1. Method Validation of VINDTA 3C (TA/DIC) Marine Scotland Science (MSS) Experience and Concerns Pamela Walsham

2. Why Validate a Method ? Establish fitness for purpose Best Practice Legal requirements (MSFD – chemical monitoring to follow QA/QC directive) Data Comparability between Laboratories/Regions for regional and global data assessment Cheaper /better to prevent problems from happening rather than correct them afterwards. Validation helps the design and implementation of suitable quality control procedures. http://eurlex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2009:201:0036:00 38:EN:PDF

3. MSS Validation Plan Repeatability (Within Day Precision) –Seven replicates of Certified reference material (CRM) or laboratory reference material (LRM) or spiked sample analysed on same day. –NIST traceable certified reference material not available for TA/DIC used Scripps Reference Material (RM) and LRM Reproducibility (Between Day Precision) –Seven replicates of CRM or LRM or spiked sample run on different days, used RM and LRM –Control chart monitoring Bias/Recovery – Seven replicates of appropriate CRM or External Quality Assurance material (EQA) or spiked sample analysed on different days.

4. MSS Validation Plan Considerations Considered Sources Uncertainty  The error associated with each step in the procedure  Sampling  Storage (LRM)  Equipment Calibration (coulometer cell)  Changes to equipment (different cells)  Reagent Purity (accounted for in validation)  Environmental Effects  Operator Experience  Data

5. Required Accuracy Accuracy for monitoring based ICES CRR report No 319 Oct 2013 ParameterSalinity (g/kg) Temperature ( o C) Phosphate (µmol/kg) Silicate (µmol/kg) Nitrate (µmol/kg) Accuracy0.01 0.3151 ParameterTA (µmol/kg) DIC (µmol/kg) pHpCO 2 (µatm) Accuracy±3 closed cell ±1 open cell ±2±0.001 pH units (spectrophotometrically) 0.003 pH units glass electrode 1-2

6. Repeatability (within day) of Reference Materials SCRIPPS RMTADIC RM concentration (µmol/kg) 2246.742021.15 Mean concentration (µmol/kg, n = 8) 2246.702018.03 Stdev0.761.78 %RSD0.030.09 Absolute Error (µmol/kg)0.043.12 Relative Error (%)0.0020.154

7. Repeatability (within day) of Laboratory Reference Material LRMTADIC Mean concentration (µmol/kg, n = 5) 2275.022066.71 Stdev1.662.06 %RSD0.071.00

8. Reproducibility (Between day) of Reference Material SCRIPPS RMTADIC RM concentration (µmol/kg) 2246.742021.15 Mean concentration (µmol/kg, n = 9) 2247.882019.98 Stdev1.722.52 %RSD0.080.12 Absolute Error (µmol/kg)1.141.17 Relative Error (%)0.0510.057

9. Quantifying Combined Uncertainty Range of calculations available for calculating method uncertainty MSS Method - agreed with UK Accreditation Service Systematic component (uncertainty on the recovery on Reference Material i.e Recovery RSD/n 0.5 ) = C s % Random Component (Control Chart RSD) = C r % Assume linear summation: Combined standard uncertainty = (C s 2 + C r 2 ) 0.5 = X Expanded uncertainty = 2* (C s 2 + C r 2 ) 0.5 = X%

10. Sampling Precision –24 replicate samples collected in consecutive fashion on board MRV Scotia –Analysed in random order over several days –F-test to compare precision from sample replicates to precision from repeatability/reproducibility  Would determine if sampling was limiting factor for overall uncertainty of measurements.

11. Sampling Precision - Results TADIC Mean concentration (µmol/kg, n = 24) 2284.522078.38 Stdev1.264.37 %RSD0.060.21 TA precision similar to repeatability and reproducibility precision DIC precision < repeatability or reproducibility precision A simple F-test indicated DIC precision was reduced as a consequence of sampling

12. Ongoing method performance- control charts –Operator error, reference electrode solutions low

13. MSS concerns with TA method validation and routine analysis Impact of bottle on nutrient concentration Silicate concentration increases over time Should samples be analysed for nutrients from sample bottle? What about concentration in RMs?

14. MSS concerns with DIC method validation and routine analysis Major Concern is Coulometer Calibration Suggested calibration - Sodium carbonate calibration Time consuming, would be required daily No time for analysis! Typically labs are correcting to RM? Is there an alternative approach?

15. MSS concerns with analysis for routine monitoring purposes No regular external Quality Assurance scheme RM available at single salinity Need extensive QC – time consuming. 1 x sub-standard (to condition DIC cell, data not recorded) 1 x Blank 3-5 x LRM (to monitor daily precision) 1 x RM (used as calibrant) 7-8 x Samples (with a blank run in random position) 1 x RM (to monitor instrumental drift)

16. Discussion Time