SATELLITE MEASUREMENTS
185
every L2P and L4 product produced by the Medspiration processor. Each
product will be resampled (nearest neighbour) onto a common grid to
facilitate analysis. Where possible the DDS sites have been chosen to
coincide with known locations of regular in situ measurements (e.g.
PIRATA buoys, Ferry-Box routes).
Within Medspiration the DDS will allow close comparison to be made
between the different input streams of SST and with the L4 output. One
objective is to provide a prompt alert should a particular sensor start to
deviate seriously from the others. Another is to facilitate the research
needed to test and improve the data merging methods, diurnal variability
parameterisations and OI configuration used both in Medspiration and more
widely in GHRSST-PP.
4.4
Lessons to be learned from the GHRSST-PP and
Medspiration approach
Although still in the development stage and not scheduled to commence
a demonstration of operational capability before early 2005, GHRSST-PP
and its European contributor Medspiration have been singled out for this
case study because they provide a pathfinding example of what can be
achieved when data providers and the remote sensing scientists working with
them turn their attention to the needs of users, and in this case the
requirements of the ocean modelling community for SST data to be
assimilated into ocean forecasting models. It is worth noting that, although
GHRSST-PP has been driven largely by the science team of satellite
oceanographers and data providers, it was spawned in the first place by the
modellers in GODAE and has at all stages interacted strongly with its
intended users in the ocean modelling community, who have attended the
science team meetings and endorsed the implementation plan and the data
processing model. This was recognised by ESA when they decided to
initiate the Medspiration Project within their DUE Programme. The
Medspiration project is thus not so much a science development project as
primarily a software contract to develop a system to meet the needs
identified in the Medspiration User Requirements Document.
Stepping back from the details which are specific to SST measurements,
there are more general lessons to be learned about the preparation of satellite
data in order to make it more useful to the modelling community, as follows.
1. Attempts should be made to widen the spatial and temporal coverage
of remote sensing by using data from different sources, especially
when they offer complementarity of sampling. Ideally use should be
made of the products being produced by agencies already, without
changing them apart from adding error statistics and confidence
flags. This implies an active partnership with the data supplying
agencies.
185
every L2P and L4 product produced by the Medspiration processor. Each
product will be resampled (nearest neighbour) onto a common grid to
facilitate analysis. Where possible the DDS sites have been chosen to
coincide with known locations of regular in situ measurements (e.g.
PIRATA buoys, Ferry-Box routes).
Within Medspiration the DDS will allow close comparison to be made
between the different input streams of SST and with the L4 output. One
objective is to provide a prompt alert should a particular sensor start to
deviate seriously from the others. Another is to facilitate the research
needed to test and improve the data merging methods, diurnal variability
parameterisations and OI configuration used both in Medspiration and more
widely in GHRSST-PP.
4.4
Lessons to be learned from the GHRSST-PP and
Medspiration approach
Although still in the development stage and not scheduled to commence
a demonstration of operational capability before early 2005, GHRSST-PP
and its European contributor Medspiration have been singled out for this
case study because they provide a pathfinding example of what can be
achieved when data providers and the remote sensing scientists working with
them turn their attention to the needs of users, and in this case the
requirements of the ocean modelling community for SST data to be
assimilated into ocean forecasting models. It is worth noting that, although
GHRSST-PP has been driven largely by the science team of satellite
oceanographers and data providers, it was spawned in the first place by the
modellers in GODAE and has at all stages interacted strongly with its
intended users in the ocean modelling community, who have attended the
science team meetings and endorsed the implementation plan and the data
processing model. This was recognised by ESA when they decided to
initiate the Medspiration Project within their DUE Programme. The
Medspiration project is thus not so much a science development project as
primarily a software contract to develop a system to meet the needs
identified in the Medspiration User Requirements Document.
Stepping back from the details which are specific to SST measurements,
there are more general lessons to be learned about the preparation of satellite
data in order to make it more useful to the modelling community, as follows.
1. Attempts should be made to widen the spatial and temporal coverage
of remote sensing by using data from different sources, especially
when they offer complementarity of sampling. Ideally use should be
made of the products being produced by agencies already, without
changing them apart from adding error statistics and confidence
flags. This implies an active partnership with the data supplying
agencies.