A Bayesian Network approach to integrating economic and biophysical modelling

It has been recognised that the complex interrelationships between environmental processes and socio-economic systems require integrated approaches to catchment management (Cai et al., 2003; Jakeman and Letcher, 2003;). Until recently, integrated modelling tools were often limited to either biophysical processes to assess environmental changes, or to economic models focussing on socio-economic systems. Despite the policy interest in integrated catchment management, there is still limited experience in linking environmental and socio-economic systems in one modelling framework (Heinz, 2007; Reinhard, 2006). The study reported in this paper aims to demonstrate how biophysical science can be linked with economic non-market valuation in an integrated Bayesian Network (BN) framework. In the context of this study, a BN was deemed a suitable modelling technique to integrate the various systems impacted by catchment management changes. The model can incorporate data of different types and quality, and its structure provides an explicit depiction of the uncertainties in environmental and economic systems. We develop an integrated model for a case study of the George catchment in northeast Tasmania, Australia. The modelling framework incorporates a water quality model, ecological information and economic data (Figure 1). The major innovations of the research reported here are the parallel development of the various models, enabling increased integration, and the use of an environmental valuation technique known as Choice Experiments (CE) to elicit economic information on the non-market costs and benefits of catchment management changes. To the best of our knowledge, CE data has not previously been linked to biophysical modelling in a BN framework. Furthermore, the biophysical modelling provides more scientific foundation for the valuation study than is typically available (Brookshire, 2007). In this paper, an overview is given of the BN development process that integrates CE with natural science modelling. Specific issues related to linking economic and biophysical knowledge include the selection and description of catchment environment indicators relevant for all research components, and matching the variables' states in the BN to the levels of the environmental attributes in the CE survey. Several of the challenges in developing an integrated, multidisciplinary model are discussed in this paper.