A Belief Network Approach to Modeling of Environmental Change: The Methodology and Prospects for Application

Varis, O. (1994). A Belief Network Approach to Modeling of Environmental Change: The Methodology and Prospects for Application. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-94-040

Preview

Text WP-94-040.pdf Download (793kB) | Preview

Abstract

In environmental management, assessments and far-reaching decisions must typically be made under very high or extreme uncertainty. The future development of the environment in interaction with societies in transition is very difficult to forecast. This is the case regardless of whether the change is introduced actively at the project or policy level, or passively through accumulated environmental deterioration or climatic change. This study presents a belief network methodology designed specifically for modeling environmental change. Belief networks contain a set of interlinked nodes. Prior probability distributions of nodes are updated with information from the rest of the network, according to transfer information in links. A link can transmit information in two directions. The existing belief network methodology was extended in several ways to meet the multiple requirements of environmental modeling. Most notably, two-layered parallel linking of nodes was allowed: the conventional probabilistic linking, and linking of outcomes of probability distributions using deterministic or logical relations. Moreover, several decision analysis techniques were included. The applicability of the methodology is discussed in reference to the following topics: knowledge acquisition, decision analytic modeling, mechanistic and process modeling, topological and spatial modeling, learning and adaptive modeling, and hybrid use.

Actions (login required)

View Item