1. Lessons from the European Union Linda Green URI Watershed Watch November 16, 2007 presented with permission…

2. A Graphical Presentation of Water Quality Data in Time and Space Dr Peter G Stoks RIWA/IAWR stoks@riwa.org Enhancing the States’ Lake Management Programs: Interpreting Lake Quality Data for Diverse Audiences April, 2007; Chicago, IL

3. Umbrella organization of 3 Associations RIWA:Netherlands ARW:lower Germany AWBR:upstream Germany, Switzerland Mission: Source water quality should allow drinking water production using simple treatment only! 120 utilities 30 million consumers

4. RIWA WQ monitoring network (part of IAWR basinwide network) Cooperation with Nat’l Dutch and German water authorities –Harmonized program (WQ variables, methods, data exchange,…) Five locations in the Dutch part of the Rhine basin –German-Dutch border, intake sites Trend detection and compliance testing –“legal standards” & “emerging contaminants” –Chemical & biological

5. Presenting Water Quality Data... the usual way Annual WQ reports – Text, graphs, tables – Executive summaries containing demands and/or recommendations Shift observed in target group’s WQ background –From directors knowledgeable about water quality –Appointed executives

6. The old way to publish our data

7. Management is usually not too bright…

8. …and has the attention span of a hamster

9. Need for simple WQ visualization So, this does not work anymore (if it ever really did…)

10. Solution: Color palettes combining locations and variables allows for basinwide overview of multiple variables color indicates WQ development over time –WQ improvement or deterioration at a glance area coverage indicates accuracy of statements –The less data, the less area coverage

11. Combining compliance ratio and trend yields Status Good status: - compliance ratio “good” and trend stable - compliance ratio “moderate” and trend improving Bad status: - compliance ratio “bad” and trend stable - compliance ratio “moderate” and trend deteriorating Moderate status: - all other combinations

12. above standard = non-compliance (except for O 2 ) between 0.8 x standard and standard = OK but look out! = OK but look out! below 0.8 x standard = compliance Compliance testing based on maximum values (or other criteria)

13. Compliance ratio max value of variable r max = divided by its WQstandard (or objective) r max interpretationcolor < 0.8good 0.8 - 1.0moderate > 1.0bad

14. Combined with trend detection Trend detection based on 5 year period - commercially available STAT package - using parametric as well as non parametric statistics - 13 data points / year minimum, reduced to quarterly avg

15. Trend determined using a STAT-package (modified linear regression model, 95% conf interval)

16. Combined with quantity of data n < 10 n ≥ 20 20 > n ≥ 10 Quantity

17. Trend detection 1 5 years period equidistant data (quarterly)

18. Trend detection 2 Aspects: Seasonal effects Serial (auto) correlation Residuals from a normal distribution

19. Trend detection Timeseries advanced linear regression normal residuals n y output yearly timescale autocor. residuals n y seasonal Mann-Kendall seasonal & autocorrelation n >=20 trend autocor. Mann-Kendall seasonal Mann-Kendall output y yy y autocor. n yn nn n bigger timescale parametric non parametric

20. Trend detection advanced linear regression intercept trend componentseasonal component autoregressive component model residue Software: Trendanalist by ICASTAT Or Y= a + bx (quarterly data) + b1*Q1 + b2*Q2 + b3*Q3( + b4*Q4)

21. Put simply…create a pictogram above standard 0.8-1.0 of standard below 0.8 of standard Compliance uptrend downtrend no trend or not detectable Trend n < 10 n ≥ 20 20 > n ≥ 10 Quantity

22. Used in annual reports

23. But wait, there’s more… Most current graphs plot 2 or 3 dimensions of a single WQ variable - change over time at given location - change over time in space

24. 2 dimensions: concentration vs time at 1 location

25. 3 dimensions: concentration vs time & location Date Downstream Qty

26. Pictograms combine WQ variables, locations AND time allows for basinwide overview of multiple variables – y-axis might list different WQ variables – x-axis might list different locations (e.g. along a river) color indicates compliance or violation of standards – easy overview “at a glance” of status arrow indicates trends – easy overview “at a glance” of development over time area coverage indicates accuracy of statements – the less data, the less area coverage

27. Also applicable at technical level plotting a single WQ variable at different locations OR plotting different (related) WQ variables at same location over time gives info on reliability of data → odd values easily detected

28. What it looks like Ammonia in the Rhine 1972 – 2004 Downstream year

29. Standard violations and trends can be presented - for a single variable at different locations over time - for several variables at a fixed location over time Easily understood by non-expert decision makers Also: triggers WQ analist about curious events ( with software built into database management system)

30. In summary above standard 0.8-1.0 of standard below 0.8 of standard Compliance uptrend downtrend no trend or not detectable Trend n < 10 n ≥ 20 20 > n ≥ 10 Quantity

31. Lessons from the European Union above standard 0.8-1.0 of standard below 0.8 of standard Compliance uptrend downtrend no trend or not detectable Trend n < 10 n ≥ 20 20 > n ≥ 10 Quantity Dr Peter G Stoks Assn of Rhine Water Works RIWA/IAWR stoks@riwa.org