Using Argo Data
Argo data is distributed by Global Data Access Centres (GDACs) in
netCDF files - usually one file per profile. Their format
is described in the Argo Users Manual found at
International Data Management site. See also the
Argo data beginners guide.
It is essential that attention is paid to the quality of the
data, both using in-file quality indicators and the processing mode
of the files themselves.
Argo filenames start with R or D, for
Realtime or Delayed-mode. R files are generated within 24 hours of
message transmission and the data has only automated corrections and
quality flagging. D files result from reprocessing the data no less than
6 months after collection, using more rigourous tests and
calibrations and involving data inspection by analysts under the
direction of a Principal Investigator. So, D files cannot be used
for near-realtime work, and R files should not be used for studies
involving quality-assured data.
Raw data and data after adjustments are
stored for each property. For example, PRES contains pressure readings
as sent by the float, and PRES_ADJUSTED data has had any known
corrections applied. Unless you have a very specific reason do not
use the raw data, even if no values are present in the _ADJUSTED variable.
Each property profile variable also has
a quality flag variable for the profile as a whole,
eg PROFILE_PRES_QC, and a variable with flags for each individual
measurement, eg PRES_QC. (See QC flag values).
It is essential that these variables are
consulted or you could use data known to be bad.
However, if the
QC flag is '4' or greater ("known bad data"), then the corresponding
_ADJUSTED value should have already been blanked.
Each adjusted property profile also has an uncertainty
variable, eg PRES_ADJUSTED_ERROR. This is often, but not always, a
default value for TEMP and PSAL, but for PRES commonly contains values
(eg 20 dbar) which imply that the data should NOT be used for some