166
IAN ROBINSON
level 3. In all cases some measure of C is produced and an estimate of K is
derived globally. These are generally reliable products and C approaches the
target accuracy of 30% in open sea Case 1 waters. However great care must
be taken when using the products over coastal and shelf sea (possibly Case
2) waters where the potentially large errors could give misleading
information. Some of the agencies attempt to provide a number of other
products such as SPM and CDOM but these are yet to be proven.
3.2.4
Using satellite ocean colour data in ocean models
The use of ocean colour derived data products in ocean models is in its
infancy. The long and successful deployment of SeaWiFS has given some
confidence that ocean colour sensors are capable of supplying data for
operational applications. However the disappointing loss of two excellent
sensors (OCTS and GLI) through spacecraft failure, and the difficulties with
calibrating the MODIS/Terra products, have slowed down any moves in this
direction. It is to be hoped that, when MODIS/Aqua and MERIS are fully
proven and delivering data products routinely within a few hours of
acquisition, they will establish an even better operational supply of data than
SeaWiFS which is approaching the end of its operational life.
There are two main ways in which ocean colour data are likely to be used
operationally. The first is to use measurements of C to improve the
modelling of phytoplankton biomass in numerical ocean models which
contain a biogeochemical, phytoplankton or carbon cycle component. The
uncertainties associated with modelling biological populations are such that
improvements can be gained by assimilating or otherwise ingesting
measurements of C even when their accuracy is no better than 30%. So far
the most promising approach has been to use the satellite observations to
identify when and where a phytoplankton bloom emerges, since it is
particularly difficult for a model to trigger the initiation of a bloom. Satellite
data are used to update or re-initialise a model with the newly emerged
bloom conditions. The problem of cloud cover and the uncertain accuracy
make the data less useful to the model for the stage when the modelled
bloom has peaked and then gradually decays.
The other main use of colour data is in providing K for physical models
which need to know how far solar radiation penetrates into the sea. Thus
models of mixed layer development and the occurrence of a diurnal
thermocline, which are sensitive to K, can benefit from being regularly
updated with information about how K is distributed and varies in space and
time.
Another factor which still needs more investigation is the relationship
between the near-surface measurements of C which satellites provide and the
distribution of C with depth. Because phytoplankton need light to thrive,
then it is reasonable to suppose that most will be detected by ocean colour
IAN ROBINSON
level 3. In all cases some measure of C is produced and an estimate of K is
derived globally. These are generally reliable products and C approaches the
target accuracy of 30% in open sea Case 1 waters. However great care must
be taken when using the products over coastal and shelf sea (possibly Case
2) waters where the potentially large errors could give misleading
information. Some of the agencies attempt to provide a number of other
products such as SPM and CDOM but these are yet to be proven.
3.2.4
Using satellite ocean colour data in ocean models
The use of ocean colour derived data products in ocean models is in its
infancy. The long and successful deployment of SeaWiFS has given some
confidence that ocean colour sensors are capable of supplying data for
operational applications. However the disappointing loss of two excellent
sensors (OCTS and GLI) through spacecraft failure, and the difficulties with
calibrating the MODIS/Terra products, have slowed down any moves in this
direction. It is to be hoped that, when MODIS/Aqua and MERIS are fully
proven and delivering data products routinely within a few hours of
acquisition, they will establish an even better operational supply of data than
SeaWiFS which is approaching the end of its operational life.
There are two main ways in which ocean colour data are likely to be used
operationally. The first is to use measurements of C to improve the
modelling of phytoplankton biomass in numerical ocean models which
contain a biogeochemical, phytoplankton or carbon cycle component. The
uncertainties associated with modelling biological populations are such that
improvements can be gained by assimilating or otherwise ingesting
measurements of C even when their accuracy is no better than 30%. So far
the most promising approach has been to use the satellite observations to
identify when and where a phytoplankton bloom emerges, since it is
particularly difficult for a model to trigger the initiation of a bloom. Satellite
data are used to update or re-initialise a model with the newly emerged
bloom conditions. The problem of cloud cover and the uncertain accuracy
make the data less useful to the model for the stage when the modelled
bloom has peaked and then gradually decays.
The other main use of colour data is in providing K for physical models
which need to know how far solar radiation penetrates into the sea. Thus
models of mixed layer development and the occurrence of a diurnal
thermocline, which are sensitive to K, can benefit from being regularly
updated with information about how K is distributed and varies in space and
time.
Another factor which still needs more investigation is the relationship
between the near-surface measurements of C which satellites provide and the
distribution of C with depth. Because phytoplankton need light to thrive,
then it is reasonable to suppose that most will be detected by ocean colour