192
UWE SEND
properties like heat content, stratification, transports (mass and heat),
mixing, and air-sea fluxes. They need to have accurate enough forcing and
fluxes to run without bias for weeks and months. The biogeochemical
components of global models are still uncertain enough to make data
requirements less stringent – the most useful contribution of data is a good
representation of the overall biogeochemical regime conditions on the large
scale for a few basic variables. Coastal models usually are run only for a few
days but focus on small scales, and often try to represent detailed ecosystem
species, both of which require more detailed observations. These models are
strongly affected by the small-scale advective processes and less sensitive to
small errors in the forcing. However, mixing, which changes stratification
but also affects e.g. nutrients, is equally critical as in global models.
Regional applications are somewhere in the middle. They typically address
conditions and changes in sub-basins (e.g. Nordic Seas) or marginal seas
(e.g. the Mediterranean). Mesoscale features generally still need to be
resolved and correctly represented.
1.1 Variables
The description of the physical state of the ocean requires the density and
temperature (T)/salinity (S) fields (not independent of course), as well as the
absolute currents. Closely coupled to this is the physical forcing at the
surface (e.g. wind, radiation, heat), which is covered in the chapter by W.
Large (this volume). For biogeochemical models, the basic state is described
by the variables, such as nutrients, phytoplankton, zooplankton and detritus
in the simplest cases. These state variables, which take values everywhere in
the model domain, need to be distinguished from the quantities which are
forecast – these are sometimes derived quantities or not predicted
everywhere (e.g. only at the surface).
The prime variables forecast in current operational systems are
temperature and currents, with a focus on surface fields. The rationale for
this emphasis is a combination of models being primarily physical, remote
sensing delivering these variables at the surface, and many applications
needing this information at the surface. For predicting ocean circulation,
however, the associated interior density field needs to be known. Many
applications also require the vertical stratification of temperature (i.e. heat
content), and density stratification (for pollutant dispersal/mixing or
fisheries). Some defense applications also seek the interior sound speed
distribution which is calculated from subsurface T and S. Thus a minimum
data requirement for physical models is the full density field and absolute
currents at some level, unless data are only used for validation (in this case,
selected locations or layers may be sufficient). Conceptually also integral
properties should be important for constraining and initializing models, like
a water-mass or basin heat content, or transports of mass and heat in major
UWE SEND
properties like heat content, stratification, transports (mass and heat),
mixing, and air-sea fluxes. They need to have accurate enough forcing and
fluxes to run without bias for weeks and months. The biogeochemical
components of global models are still uncertain enough to make data
requirements less stringent – the most useful contribution of data is a good
representation of the overall biogeochemical regime conditions on the large
scale for a few basic variables. Coastal models usually are run only for a few
days but focus on small scales, and often try to represent detailed ecosystem
species, both of which require more detailed observations. These models are
strongly affected by the small-scale advective processes and less sensitive to
small errors in the forcing. However, mixing, which changes stratification
but also affects e.g. nutrients, is equally critical as in global models.
Regional applications are somewhere in the middle. They typically address
conditions and changes in sub-basins (e.g. Nordic Seas) or marginal seas
(e.g. the Mediterranean). Mesoscale features generally still need to be
resolved and correctly represented.
1.1 Variables
The description of the physical state of the ocean requires the density and
temperature (T)/salinity (S) fields (not independent of course), as well as the
absolute currents. Closely coupled to this is the physical forcing at the
surface (e.g. wind, radiation, heat), which is covered in the chapter by W.
Large (this volume). For biogeochemical models, the basic state is described
by the variables, such as nutrients, phytoplankton, zooplankton and detritus
in the simplest cases. These state variables, which take values everywhere in
the model domain, need to be distinguished from the quantities which are
forecast – these are sometimes derived quantities or not predicted
everywhere (e.g. only at the surface).
The prime variables forecast in current operational systems are
temperature and currents, with a focus on surface fields. The rationale for
this emphasis is a combination of models being primarily physical, remote
sensing delivering these variables at the surface, and many applications
needing this information at the surface. For predicting ocean circulation,
however, the associated interior density field needs to be known. Many
applications also require the vertical stratification of temperature (i.e. heat
content), and density stratification (for pollutant dispersal/mixing or
fisheries). Some defense applications also seek the interior sound speed
distribution which is calculated from subsurface T and S. Thus a minimum
data requirement for physical models is the full density field and absolute
currents at some level, unless data are only used for validation (in this case,
selected locations or layers may be sufficient). Conceptually also integral
properties should be important for constraining and initializing models, like
a water-mass or basin heat content, or transports of mass and heat in major