common error. This is a hazard of specifying the
units to be used in the data format instead of
requiring/allowing the data originator to supply
the units.
The lack of a community standard CTD/hydrographic/tracer data exchange format prior to
WOCE was recognized as a problem. WHP planners attempted to rectify this with the WHP data
formats described in WHPO (1994b) for CTD and
bottle data. The intention was to specify a format
that included all the relevant WOCE parameters
and supporting information thought desirable, and
which an inexperienced user could write and read
reasonably easily. Unfortunately, the WHP format
definitions were not sufficiently rigorous, and so
many users had trouble writing and reading files in
a consistent manner. Many files submitted by PIs
contain small or large discrepancies that make
them hard to read, although some of them are
technically legal within the specification. This
was a liability throughout the WOCE observational phase, hindering exchange of data. When
PIs started submitting large numbers of data files,
the WHP DAC was unable to keep up with the
task of fixing the format problems in the data files.
Only during the WOCE AIMS phase, when the
rate of new data submissions slowed, has the DAC
been able to address the task of converting the
data files to a more consistent interpretation of
the formats. The increasing use of NetCDF may
provide a format that can be widely read with a
minimum of user effort.
So what recommendation can be made for the
future? In order for a format to be easily and reliably read, it must be strictly defined and adhered
to when the data files are written. This is difficult
for inexperienced data originators. If the burden is
too great they may be unwilling or even unable to
cooperate. Indeed, it is difficult to imagine a community of hundreds of PIs all managing to do this
flawlessly. Conversely, data will only be widely
used if they are available in a common format.
Data centres will have to continue to accept data
in somewhat flexible formats, working with data
originators to resolve ambiguities, and making the
data available back to the community in a single
format that each user need only learn to read once.
Data centres need to be provided with resources
on the assumption that every file provided to them
will need to be read in, debugged and written out
before the data can be made available to users.
Accuracy of data in the WHP
There is no single conclusive method for determining the accuracy of the WHP water property measurements. Instead, the overall quality evaluation
relies heavily on a consistency check. The WHP
sampling grid (Fig. 3.1.4, see Plate 3.1.4, p. 172)
includes many cross-overs, where nearby measurements were made at different times. Above the
main thermocline, the differences in water properties at these cross-overs are expected to be large;
the upper ocean is highly variable due to the
annual cycle, eddies, and other short-term phenomena. In and below the thermocline, however,
temporal variability is expected to be much
weaker, and to be manifest primarily as vertical
displacements of the stratification. As a function
of density, deep water properties are expected to
have been approximately constant during the
WHP; indeed, this is a fundamental assumption in
the design of the one-time survey. Comparison of
water properties at cross-overs as a function of
density is therefore a check of this assumption – the
ability of a one-time survey to represent a mean circulation – and also a check of measurement accuracy. The results presented need to be read with
some caution since there is not universal agreement
on how such comparisons should be performed or
interpreted. But they remain a useful guide to the
quality and consistency of the measurements.
Because density is a function of temperature,
salinity and pressure, these three quantities cannot
be compared independently on density surfaces.
Temperature and pressure measurements are relatively direct, stable and well calibrated. Their
uncertainties, typically 0.002°C for temperature
and 3 dbar for pressure, are believed to be fairly
well known a priori. Attention is therefore focused
primarily on salinity, which is measured indirectly
via electrical conductivity, temperature and pressure. In-situ CTD salinity measurements are calibrated by comparison with shipboard laboratory
measurements on water samples from the hydrographic cast, which are in turn compared with
measurements on standard seawater samples that
are prepared in large batches.
Accuracy of the WHP one-time salinity measurements
In the Pacific Basin, WHP cross-overs were
examined for salinity differences by Aoyama et al.
(2000). Reported bottle salinities having potential
temperatures between 1.0°C and 1.6°C were used
3.1 Shipboard Observations during WOCE
109
King, Firing and Joyce
units to be used in the data format instead of
requiring/allowing the data originator to supply
the units.
The lack of a community standard CTD/hydrographic/tracer data exchange format prior to
WOCE was recognized as a problem. WHP planners attempted to rectify this with the WHP data
formats described in WHPO (1994b) for CTD and
bottle data. The intention was to specify a format
that included all the relevant WOCE parameters
and supporting information thought desirable, and
which an inexperienced user could write and read
reasonably easily. Unfortunately, the WHP format
definitions were not sufficiently rigorous, and so
many users had trouble writing and reading files in
a consistent manner. Many files submitted by PIs
contain small or large discrepancies that make
them hard to read, although some of them are
technically legal within the specification. This
was a liability throughout the WOCE observational phase, hindering exchange of data. When
PIs started submitting large numbers of data files,
the WHP DAC was unable to keep up with the
task of fixing the format problems in the data files.
Only during the WOCE AIMS phase, when the
rate of new data submissions slowed, has the DAC
been able to address the task of converting the
data files to a more consistent interpretation of
the formats. The increasing use of NetCDF may
provide a format that can be widely read with a
minimum of user effort.
So what recommendation can be made for the
future? In order for a format to be easily and reliably read, it must be strictly defined and adhered
to when the data files are written. This is difficult
for inexperienced data originators. If the burden is
too great they may be unwilling or even unable to
cooperate. Indeed, it is difficult to imagine a community of hundreds of PIs all managing to do this
flawlessly. Conversely, data will only be widely
used if they are available in a common format.
Data centres will have to continue to accept data
in somewhat flexible formats, working with data
originators to resolve ambiguities, and making the
data available back to the community in a single
format that each user need only learn to read once.
Data centres need to be provided with resources
on the assumption that every file provided to them
will need to be read in, debugged and written out
before the data can be made available to users.
Accuracy of data in the WHP
There is no single conclusive method for determining the accuracy of the WHP water property measurements. Instead, the overall quality evaluation
relies heavily on a consistency check. The WHP
sampling grid (Fig. 3.1.4, see Plate 3.1.4, p. 172)
includes many cross-overs, where nearby measurements were made at different times. Above the
main thermocline, the differences in water properties at these cross-overs are expected to be large;
the upper ocean is highly variable due to the
annual cycle, eddies, and other short-term phenomena. In and below the thermocline, however,
temporal variability is expected to be much
weaker, and to be manifest primarily as vertical
displacements of the stratification. As a function
of density, deep water properties are expected to
have been approximately constant during the
WHP; indeed, this is a fundamental assumption in
the design of the one-time survey. Comparison of
water properties at cross-overs as a function of
density is therefore a check of this assumption – the
ability of a one-time survey to represent a mean circulation – and also a check of measurement accuracy. The results presented need to be read with
some caution since there is not universal agreement
on how such comparisons should be performed or
interpreted. But they remain a useful guide to the
quality and consistency of the measurements.
Because density is a function of temperature,
salinity and pressure, these three quantities cannot
be compared independently on density surfaces.
Temperature and pressure measurements are relatively direct, stable and well calibrated. Their
uncertainties, typically 0.002°C for temperature
and 3 dbar for pressure, are believed to be fairly
well known a priori. Attention is therefore focused
primarily on salinity, which is measured indirectly
via electrical conductivity, temperature and pressure. In-situ CTD salinity measurements are calibrated by comparison with shipboard laboratory
measurements on water samples from the hydrographic cast, which are in turn compared with
measurements on standard seawater samples that
are prepared in large batches.
Accuracy of the WHP one-time salinity measurements
In the Pacific Basin, WHP cross-overs were
examined for salinity differences by Aoyama et al.
(2000). Reported bottle salinities having potential
temperatures between 1.0°C and 1.6°C were used
3.1 Shipboard Observations during WOCE
109
King, Firing and Joyce