© 2000 Heron Publishing—Victoria, Canada
Bayesian synthesis for quantifying uncertainty in predictions from process models
Edwin J. Green (1), David W. MacFarlane (1) and Harry T. Valentine (2)
1. Department of Ecology, Evolution and Natural Resources, Rutgers University, 14 College Farm Road, New Brunswick, NJ 08901-8551,
USA / 2. USDA Forest Service, Northeastern Research Station, P.O. Box 640, Durham, NH 03824-0640, USA / Received May 7, 1998
Summary
The Bayesian synthesis method is reviewed and judged to be useful for determining posterior distributions and interval estimates
for inputs and outputs of process-based forest models. The method furnishes posterior distributions of the values of a model's
parameters and response variables. The method also provides estimates of correlation among the parameters and output variables.
Bayesian synthesis is the only type of uncertainty analysis that affords incorporation of all the information available to
the investigator, in addition to the information contained in the model itself.
Keywords:
confidence intervals, mechanistic models, posterior distributions, sensitivity analysis.
