Development of Bayesian belief networks

clean groundwater (<0.1 μg/l) for theHavelse catchment area in the future (30-50 years).This overall Boolean indicator variable provides anoverall estimate based on content of pesticides ingroundwater, shallow groundwater and surface water.The first is weighted highest for the total “score”. Therisk of pollution from surface water can eventually beminimised significantly by moving the wellfield awayfrom the present location in the creek valley.Perception of vulnerability impacts this variablesignificantly.States: “false” (must be stopped) or “true” (safedrinking can be abstracted).The variable is basedon stakeholderinvolvement. May beused for benchmarkingseveral of CE’swellfields (55 innorthern Zealand). C21 Remove pointsourcesVariable C21 describes a possible action for CEfocus: a removal of point sources both in urban andrural areas. The action to remove may involve otherstakeholders (Frederiksborg County andmunicipalities or even farmers’ organisations):States: “false” (not removed) or “true” (remove).The variable is basedon stakeholderinvolvementwww.fba.dk C22 Perception ofvulnerabilityVariable C22 is a new controlling factor that wasincluded after the collection of feedback fromstakeholders (Step 7 in the protocol). The farmers’organisations have the attitude that the deepgroundwater will not be polluted above the MACvalue because the deep aquifer is less vulnerable(compared to average Danish vulnerabilities). Maybethe KUPA project can help clarify this majoruncertainty; the limited number of samples is notenough to show which party has the most correct“perceptual model” of this problem.States: “proxy basin” (e.g. based on monitoring datafrom Denmark) or farmers’ organisation (NOLA).The variable is basedon stakeholderfeedback. Sources:GEUS (2003)Brüsch (2004)Henriksen andSonnenborg (2003)www.geus.dkwww.kupa.dkwww.vandmodel.dk With the same control variables as in Figure 6.13 the figure below (6.14) shows the effectof “perception of vulnerability”. The farmers’ organisation (NOLA) believes that the deep groundwater will not be pollutedabove the maximum limit value (probability of pesticides >0.1 μg/l = 0%). Also, supply isregarded rather safe for present conditions (probability of safe supply = true is 90%). Onthe other hand, GEUS experts and CE (based on proxy basin perception) believe that withthe current application of pesticides, the deep groundwater probably will be polluted abovemaximum limit value (probability = 7%). Also, supply is at risk because the probability forsafe supply is only 66%. See Figure 6.13.