sets the basic hydrodynamic response, the
nutrient and light levels. In the Mediterranean Sea very different coastal ecosystems coexist and in general this ratio varies an order of magnitude. We concentrate
our test case in a river dominated shelf area such as the Northem Adriatic Sea
where a monitoring approach toward environmental predictions has started.
Following Robinson and Leslie (1985), the forecasting capability can be
achieved if three basic phases of scientific research and technology development
are attained. The first phase consists of historical data collection, specific process
oriented data collection exercises in dynamically active areas, numerical and theoreti cal dynamical studies of relevant processes. This phase is called 'descriptive'
and in fisheries system analysis can be called 'conceptual model definition' (Grant,
1986). The second phase considers the calibration/validation ofthe numerical mod-
such systems should be part ofthe information database that could support the handling of marine disasters and emergencies.
The scientific strategy pursued in this paper is the one that considers monitoring
networks producing the bulk of the observations capable of sustaining the modeling and prediction system. The modeling and prediction system are also permanent, meaning that a careful calibration of model parameters and validation of
model simulation results is done before the observing and the model system is
changed, based upon the experience with the previous permanent system components. A coastal environmental prediction system is, in our mind, based upon permanent observing systems, efficiently and scientifically designed to fit the
modeling needs and continuously re-evaluated by producing forecasts which will
enable to check the quality of both the observing and modeling components. This
permanent system will support, in the future, the climate observing system in the
coastal marine areas, producing long time series of data for trend analysis and global change signals detection.
In addition to the choice of a sustained monitoring and modeling system as a
base to understand and predict the marine ecosystem, we concentrate our attention
in the marine coastal areas, where hydrodynamic and biogeochemical cycles have
short and small time and space scales. The observing system should resolve a relevant part of the dynamical field variability so that optimal estimates of the ecosystem state variables can be done in order to initialize forecasts. Both the
measurement technologies and the mode ling tools are barely adequate for coastal
marine predictions, since there is a need to measure simultaneously physical and
biogeochemical parameters at the synoptic time scales and the hydrodynamic models need high resolution.
Our study area is the Mediterranean Sea where a Mediterranean Forecasting System approach has started to be developed, which aims at the 'understanding and
prediction of coastal primary producers biomass variability from the time scales of
days to months' (Pinardi and Flemming, 1998). Brink and Robinson (1998) have
recently discussed the coastal areas ofthe World Ocean in a global overview. In our
view, the different coastal area ecosystems could be classified on the basis of the
different values taken by a ratio between river and large scale flow field forcing in
the shelf/coastal areas. This ratio in fact sets the basic hydrodynamic response, the
nutrient and light levels. In the Mediterranean Sea very different coastal ecosystems coexist and in general this ratio varies an order of magnitude. We concentrate
our test case in a river dominated shelf area such as the Northem Adriatic Sea
where a monitoring approach toward environmental predictions has started.
Following Robinson and Leslie (1985), the forecasting capability can be
achieved if three basic phases of scientific research and technology development
are attained. The first phase consists of historical data collection, specific process
oriented data collection exercises in dynamically active areas, numerical and theoreti cal dynamical studies of relevant processes. This phase is called 'descriptive'
and in fisheries system analysis can be called 'conceptual model definition' (Grant,
1986). The second phase considers the calibration/validation ofthe numerical mod-