Developing Local Environmental Decision Tools Using Bayesian Belief Networks
Rognvald I. Smith, Jan M. Dick
Centre for Ecology and Hydrology, Edinburgh, United Kingdom

Following the Millennium Ecosystem Assessment, the concept of ecosystem services promoted the idea of a holistic assessment of an ecosystem to inform environmental management. It can be difficult to put this idea into practice as the manager has to consider all aspects including the functioning of the environmental components and how these may react to change, as well as the social and economic context for any decision process. Different actors in any community have different perspectives on the exploitation of their surroundings and there are often many layers of environmental regulation reflecting the policy concerns of differing governance levels.

We propose the use of a Bayesian belief network (BBN) set within a decision support tool for local communities and individual land or environmental managers. While these users do not have access to the range of complex models available to policymakers and environmental scientists, they do benefit from local environmental knowledge and data. The BBN is set up with the user in the central position. The user constructs a network of the required complexity using an ecosystem services framework, so the basic graphical structure is enhanced by more specific detail relevant to the proposed decision. The user is also empowered to choose between data sources, using either their own local knowledge to build parts of the network structure or retrieving information from libraries of accessible general data resources. The tool exploits the advantage of a BBN allowing use of different types of information (including expert opinion) with various levels of uncertainty and tracks that statistical error into multiple outcomes, allowing transparency in balancing these values when making a decision. The decision tool also has the ability to flag regulatory constraints on any proposed course of action.

The aim of this decision tool is to exploit the BBN as a rigorous statistical method for pooling environmental, social and economic data and to develop the information sources to promote its use within local communities. Improved information and better local decision processes also benefit all levels of rural environmental governance.

Some of this work has been developed under the EU FP7 project TESS: Transactional Environmental Support System [212304]

Smith RI, Dick JM & Scott EM (2011) The role of statistics in the analysis of ecosystem services. Environmetrics. DOI:10.1002/env.1107

Keywords: Local environmental decisions tools; Bayesian belief network

Biography: Ron Smith is an environmental statistician at the Centre for Ecology and Hydrology in the UK. He has worked on an extensive range of environmental problems including air pollution effects research, biodiversity and, more recently, ecosystem function and services. He is an active member of the Royal Statistical Society, leading their Panel on Statistics for Ecosystem Change, and of the International Environmetric Society.