Evolutionary Bayesian Belief Networks for Participatory Water Resources Management under Uncertainty

Abstract

A participatory integrated (social, economic, environmental) approach based on causal loop diagram, Bayesian belief networks and evolutionary multiobjective optimisation is proposed for efficient water resources management. The proposed methodology incorporates all the conflicting objectives in the decision making process. Causal loop diagram allows a range of different factors to be considered simultaneously and provides a framework within which the contributions of stakeholders can be taken into account. Bayesian belief networks takes into account uncertainty by assigning probability to those variables whose states are not certain. The integration of Bayesian belief network with evolutionary multiobjective optimisation algorithm allows analysis of trade-off between different objectives and incorporation and acknowledgement of a broader set of decision goals into the search and decision making process. The proposed methodology is used to model decision making process for complex environmental problems, considering uncertainties, addressing temporal dynamics, uncovering discrepancies in decision analysis process (e.g. completeness or redundancy of the model based on utility function) and generating policy options that trade-off between conflicting objectives. The effectiveness of the proposed methodology is examined in several water resources management problems. The case studies include optimum water demand management, UK; management of groundwater contamination of Copenhagen source capture zone areas, Denmark and simultaneous optimum management of four overexploited aquifers in Spain. It is shown that the proposed methodology generates large number of management options that trade-off between different objectives. The remaining task is to choose, depending on the preference of decision makers, a group of solutions for more detailed analysis.