Object-Oriented Bayesian Networks for Participatory Water Management: Two Case Studies in Spain

Abstract

Object-oriented Bayesian networks (OOBNs) have recently been introduced to model water systems that can be represented as repetitive patterns. This paper shows the way in which OOBNs can be used as a groundwater management decision support system in two Spanish case studies. The two areas, in the southern and eastern parts of inland Spain, are characterized by a semiarid climate, water scarcity, and frequent droughts; consequently, the agrarian economy in both cases depends on the provision of irrigation from groundwater sources. Both case studies are illustrative examples of conflict among various water actors, complexity, and uncertainty about the consequences of water management actions. Each study is approached from a different viewpoint: one from an agroeconomic and the other from a hydrogeological perspective. The sites display different degrees of aquifer overexploitation and agrarian profitability. This indicates that, in each case, the effects generated by water management interventions and the time needed for recovery to natural conditions are different. The processes governing both systems can be represented as a series of repeating patterns, which makes them good candidates for an OOBN analysis. The OOBNs developed have been constructed with the participation of stakeholders to help minimize conflicts and make management decisions more understandable and acceptable for all users. The innovative nature of this research lies with the implementation of OOBNs for groundwater management. Results of the study demonstrate that the OOBN tool is a powerful decision support system that can help managers make decisions in cases for which the consequences of alternative interventions are unknown. It provides the probabilities of obtaining certain outcomes from alternative management actions for the economy and for the state of the environment. OOBNs meet the requirements of the European Water Framework Directive as a simple, participative, and integrative tool. Finally, this research represents a starting point for additional applications to support the integrated groundwater management of other complex water resources systems.