Bayesian belief networks in environmental modelling: A review of recent progress

Pages: 13-50

ISBN: 9781606920343

Abstract:

Bayesian Belief Networks (BBNs) are graphical models that incorporate probabilistic relationships among variables. Since their development in the late 1980s, they have increasingly been employed as tools for environmental modelling. This chapter provides a summary of recent progress in using the technique, based on a review of relevant literature. A total of 50 publications were identified describing the application of BBNs to environmental problems, which were examined with respect to: (i) the development of model structures, (ii) identification of the probabilistic relationships between variables, (iii) the approaches used to test or validate models, (iv) sensitivity analysis, and (v) perceptions regarding the overall value of the approach. In the environmental sector, BBNs have to date primarily been used to examine management of natural resources, especially fisheries and water resources. The approach is widely considered to be useful, particularly in modelling domains characterized by high uncertainty, and where model parameterization is strongly dependent on expert knowledge. BBNs possess a number of features that make them particularly valuable in this context, especially where there is a need for models to support decision-making. However, many of the BBN models published to date are preliminary in nature, and there is a widespread lack of rigorous model testing, hindering a full evaluation of the approach. Current trends suggest that use of BBNs is likely to increase in future, extending to a wider range of domains, and involving increased integration with other modelling approaches and analysis of spatial data. However, increased emphasis on model testing is required if the approach is to fulfill its potential as a tool for environmental modelling. © 2009 by Nova Science Publishers, Inc. All rights reserved.

Source: Scopus

Bayesian Belief Networks in Environmental Modelling: A Review of Recent Progress

Authors: Newton, A.

Editors: Findley, P.N.

Pages: 13-50

Publisher: Nova Science Publishers

Place of Publication: New York

ISBN: 978-1606920343

Source: Manual

Preferred by: Adrian Newton