IN-SITU OBSERVATIONS
193
current systems or passages. This however is not exploited in current
assimilation and forecasting approaches.
Increasingly, operational systems also need to and do address ecosystem
dynamics and biogeochemical cycles in the ocean. This is especially true for
the regional and coastal applications. In such cases, a larger range of
variables is required from an observing system. At the minimum level this
includes the prime variables like oxygen, nutrients, chlorophyll or
phytoplankton biomass, and zooplankton biomass (e.g. Fasham et al, 1993).
In more small-scale applications, it may require the knowledge of individual
species of plankton or fish, or of certain chemicals (specific nutrients, trace
elements). For very specific purposes, long lists of variables can be drawn
up, which however is not useful here since the goal of this presentation is the
nature of typical observing systems and not the exhaustive coverage of
singular cases.
1.2 Coverage and time-space sampling
The ideal observing system, both for research and for operational
applications, covers the three space (x,y,z) and time (t) dimensions
“completely”. The word ‘covering’ usually denotes the extent/reach in the
four dimensions. All current sampling techniques are discrete, however, in
these dimensions, and ‘completely’ therefore must also be interpreted as
having sufficient resolution to reveal the smallest scales of the variabilities
of interest.
Thus, rigorously, for each application, a new system would need to be
designed which can deliver the needed observations with the accuracy and
the sampling specific to the needs. In general, this is not feasible. Also the
envisioned 4-D sampling is not possible with current technology. There are,
however, various techniques which provide different sections through this 4-D
space with useful resolution in at least some of the dimensions, see Figure 1.
Satellites have excellent x-y-t coverage, and sufficient x-y-t resolution for
many applications. However, the sampling is provided only at or near the
surface, and is restricted to few variables. The ARGO float network provides
good sampling in the vertical (profiles), with global coverage and hopefully
long (sustained) coverage in time, but has sparse horizontal and temporal
resolution despite the large number of platforms (e.g. not eddy-resolving,
unable to resolve the timescale of short events, etc). Also the number/types
of variables observable with the ARGO system will remain very limited.
Fixed (moored) instruments can deliver excellent (probably complete)
sampling of the time domain and may have good coverage of the
z-dimensions, but can only be installed in a few number of x-y locations.
Therefore, a coordinated and deliberate use of several observing techniques
often is required to provide the information needed.
193
current systems or passages. This however is not exploited in current
assimilation and forecasting approaches.
Increasingly, operational systems also need to and do address ecosystem
dynamics and biogeochemical cycles in the ocean. This is especially true for
the regional and coastal applications. In such cases, a larger range of
variables is required from an observing system. At the minimum level this
includes the prime variables like oxygen, nutrients, chlorophyll or
phytoplankton biomass, and zooplankton biomass (e.g. Fasham et al, 1993).
In more small-scale applications, it may require the knowledge of individual
species of plankton or fish, or of certain chemicals (specific nutrients, trace
elements). For very specific purposes, long lists of variables can be drawn
up, which however is not useful here since the goal of this presentation is the
nature of typical observing systems and not the exhaustive coverage of
singular cases.
1.2 Coverage and time-space sampling
The ideal observing system, both for research and for operational
applications, covers the three space (x,y,z) and time (t) dimensions
“completely”. The word ‘covering’ usually denotes the extent/reach in the
four dimensions. All current sampling techniques are discrete, however, in
these dimensions, and ‘completely’ therefore must also be interpreted as
having sufficient resolution to reveal the smallest scales of the variabilities
of interest.
Thus, rigorously, for each application, a new system would need to be
designed which can deliver the needed observations with the accuracy and
the sampling specific to the needs. In general, this is not feasible. Also the
envisioned 4-D sampling is not possible with current technology. There are,
however, various techniques which provide different sections through this 4-D
space with useful resolution in at least some of the dimensions, see Figure 1.
Satellites have excellent x-y-t coverage, and sufficient x-y-t resolution for
many applications. However, the sampling is provided only at or near the
surface, and is restricted to few variables. The ARGO float network provides
good sampling in the vertical (profiles), with global coverage and hopefully
long (sustained) coverage in time, but has sparse horizontal and temporal
resolution despite the large number of platforms (e.g. not eddy-resolving,
unable to resolve the timescale of short events, etc). Also the number/types
of variables observable with the ARGO system will remain very limited.
Fixed (moored) instruments can deliver excellent (probably complete)
sampling of the time domain and may have good coverage of the
z-dimensions, but can only be installed in a few number of x-y locations.
Therefore, a coordinated and deliberate use of several observing techniques
often is required to provide the information needed.