back to the data originator, and the data set
improved by iteration. Eventually, data-quality
flags would be added to the data set to identify
suspect or bad data.
By mid-way through the WHP, it was apparent
that there were not enough scientists willing to act
as DQEs to enable all the data to be examined in
detail. Accordingly, the WHP Planning Committee
decided that DQE effort should be concentrated
on the one-time data. It is expected that all onetime bottle data will have been through the DQE
process and all one-time CTD data will have been
checked at the DAC by the time the final WOCE
data archive is written to CD-ROM in 2003.
Where the DQE process has taken place, it is
not clear at the time of writing what the long-term
benefits will be. What becomes of the DQE comments and suggestions? They can be archived, and
some changes are made after the DQE process,
although the number is actually rather small: the
data reported by PIs are generally of a very high
quality. Because of the time lag, changes made to
files held at the WHP DAC and subsequently
archived may well diverge from versions held by
originating PIs or in a national archive. Original
versions may have been published by a PI and be
in widespread circulation without the DQE modifications. Another drawback of the DQE delay is
that many scientists have found it impossible to
respond adequately to DQE questions. Critical
personnel may have moved on to other projects
and be unavailable to sort out queries. It has been
suggested that the mere concept of the DQE
process made it less necessary: knowing that their
data submissions would be scrutinized by colleagues, PIs may have taken extra care to submit
carefully checked data sets.
For repeat hydrography, the formal DQE process has been abandoned; the majority of repeat
stations will be archived as reported by the investigators. Some of the repeated sections were carried
out with the same personnel, facilities, and standards as the one-time surveys, and are likely to be
of correspondingly high quality. Other sections
may be of lesser quality for a variety of reasons.
Whatever the origin of the measurements, final
quality assessment of the repeat hydrography is
left to the data users.
One of the effective data-quality checks emerging
during the AIMS phase is the various atlas exercises: small teams have been funded to produce
basin-wide atlases of WOCE data, with the Pacific
and Southern Oceans the most advanced at present. The requirement to examine every section of
data in the course of atlas production has already
highlighted some errors or inconsistencies in the
data sets, with the information being passed back
to the WHP DAC for investigation.
Documentation, data reporting and formats
The level of documentation stored with the WHP
data is not as great as originally conceived. Prior
to WOCE, many laboratories would routinely
produce printed data reports for all major cruises.
These would include detailed accounts by data
processors and analysts of procedures used, calibrations applied and problems encountered. The
original intention was to include this information
in cruise reports, made available alongside the
data in electronic form. However, the wide range
of word processing packages in which PIs prepare
the documentation was incompatible with the
need to make the information available in a universally readable format. The task of converting
the PI’s version was beyond available personnel
resources. A compromise was therefore reached in
which the WHP Office produced minimal summaries of cruise documentation: ship, dates, personnel, PIs and so on. The WHP Office store any
further documentation supplied by the PIs, but the
level of detail is variable. Wherever possible, the
WHPO has obtained supplementary cruise and
data documentation and made it available as PDF
files, including scanning in paper copies of older
material if necessary.
Furthermore, the ease of access to electronic
forms of the data means that many laboratories
now no longer produce printed data reports. As a
result, the link to the original analyst’s records is
now rather weaker, with possible consequences for
the accuracy of data reporting. Details of blank
corrections may be lost, as may formulae used for
corrections and conversions. In the requirements
for data reporting, the WHP was careful to specify
names and units of variables to be reported. This
proved to be a mixed blessing, because it gave rise
to the possibility that data will be reported as having been converted to the approved units when in
fact no conversion has taken place: concentration
per unit volume (as commonly measured by analysts) instead of per unit mass (as requested by
WHP reporting protocols) is probably the most
SECTION 3 NEW WAYS OF OBSERVING THE OCEAN
108
improved by iteration. Eventually, data-quality
flags would be added to the data set to identify
suspect or bad data.
By mid-way through the WHP, it was apparent
that there were not enough scientists willing to act
as DQEs to enable all the data to be examined in
detail. Accordingly, the WHP Planning Committee
decided that DQE effort should be concentrated
on the one-time data. It is expected that all onetime bottle data will have been through the DQE
process and all one-time CTD data will have been
checked at the DAC by the time the final WOCE
data archive is written to CD-ROM in 2003.
Where the DQE process has taken place, it is
not clear at the time of writing what the long-term
benefits will be. What becomes of the DQE comments and suggestions? They can be archived, and
some changes are made after the DQE process,
although the number is actually rather small: the
data reported by PIs are generally of a very high
quality. Because of the time lag, changes made to
files held at the WHP DAC and subsequently
archived may well diverge from versions held by
originating PIs or in a national archive. Original
versions may have been published by a PI and be
in widespread circulation without the DQE modifications. Another drawback of the DQE delay is
that many scientists have found it impossible to
respond adequately to DQE questions. Critical
personnel may have moved on to other projects
and be unavailable to sort out queries. It has been
suggested that the mere concept of the DQE
process made it less necessary: knowing that their
data submissions would be scrutinized by colleagues, PIs may have taken extra care to submit
carefully checked data sets.
For repeat hydrography, the formal DQE process has been abandoned; the majority of repeat
stations will be archived as reported by the investigators. Some of the repeated sections were carried
out with the same personnel, facilities, and standards as the one-time surveys, and are likely to be
of correspondingly high quality. Other sections
may be of lesser quality for a variety of reasons.
Whatever the origin of the measurements, final
quality assessment of the repeat hydrography is
left to the data users.
One of the effective data-quality checks emerging
during the AIMS phase is the various atlas exercises: small teams have been funded to produce
basin-wide atlases of WOCE data, with the Pacific
and Southern Oceans the most advanced at present. The requirement to examine every section of
data in the course of atlas production has already
highlighted some errors or inconsistencies in the
data sets, with the information being passed back
to the WHP DAC for investigation.
Documentation, data reporting and formats
The level of documentation stored with the WHP
data is not as great as originally conceived. Prior
to WOCE, many laboratories would routinely
produce printed data reports for all major cruises.
These would include detailed accounts by data
processors and analysts of procedures used, calibrations applied and problems encountered. The
original intention was to include this information
in cruise reports, made available alongside the
data in electronic form. However, the wide range
of word processing packages in which PIs prepare
the documentation was incompatible with the
need to make the information available in a universally readable format. The task of converting
the PI’s version was beyond available personnel
resources. A compromise was therefore reached in
which the WHP Office produced minimal summaries of cruise documentation: ship, dates, personnel, PIs and so on. The WHP Office store any
further documentation supplied by the PIs, but the
level of detail is variable. Wherever possible, the
WHPO has obtained supplementary cruise and
data documentation and made it available as PDF
files, including scanning in paper copies of older
material if necessary.
Furthermore, the ease of access to electronic
forms of the data means that many laboratories
now no longer produce printed data reports. As a
result, the link to the original analyst’s records is
now rather weaker, with possible consequences for
the accuracy of data reporting. Details of blank
corrections may be lost, as may formulae used for
corrections and conversions. In the requirements
for data reporting, the WHP was careful to specify
names and units of variables to be reported. This
proved to be a mixed blessing, because it gave rise
to the possibility that data will be reported as having been converted to the approved units when in
fact no conversion has taken place: concentration
per unit volume (as commonly measured by analysts) instead of per unit mass (as requested by
WHP reporting protocols) is probably the most
SECTION 3 NEW WAYS OF OBSERVING THE OCEAN
108