|
|
|
In support to SMOS cal/val, the
objective is to provide an extended and validated in-situ dataset and
the corresponding gridded fields at the sea surface. The Coriolis
data centre mainly provide the in-situ observations. Argo
international array of profiling floats is the principal source of data,
reporting temperature and salinity from 2000m to near surface level
(4-5m). More isolated measurments from CTDs, sensor equipped marine
mammals and moorings (such as TAO/PIRATA in the tropics) complement
the dataset but provide little information on the surface levels.
Within the GLOSCAL context, we intent to incorporate data from
surface dedicated instruments that are not yet in the Coriolis
standard flow. First priority has been put on those collected with
Thermosalinographs.
Quality control of these data is a
prerequisite before they could be intergrated to the objective
analysis ISAS.
|
ISAS surface processing
As
ISAS has not been specifically developed for the surface (nor data
neither processing), a preliminary work was required to adapt it to
this need.
We
considered a surface layer extending from 0 to 20 meters. For profile
data, the surface layer is assumed perfectly mixed above the level of
the last measurement, which is then repeated up to the surface.
For TSG data, the level sampled by the instrument is assumed to
represent the whole surface layer and is repeated over this layer.
Most of the time the thermosalinographs measure temperature and
salinity at levels varying between 3m and 15m, depending on the
vessels and their draught. The error associated with extrapolated
data is increased proportionally to the distance from the measurement
level.
Furthermore,
ISAS has been adapted to include the thermosalinographs data because
of their specificity (timeserie data, data format, high frequency
sampled).
The ISAS tool
ISAS
(In Situ Analysis System) has been developed by Gaillard et al.
(2009b), as a tool to produce these gridded fields from objective
analysis of in-situ data coming from multiple sources. All the
documentation could be find here.
The ISAS grid covers the
globe from 80°S to 90°N with 1/2° Mercator resolution, on 151
standard depth levels between 0 and 2000m. A priori statistics are
needed for the analysis: they were obtained from a previous analysis
of the period 2002-2008 (von Schuckmann, et al. 2010), providing the
following reference fields:
- a mean seasonal cycle, or climatology, of temperature and salinity representative of the period
- the corresponding variances (deviation between profiles and monthly mean field)
- spatial scales deduced from the Rossby radius of the annual climatology
These
reference fields and statistics are also used by ISAS for
preliminary controls on the dataset and by the CATDS (Centre Aval de
Traitement et des Données SMOS) for the real time validation of SMOS
data.
We can consider that the ISAS processing
constitutes an additional control in the sense that it allows to
check with a single processing the consistency of simultaneous
datasets and the agreement with climatology.
In the context of SMOS cal/val, we provide monthly
gridded fields of SSS and SST and the corresponding dataset in
near real time, taking into account as many validated observations as
possible.
In-situ data and Quality Control
Coriolis Near Real Time dataset
The in-situ
observations used to build our dataset are mainly collected and
processed by the Coriolis data centre, which apply a real
time quality controls in two steps. First, a set of automatic tests
(location and date, spikes...) is applied to the database, followed
by a visual checking. The result is that quality flags ranging from 0
(no control) to 9 (missing value) are assigned to each
individual measurement. Then, after running daily objective analysis,
a diagnostic test detects outliers by screening the analysis
residuals. Anomalous
profiles are visually checked by an operator. The Coriolis Processing
is described in Coriolis report-04-047.
For
the needs of GLOSCAL, Coriolis data centre has implemented a
near-real time data flow. At the beginning of each month, a new
analysis is performed with the data of the previous month, thus
providing a dataset of improved quality.
Thermosalinograph
datasets
The TSG data are
collected within the context of GOSUD international project (Global
Ocean Surface Underway Data). The GOSUD dataset is hosted by Coriolis
that performs the project agreed real time quality controls on this
dataset. For our analysis, only data that have passed delayed mode
processing are used. The delayed mode quality control on TSG
data is performed by the scientists at LPO, IRD Centres and LEGOS
using the software TSGQC
(TSG-QC link) developed by US-IMAGO (IRD). There are two levels of
control: quality flags and corrections. The
quality flags used follows the Coriolis definition and depends on the
quality of the data with respect to the climatology, spikes, noise,
etc. Then when needed, the time series are adjusted to fit the
external data which can be either the water sample analysis (taken
aboard) and/or the Argo co-localized data with ship tracks (provided
by the LOCEAN). The TSG data set so produced are a contribution
to
the GOSUD project and are available in NetCDF Gosud format
The
processing flow can be summarized as follows:
| |
Further information on the LPO website. |
