Presentation is loading. Please wait.

Presentation is loading. Please wait.

Rivers X-GIG phytobenthos intercalibration

Similar presentations


Presentation on theme: "Rivers X-GIG phytobenthos intercalibration"— Presentation transcript:

1 Rivers X-GIG phytobenthos intercalibration
Martyn Kelly

2 Phase 1 outcomes CB & N GIGs Several open issues
Partial intercalibration (diatoms only) for 12 MS (CB) and 4 MS (N) Several open issues Quality of reference screening Relevance of IC typology Effect of non-intercalibrated components and interaction with macrophytes Are phase 1 results consistent with phase 2 criteria?

3 Intercalibration Common Metric
ICM = average of EQR-IPS and EQR-TI Regression between national metric and ICM used to convert national boundary values to “common currency” e.g. Flanders national metric

4 Phase I achievements

5 Validation of phase 1 reference criteria
X-GIG exercise Developed a common database Reference screening by University of Vigo Taxonomic harmonisation exercise

6 NDMS of screened reference sites
Some separation between GIGs; effect of MS overrides effect of IC type

7 GIG / IC type differences in pICM

8 Threshold values for pICM
Mean value for GIG ± 2 standard deviations ALP CB EC MED N X-GIG IPS 15.2 14.4 10.7 13.3 16.3 13.9 IPS-adj 15.6 12.3 13.5 16.1 14.3 TI 2.33 3.55 4.60 3.03 2.23 3.22 TI-adj 3.48 4.69 3.05 2.06 3.18

9 Application to phase 1 reference data

10 Evaluation of “phase 1” results using “phase 2” criteria
MS H/G G/M AT Austria 0.11 0.03 BE-FL Belgium-Flanders 0.40 0.62 BE-WA Belgium-Wallonia 0.44 0.30 DE Germany 0.15 0.24 EE Estonia -0.15 ES Spain FR France -0.31 -0.01 IE Ireland 0.08 0.72 LU Luxembourg -0.26 -0.02 NL Netherlands 0.04 -0.03 SE Sweden -0.05 UK United Kingdom Median H/G and G/M recalculated excluding LU; “phase 2” calculations applied. MS who contribute to calculation of medians tend to have acceptable bias Other MS tend to be more stringent

11 New and adjusted methods
MS Phase Boundary as national EQR Bias as class width H/G G/M EE Estonia 1 0.85 0.70 -0.36 0.95 2 0.38 0.57 ES Spain 0.93 0.7 0.40 0.30 (Types 1 & 2) -0.29 0.10 (Types 3 & 4) 0.92 1.56 IT Italy 0.84 0.65 -0.26 -0.07 FR France -0.45 -0.02 0.94 0.78 0.02 0.19 LU Luxembourg -0.33 -0.06 0.90 0.00 -0.05 PL Poland (RC1, RC3, RC99) 0.80 0.58 0.03 Poland (RC4, RC5) 0.8 UK United Kingdom 0.42 1.00 0.75

12 Effect of initial assumptions
Phase 2 work started with the assumption that boundaries set by MS in good faith during phase 1 should be respected unless serious failings were revealed The tendency towards stringent boundaries for MS who did not contribute to the mean boundary has also been assessed …. … mean boundary value recalculated using all MS with reference sites (excluding IT – v. small dataset)

13 MS with acceptable (“OK”) or unacceptable (“not”) levels of bias
MS with ref sites MS without ref sites bias H/G OK not Not original calculations (phase 1) 6 2 3 new common boundary - mean 8 4 1 G/M original calculations (phase 1) 5 Re-calculated boundaries are slightly higher than phase 1 boundaries. Using these, there is a slight reduction in overall bias

14 N GIG Similar principles to CB GIG
Fewer MS (4 in phase 1; 5 in phase 2) Fewer problems with reference screening Different ICM: TISImin – minimum of TI and SI TISImean – average of TI and SI

15 Re-analysis of N GIG results
TISImin MS Phase Reference samples H/G G/M FI Finland 1 OK -0.04 -0.16 IE Ireland -0.45 SE Sweden 0.02 0.17 UK United Kingdom 0.01 0.08 2 TISImean MS Phase Reference samples H/G G/M FI Finland 1 OK -0.19 IE Ireland -0.41 -0.03 SE Sweden 0.18 1.75 UK United Kingdom 0.00 0.06 2 0.01 -0.07

16 Adding Norway to N GIG Norway has a metric that does not use diatoms.
Intercalibration involved collecting a parallel set of diatom samples, and using these to establish the position of the NO boundaries on the pICM scale.

17 Evaluation of NO boundaries
TISImin MS Step Boundary as national metric Bias, as class width H/G G/M NO Ca ≤ 1 0.99 0.83 0.22 0.06 Ca > 1 0.95 0.08 TISImean MS Step Boundary as national metric Bias, as class width H/G G/M NO Ca ≤ 1 0.99 0.83 0.38 0.10 Ca > 1 0.95 0.14

18 Summary CB-GIG N-GIG Phase 1 results are mostly sound
Several MS have voluntarily adjusted boundaries DK, CZ, LV, LI, SI and SK have not submitted methods CZ, IE, ES have new/modified methods close to completion N-GIG NO has been included in phase 2; UK has adjusted phase 1 boundaries All MS in N GIG have now submitted methods


Download ppt "Rivers X-GIG phytobenthos intercalibration"

Similar presentations


Ads by Google