Abstract
The objective of this inter-laboratory calibration exercise is to evaluate and improve comparability of global nutrients data in the world ocean. IOCCP and JAMSTEC co-organized an inter-laboratory calibration exercise of nutrients in seawater using four lots of recently certified RM produced by KANSO and three CRMs provided by National Metrology Institute of Japan which are certified in Marine 2014. 71 laboratories in 28 countries had replied to the call for participants. Results were returned from 59 laboratories. Korean Institute of Ocean Science and Technology, KIOST, also offered to provide their recently developed RMs to this I/C exercise. The Royal Netherlands Institute for Sea Research, NIOZ, also offers to provide silicate stock solution to contribute to the overall assessment of results regarding with this I/C exercise.
Mean, median and standard deviation were calculated, robust mean and standard deviation were also calculated. Successive t-tests at the 95% confidence level were applied to the results before estimating the consensus mean, consensus median, and consensus standard deviation. Z-scores were also calculated to evaluate the performance of laboratories as in the previous inter-comparison studies.
The ranked concentration plots for a particular nutrient would be proportional and roughly parallel to each other for samples with different nutrient concentrations if each laboratory appropriately compensated for the non-linearity of the calibration curves. However there are non-proportional results from some laboratories for all of the determinants as well as observed in the previous I/C studies. These results indicate that non-linearity of the calibration curves for nutrient analysis is one of significant sources of less comparability of nutrients data. This implies that we need to use a set of CRM of which nutrients concentrations can cover whole range of measurements of nutrients concentration to keep comparability of whole range of nutrients concentration in the world ocean. It is clear that present comparability among the participants in 2015 I/C exercise is quite similar with previously obtained comparability in 2012 I/C study and previous I/C studies. Consensus standard deviations of all determinands are one order of magnitude large rather than homogeneity of the CRMs distributed and consensus standard deviations are about double of reported precision of measurements of the laboratories. Therefore these I/C results show that use of CRM will be able to greatly improve comparability of nutrient data among the laboratories of the world.
There are good signal in the results that although consensus standard deviations are relatively large, consensus median/mean of each samples showed good agreement with certified values of the samples within consensus SDs. This implies that majority of the participating laboratories have good capability to measure nutrients concentration in seawater and using CRM will increase more on the comparability and could be their results to be SI traceable quickly.
Mean, median and standard deviation were calculated, robust mean and standard deviation were also calculated. Successive t-tests at the 95% confidence level were applied to the results before estimating the consensus mean, consensus median, and consensus standard deviation. Z-scores were also calculated to evaluate the performance of laboratories as in the previous inter-comparison studies.
The ranked concentration plots for a particular nutrient would be proportional and roughly parallel to each other for samples with different nutrient concentrations if each laboratory appropriately compensated for the non-linearity of the calibration curves. However there are non-proportional results from some laboratories for all of the determinants as well as observed in the previous I/C studies. These results indicate that non-linearity of the calibration curves for nutrient analysis is one of significant sources of less comparability of nutrients data. This implies that we need to use a set of CRM of which nutrients concentrations can cover whole range of measurements of nutrients concentration to keep comparability of whole range of nutrients concentration in the world ocean. It is clear that present comparability among the participants in 2015 I/C exercise is quite similar with previously obtained comparability in 2012 I/C study and previous I/C studies. Consensus standard deviations of all determinands are one order of magnitude large rather than homogeneity of the CRMs distributed and consensus standard deviations are about double of reported precision of measurements of the laboratories. Therefore these I/C results show that use of CRM will be able to greatly improve comparability of nutrient data among the laboratories of the world.
There are good signal in the results that although consensus standard deviations are relatively large, consensus median/mean of each samples showed good agreement with certified values of the samples within consensus SDs. This implies that majority of the participating laboratories have good capability to measure nutrients concentration in seawater and using CRM will increase more on the comparability and could be their results to be SI traceable quickly.
| Original language | English |
|---|---|
| Place of Publication | Yokosuka |
| Publisher | Japan Agency for Marine-Earth Science and Technology |
| Commissioning body | Japan Agency for Marine-Earth Science and Technology |
| ISBN (Electronic) | 978-4-901833-23-3 |
| Publication status | Published - Jun 2016 |
