There is an ever-growing demand for high-quality marine scientific data to contribute to global programmes. The need for harmonisation of those data is of paramount importance. An essential component of that harmonisation is the adoption of internationally agreed calibration standards and techniques for the instrumentation. This article explains the advantages and pitfalls in the calibrations of modern oceanographic instruments.
WOCE is a major component of the World Climate Research Programme (WCRP). It was established in 1979. It is a long-term global research study to develop ocean models useful for predicting long term, decadal, climatic changes. In order to provide the data required to test these models, oceanographers from many countries were involved in deploying instruments over the full ocean depth in all the world's oceans. It is estimated that some 10 years of ship time were carried out during the period 1990 - 1997 for the one-time hydrographic sections and a further 15 years of ship time for the repeated sections where the ocean variability is measured. Measurements were carried out on the physical parameters of conductivity, temperature and depth (CTD) as well as on chemical tracers, such as oxygen, nitrate, phosphate and CFC's.
Although the various components of WOCE combine to form the biggest and most comprehensive oceanographic experiment undertaken to date, programmes such as the Global Ocean Observing System (GOOS) promise an extensive dataset for coastal seas, the health of the ocean, living marine resources, climate and services. There are many aspects of GOOS currently under review by various international committees concerned with the politics, finance and practicality of implementation, data collection and handling but the fundamental issue of data quality will still depend upon the reliability of instrument calibration.
With the international effort required to make so many measurements, it is essential that expectations for data quality are consistent and realistic. With guidance from the Planning Committee, the specification produced by the WOCE Hydrographic Programme Office (WHPO) covers all the standards for water samples as well as the specifications for conductivity, temperature and pressure sensor calibration accuracy (Table 1).
Table 1 - The WOCE CTD specifications. PARAMETER ACCURACY PRECISION
Salinity 0.002 PSS-78 0.001 PSS-78
Temperature 0.002oC 0.0005oC
Pressure 3dbar 0.5dbar
Inter-comparability of data is of paramount importance to programmes such as WOCE and GOOS and the cost of acquiring data at sea is considerable. Therefore, it is vital that the calibration of probes is carried out to a satisfactory standard.
The conductivity-temperature-depth (CTD) instrument, extensively used in WOCE, is calibrated, like all calibrations, by comparing the device being measured to a standard or reference. As a general rule, the reference should perform significantly better (ideally one order of magnitude) than the instrument under calibration. The reference must also be traceable to national and international standards. In practice, the CTD instrument is calibrated against a transfer standard which fulfils the above criteria and which is itself calibrated against a primary standard. Here are some of the procedures for calibrating CTD's for WOCE.
Temperature
High-precision temperature calibrations use well established bridge techniques involving the inductively coupled AC current divider principle. The transfer standards presently in use at OSIL include an automatically balancing AC bridge system with platinum resistance thermometer (PRT) and external reference resistor. The resolution is equivalent to 0.00025oC with a quoted accuracy better than 0.001oC.
The transfer standards are themselves calibrated in-house, against primary standards. The WOCE requirement is to measure temperatures below 0oC and the International Temperature Scale 1990 (ITS90) stipulates that the PRT must be calibrated over the range covered by the triple points of mercury (-38.8344oC), water (0.0100oC) and the melting point of gallium (29.7646oC). In addition, two secondary standard triple point cells, phenoxybenzene (26.8625oC) and ethylene carbonate (36.3135oC), are used to provide extra reference points.
The transfer standard calibration procedure itself is fairly labour intensive, taking several days to complete for one PRT. For each PRT, bridge balance ratio readings are recorded at each standard temperature. Corrections are made for the hydrostatic head appropriate to the immersion depth of the PRT. Using a 'best fit' polynomial equation that relates the standard temperatures to the derived values of PRT resistance, a derivation of intermediate temperatures is obtained. In this way, the transfer standards are calibrated against and traceable to, the primary reference standards.
The sea-going sensors are calibrated against these transfer standards.
Uncertainties occur throughout any calibration process. Systematic errors arise from the primary standards, bridges, hydrostatic head, self-heating effects of the PRT and temperature effects on the reference resistor; all of which must be identified, quantified and corrections made wherever possible.
Conductivity
Two standard measurements are needed for conductivity calibration - temperature and salinity.
Temperature is measured using an automatically balancing AC bridge with external standard resistor and platinum resistance thermometer (PRT).
 |
Salinity is measured using the latest model of Guildline Autosal or Portasal salinometer. The manufacturer's quoted accuracy for the Guildline Autosal Model 8400B is 0.002 in salinity, although experience has shown that analysts can consistently obtain agreement of better than 0.001 in salinity. The salinometers, which are kept in optimum temperature controlled conditions, are standardised using IAPSO Standard Seawater at 35 salinity. The primary standard for salinity is the potassium chloride solution specifically defined by the Practical Salinity Scale 1978 (PSS-78). IAPSO Standard Seawater is the only approved transfer standard for salinity measurement. It is calibrated, only by the IAPSO Standard Seawater Service, against the primary standard potassium chloride solution to provide traceability. Linearity errors in the salinometers are corrected by the use of higher and lower salinity Standard Seawaters.
For the calibration of conductivity sensors, baths containing approximately 120 litres of filtered natural Atlantic seawater are used. The salinity of the baths is adjusted to provide a range appropriate for each particular instrument or its application. The contents of each bath are mixed by re-circulation and all calibrations are carried out in a temperature-controlled environment. The standard conductivity is calculated from the temperature and salinity measurements and compared to the CTD reading to obtain a best fit polynomial equation. |
Pressure
The pressure calibration accuracy as specified for WOCE is 3dbar over the full ocean range. This is approximately equivalent to 0.05% of 6000m, or three metres water depth.
The pressure calibration of sensors at OSIL is carried out using a dead weight tester (dwt) and a quartz pressure transfer standard. Both of these instruments are returned for re-calibration to a UKAS (United Kingdom Accreditation Service) accredited laboratory at regular intervals. Present certificated pressure differences between the dwt and the UKAS pressure vary from 0.03% of reading at pressures below 60 bar to 0.016% of reading at pressures of up to 600 bar. For the quartz standard, the pressure differences do not exceed 0.005% of full scale throughout the range and, the hysteresis 0.01% full scale.
When carrying out a high accuracy pressure calibration, the quartz gauge is placed in the high-pressure line from the dwt to the instrument to be calibrated. Pressure steps are generated by the dwt, for a sufficient length of time, to enable the quartz unit to make an absolute pressure measurement. This has the attraction that the errors associated with a dwt can be ignored if the environment remains reasonably stable.
The calibration for these parameters requires precise instruments, controlled laboratory conditions and highly-trained personnel. Of equal importance are the production, distribution and storage of all relevant documentation. This includes production of certificates and accurate record keeping for the standards. Ideally the calibration laboratory should operate a quality management system (e.g. ISO 9000) which formalises the documentation and procedures.
Traceable standards perform a key role in the production of consistent and comparable data. A good example of such a standard is the IAPSO Standard Seawater Service. This has been in operation for almost a century, providing one common source salinity standard throughout the world. Standard Seawater, as approved by the International Association for Physical Sciences of the Ocean (IAPSO), is the only transfer standard for Practical Salinity that is recognised by all the major oceanographic bodies. For over 90 years, the comparability of salinity data worldwide has depended on the widespread use of this standard for calibrating laboratory and in-situ instruments. IAPSO Standard Seawater is available in sealed glass ampoules, each containing about 275ml of natural seawater. The ampoule label carries information on its conductivity ratio (K15) and salinity according to the Practical Salinity Scale 1978 (PSS78).
However, other parameters, which will become increasingly important in global studies, are not so well defined. The measurement of the carbon dioxide system in the oceans has significant importance for the question of global warming. It is well understood that the fate of CO2 as it builds up in the atmosphere, lies within the oceans. Until recently, measurements of the CO2 system in the sea, were few and far between, not least because of the difficulties encountered during analysis. Some of those difficulties are now being overcome by the use of a common source standard seawater for carbon dioxide, currently being produced by Scripps Institution of Oceanography.
| The measurement of dissolved nutrients (nitrate, nitrite, silicate and phosphate) in studies of primary production is being improved by the more widespread use of seawater nutrient standards. Most nutrient determinations are carried out with automatic analysis systems, which are particularly susceptible to changes in refractive index between fresh and saline waters. Therefore it is important that standards are prepared in waters with similar salinity to that of the sample. To overcome problems of stability OSIL have prepared a Marine Nutrient Standards Kit (MNSK) which comprises concentrates of nutrient salts (nitrate, nitrite, silicate and phosphate) and Low Nutrient Seawater (LNS) which are mixed to produce fresh working standards for each analysis. |
 |
The calibration of instruments for chlorophyll determination or dissolved oxygen, is still inadequate, with methodology so far removed from the real world as to provide only an indication of instrument performance rather than meaningful calibration.
As we continue to collect ever-increasing volumes of data from more sensitive instruments, it is important to reflect on the quality of those data. Calibration is a fundamental prerequisite of data quality and should help to provide the comparability and relevance of scientific evidence from many sources as we strive to improve our understanding of the environment.
By Paul Ridout, OSIL.
As published in International Ocean Systems Design, v.2 No. 4, pp 8 - 10, (July/August 1998)