Integrated water resources management of overexploited hydrogeological systems using Object-Oriented Bayesian Networks

Abstract

Object-Oriented Bayesian Networks (OOBNs) have been used increasingly over the past few decades in fields as diverse as medicine, transport and aeronautics. In this paper, OOBNs are applied to the domain of integrated water management and used as a Decision Support System (DSS). This pioneering study, set in the Altiplano region of Murcia in Southern Spain, describes a method for the integrated analysis of a complex water system supplied by groundwater from four aquifers. This method is based on the development of a multivariable integrated technique based on Bayes' theorem. After identifying all relevant factors related to water management in the area these were then translated to variables within a Bayesian Network (BN) and the relationships between them investigated. Each network represented one of the four aquifer units. These individual BNs were then linked to form an OOBN which was used to represent the complex real-world situation. In this way a DSS to simulate the entire water system was constructed using a group of conventional Bns, linked to produce an OOBN. The main stakeholders of the region contributed to network design and construction throughout the entire process. The paper shows how this type of DSS can be used to evaluate the impacts of a range of management strategies that are available to local planners.