184
IAN ROBINSON
facilitate the assimilation of SST products into operational ocean models and
forecasting systems.
Medspiration will produce four types of data product. These are
elaborated as follows:
Level 2 pre-processed SST data: ESA’s Medspiration project will
generate L2P data, as defined in Section 4.2.3, within a few hours of the
input L2 SST products being served by their respective agencies. The area
coverage is the whole Atlantic Ocean and adjacent seas. In the User
Requirements Document (URD) produced by GHRSST-PP the EurDAC
area is defined as encompassing the area 70ºS to 90ºN (to the ice limit) and
from 100ºW to 45ºE (to include the adjacent seas but excluding any part of
the Pacific Ocean).
Level 4 gridded, interpolated SST data product: An L4 product will
be produced for the Mediterranean Sea, consisting at each grid cell of a
single daily value of SSTfnd produced by the optimal interpolation process,
plus an estimate of the expected diurnal stratification and skin effects every
two hours. Error estimates will also be provided for every temperature
estimate. The ultra-high resolution (UHR) grid for this product is 2 × 2 km.
Delivery is due by noon on the day following that to which it refers.
The software for generating the L4 product is designed with sufficient
flexibility to allow the parameters in the OI scheme to be changed. Similarly
the form of the model defining the differences between skin, subskin and
foundation SST can be adjusted in future following operational experience
and analysis of the resulting L4 products. The main purpose of the L4
product in GHRSST is to meet the demand from general users of SST for a
product which best combines all of the available primary sources of SST
from different sensors.
A match-up data base (MDB) will be produced, which assembles
matched pairs of all available in situ measurements of SST that are
coincident with satellite measurements from the ingested L2 SST data within
the Medspiration domain. It is most convenient to create the MDB in nearreal time when the L2P data have just been produced. Medspiration will
make use of the French Coriolis Data Centre used to assemble in situ data
although, the difficulty of timely delivery of some in situ SST samples is
likely to delay the MDB creation by up to a week after real time.
The MDB will be analysed regularly to allow GHRSST to update
frequently (say once per week or every few weeks) the sensor specific error
statistics (SSES) attached to the L2P product, needed for assimilation and
also in the OI procedure. It is also important that a broad spatial coverage is
achieved by the MDB so that the SSES are properly spatially representative
and can also be stratified by confidence value.
A diagnostic data set (DDS) of SST observations will be compiled for
the Medspiration L2P product area by extracting 1/100º × 1/100º gridded sub
scenes for a set of predefined small areas (typically 2º × 2º in extent) from
IAN ROBINSON
facilitate the assimilation of SST products into operational ocean models and
forecasting systems.
Medspiration will produce four types of data product. These are
elaborated as follows:
Level 2 pre-processed SST data: ESA’s Medspiration project will
generate L2P data, as defined in Section 4.2.3, within a few hours of the
input L2 SST products being served by their respective agencies. The area
coverage is the whole Atlantic Ocean and adjacent seas. In the User
Requirements Document (URD) produced by GHRSST-PP the EurDAC
area is defined as encompassing the area 70ºS to 90ºN (to the ice limit) and
from 100ºW to 45ºE (to include the adjacent seas but excluding any part of
the Pacific Ocean).
Level 4 gridded, interpolated SST data product: An L4 product will
be produced for the Mediterranean Sea, consisting at each grid cell of a
single daily value of SSTfnd produced by the optimal interpolation process,
plus an estimate of the expected diurnal stratification and skin effects every
two hours. Error estimates will also be provided for every temperature
estimate. The ultra-high resolution (UHR) grid for this product is 2 × 2 km.
Delivery is due by noon on the day following that to which it refers.
The software for generating the L4 product is designed with sufficient
flexibility to allow the parameters in the OI scheme to be changed. Similarly
the form of the model defining the differences between skin, subskin and
foundation SST can be adjusted in future following operational experience
and analysis of the resulting L4 products. The main purpose of the L4
product in GHRSST is to meet the demand from general users of SST for a
product which best combines all of the available primary sources of SST
from different sensors.
A match-up data base (MDB) will be produced, which assembles
matched pairs of all available in situ measurements of SST that are
coincident with satellite measurements from the ingested L2 SST data within
the Medspiration domain. It is most convenient to create the MDB in nearreal time when the L2P data have just been produced. Medspiration will
make use of the French Coriolis Data Centre used to assemble in situ data
although, the difficulty of timely delivery of some in situ SST samples is
likely to delay the MDB creation by up to a week after real time.
The MDB will be analysed regularly to allow GHRSST to update
frequently (say once per week or every few weeks) the sensor specific error
statistics (SSES) attached to the L2P product, needed for assimilation and
also in the OI procedure. It is also important that a broad spatial coverage is
achieved by the MDB so that the SSES are properly spatially representative
and can also be stratified by confidence value.
A diagnostic data set (DDS) of SST observations will be compiled for
the Medspiration L2P product area by extracting 1/100º × 1/100º gridded sub
scenes for a set of predefined small areas (typically 2º × 2º in extent) from