By: David Barton
WWW: Martin Karlsen & Anders L Madsen
The BBN illustrates an operational interpretation of integrated valuation of ecosystem services from eutrophication mitigation measures.
An object-oriented Bayesian network is used to link a cascade of sub-models across drivers, pressures, states, impacts and societal responses to lake eutrophication. Systems dynamic, empirical and expert judgement models are integrated in the network to assess trade-offs in ecological, social and economic values from improving lake ecological condition achieved using nutrient abatement measures in a catchment in South-Eastern Norway. The data used is documented in Barton et al. (2016).
The integrated valuation BBN makes it possible to assess the combined uncertainty in eutropication mitigation management predictions due to natural temporal variability, spatial heterogeneity, monitoring data resolution, sub-model prediction error and information loss at model interfaces.
This online BBN demonstrates the spatial mapping of household willingness to pay (WTP) (NOK/year) for a sewage fee (blue node, Figure 2).
Below are some HUGIN widgets - main tab, dashboard - for interacting with the model shown on the right (click on the probability bar to instantiate a node or remove evidence):
Expected Utility (NOK/year)
Expected Utility (NOK/year)
Main tab: Here you can select three programmes of measures corresponding to two different baseline situations without additional mitigation measures (post-2006/07, pre-2006/07) and a scenario where all cropping areas in the catchment are converted to pasture resulting in less fertilisation and plowing, and all blue-green structural rehabilitation measures are implemented (constructed wetlands, vegetation buffers, nutrient point sources treated). WTP for different users can be selected. Press updated map to see the result on household WTP in the map.
Things to notice: The distribution of WTP is positive when no scenario is selected due to skewed joint probability distributions across the integrated dynamic model (in principle it should be zero in a static baseline/status quo situation). WTP differs between Lake User types and at distances from the Vansjø lakes (5km bands). Notice how expected WTP doubles with the mitigation measures scenario (S4).
Dashboard tab: In this tab, some of the nodes in the model are illustrated. To the far right the measures considered in each scenario are shown (100% if implemented). Tot-P, Algal-P show the eutropication state predicted by run-off and lake water quality model and Lake Ecological Status shows the probability of a changes from bad(red), moderate(yellow) to good(green) or very good(blue) ecological status. Click on different years to see how effectiveness of measures varies year-on-year.
Things to notice: The uncertainty regarding lake water quality and improvement in lake status. The uncertain effectiveness of measures is driven by year-on-year natural variability in the catchment.
The online BBN application was made possible by the FP7 OpenNESS project (grant no. 308428). We are also grateful for financial support from several sources over the years it has taken to develop the various models included in the BBN, including research Council of Norway for the EUTROPIA project (grant no. 190028/S30)), the European Union for the FP6 AQUAMONEY (contract no. SSPI-022723), FP7 REFRESH (grant no. 244121).
D. N. Barton, T. Andersen, O. Bergland, A. Engebretsen, S. J. Moe,
G.I. Orderud, K. Tominaga, E. Romstad, and R.D. Vogt (2016) Eutropia:
Integrated Valuation of Lake Eutrophication Abatement Decisions Using
a Bayesian Belief Network. pp. 297-320. Chap.14 in Niel, Z. P. (Ed.)
Handbook of Applied Systems Science. Routledge, New York and London.
Useful references for those interested in BBN include:
Kjærulff, U. B. and Madsen, A. L. (2013) Bayesian Networks and Influence Diagrams: A Guide to Construction and Analysis. Springer, Second Edition.
For further details contact: David Barton (David(dot)Barton(at)nina(dot)no)