Aquifers Management through Evolutionary Bayesian Networks: The Altiplano Case Study (SE Spain)

An approach for the integration of Object‐Oriented Bayesian Networks (OOBNs) and Evolutionary Multiobjective Optimization (EMO) is proposed for integrated water resource management and decision support. Bayesian Networks (BNs) offer a novel and powerful tool for modelling complex water systems, facilitating the use of hierarchical modelling by improving the efficiency and communication between the different parts of a model. EMO offers a range of non‐dominated optimal management solutions on a Pareto front that facilitate the identification of tradeoffs among conflicting criteria regarding stakeholder’s preferences. The integrated tool provides new possibilities for undertaking an integrated analysis where stakeholder participation can play an important role. It is used for simultaneously analysing the whole water system, characterising uncertainty as probabilities and evaluating different management options. The tool is applied to an overexploited water system located in Southern Spain that is supplied totally by groundwater. In this study, a complex model based on BNs is designed and used as the core of the study. The transition to Evolutionary Bayesian networks (EOBNs) allows stakeholder involvement to be utilized more effectively for designing and evaluating the model’s consistency, and taking into account their conflicting interests.