Presentation is loading. Please wait.

Presentation is loading. Please wait.

EMMA Workshop April, Copenhagen

Similar presentations


Presentation on theme: "EMMA Workshop April, Copenhagen"— Presentation transcript:

1 EMMA Workshop 17-18 April, Copenhagen
WS3/9: Questions 1-3 Questions to WHS6 Comments 1. Is the selection of determinands and species adequate for the purpose of this indicator? Determinands: Cd, Pb, Hg, DDT(=DDE+DDD), Lindane, PCB7 (CB28, CB52, CB101, CB118, CB138, CB153 and CB180) or adjusted subsets Species: mussels, cod, herring and flounder 2. Are participants aware of their respective data-submission procedures (protocols, quidelines etc.) and who the data coordinators are? Cf. Data Dictionary on EEA Website + annual data request to country NFPs 3. Is there adequate spatial/temporal representation? Cf. slide 6 and 7 from Bjerkeng’s presentation EMMA Workshop April, Copenhagen

2 Overview of Hazardous substance data for biota from EIONET and ICES
Three countries have delivered partly through ICES, but also directly in EIONET data flow: Partly same contaminants Possible that data series are split between ICES and EIONET data, or occur in duplicate May also be different geographical locations Review of combined ICES+EIONET data set together with country NFP/NRC required to resolve this issue EMMA Workshop April, Copenhagen

3 Geographical coverage of Hazardous substances in marine Biota
EMMA Workshop April, Copenhagen

4 EMMA Workshop 17-18 April, Copenhagen
WS3/9: Questions 4,5 and 7 Questions to WHS6 Comments 4. Are there obvious gaps in the data? 5. If so, why has this data not been included (e.g. because of bottlenecks, funding, protocols, qualified personnel, other priorities etc.)? 7. How might the EEA and respective Marine conventions enhance data submissions? A possible solution would be to develop procedures in the EIONET/WISE data flows for reviewing complete datasets and aggregation/assessment results in communication with countries instead of merely adding incremental data submissions from countries. Build procedures to handle differences in technical details, focus more on completeness and consistency in data, simplify data dictionary. EMMA Workshop April, Copenhagen

5 EMMA Workshop 17-18 April, Copenhagen
WS3/9: Questions 8 and 9 Questions to WHS6 Comments 8. Do the figures and maps in WHS6 provide an adequate overview of concentrations of the selected hazardous substances? Cf. figures and tables and maps in doc. 9 9. Are the statistical procedures adequate? Changes in Regional (aggregated) concentrations may not correspond well with the station-by-station studies and show how aggregated values can be misleading. Also, stations in close proximity may have conflicting results. Classification: Average yearly median concentrations over latest 8-year period are classified into three groups by comparison with chosen limits + simple regional summary; no statistical analysis. Individual time series: Monotone time trends tested using Mann-Kendall statistics (ICES, 1999), modified to take into account occurrence of values reported as below reporting limit. EMMA Workshop April, Copenhagen

6 Regional summaries for each species, tissue and sea region
Statistics on single-series trends by region and species for selected determinands General linear model (GLM) with location and year as factors (to avoid spurious changes due to changes in monitoring program) Summarised in bar charts (normalised to reference period) EMMA Workshop April, Copenhagen

7 Indicator WHS6: Hazardous substance in biota
Bases on primary aggregation for Waterbase/WISE: Statistics on yearly medians for each station for selected species, tissue, determinand : Classification by average for recent years (only data from 1997 to 2004) Monotone time trends by modified Mann-Kendall statistics ( , only series with recent data) Maps from Update 2004, analysis in May-June 2006 EMMA Workshop April, Copenhagen

8 EMMA Workshop 17-18 April, Copenhagen
WS3/9: Questions 10 Questions to WHS6 Comments 10. Are the criteria for classification of concentrations adequate? The assessment of concentration levels can be improved by using agreed European classification schemes. At the moment EU-legislation provides only limits for cadmium, mercury and lead in some foodstuffs. A classification system in line with the water framework directive implementation ought to be developed. EMMA Workshop April, Copenhagen

9 Classification limits in fact sheets
EMMA Workshop April, Copenhagen

10 Methodology of data manipulation (I): Individual time series
Methods have been reported in more detail (EEA 2003a). Time series were identified by station code where available, otherwise by location coordinates. Because of this, some data that constitute parts of the same time series may have been separated into different data series in the analysis because coordinates may vary for a station. Annual concentrations in biota were used converted to dry or wet weight basis, which ever gave best temporal coverage of harmonised data within each identified time series. For assessment of current levels, the average of yearly median concentrations over the period are classified into three groups (cf. Table 2). For series where data reported as below reporting limit occurs, averages have been calculated using the reported upper limits. Monotone time trends were tested using Mann-Kendall statistics (ICES, 1999), modified to take into account occurrence of values reported as below some reporting limit. For series with such data, the method will be more conservative in detecting trends than if reporting limit had been below the measured values for all values. Only series with data for at least 3 years and with data at least up to 1997 were used for the station-by-station trend analysis. Time trends are based on all available data from 1986 to 2004 for each time series - the time coverage is very variable between series. However, some of the trends shown may be based on mainly older data. Significance of trends is based on a two-sided test with a nominal 5 % significance level, separately for each time series, without regard to serial correlation. Assessments of 'No trend' (i.e. no statistically significant trend) may be due both to actual lack of trend and to insufficient data. EMMA Workshop April, Copenhagen

11 Methodology of data manipulation (II): Regional trends
The assessment of general changes in concentrations for each sea are based on the yearly aggregate values (averages) for each combination of location, year, species and tissue. Details of the calculation of these yearly averages are given in the text above. The overall time trends have been extracted from these aggregates for each species, tissue and region. Only series with data for at least 3 years, and with data at least up to 1997 have been used. To diminish the effect on apparent time trends of changing geographical coverage between years, overall yearly average values have been extracted by variance analysis (general linear model - GLM) with location and year as factors. The aim of this is to separate the variation due to stations from the change over time, and achieve yearly averages that are adjusted for differences in geographical coverage between years. Some apparent changes in time may still be due to changes in geographical coverage between years. The analysis has been done using scaled values with the average value of each series, to get a representative average of relative variation over time, regardless of absolute levels at each station. The regional apparent changes in time may not confirm the overall picture based on significant trends found in each time series, since the regional trend will be based on all series seen together. Thus, weak trends in a number of time series that are not significant separate may dominate the results over a few series with significant time trends. (cf. cadmium in cod in Figure 2 and Figure 4 and Map 4). EMMA Workshop April, Copenhagen

12 Regional trend Example: Lindane in liver , Atlantic
(Symbols repeated cyclically over time series) Simple averages reflect changing set of stations over time. Regional trend reflects real trend better, but is still an average of two quite different trends: Stations grouped into two sets with different development in time – too large region? Uncertain identification of time series? Are jumps by factor 100 in and real, or caused by combining positions that are close but have different distance to hot-spots? Or is it a confusion about units? EMMA Workshop April, Copenhagen

13 Problems with time trends in current indicator fact sheet:
Spurious trends in single time series may be caused by: Incorrect identification of time series by coordinates Breakup of time series due to small variations in coordinates Combination of data from nearby locations with different conditions Incorrect specification of units or basis in part of time series May be difficult to judge what is behind apparent regional trends: Changing monitoring programmes over time (GLM helps, but does not completely remove effects of unbalanced data) Remaining problems with <LOD values Combination of different types of stations (impact, representative, reference) into the same regional trend Equal weighting of locations spatially uneven/arbitrary pattern, may not represent a region well as an area. EMMA Workshop April, Copenhagen

14 Improving assessments:
More refined time trend assessments Consider trends on different time scales Breakpoints Most recent significant time trend Smoothers + classification of patterns? Comprehensive trends across species and /or tissue Classify by combined time trend and level? Bad but getting better, Bad and getting worse, Good without any sign of upward trend, Good but with tendencies to get worse Consider spatially based weighting of time series Down-weight multiple series from nearby locations? Make separate assessments for reference, representative and impact stations within regions? Base regional trend assessments on more local/region-specific knowledge Continue development of agreed European classification schemes in line with the water framework directive and marine strategy implementation EMMA Workshop April, Copenhagen


Download ppt "EMMA Workshop April, Copenhagen"

Similar presentations


Ads by Google