9.10 – Integration of Bayesian Inference Techniques with Mathematical Modeling

摘要: The credibility of the scientific methodology numerical models and their adequacy to form basis public policy decisions have been frequently challenged. first part this chapter aims address issue model reliability by evaluating current state aquatic biogeochemical modeling. We provide evidence that there is still considerable controversy among modelers resource managers about how develop, evaluate, interpret mathematical models. Our arguments are (1) not always developed in a consistent manner, clearly stated purpose, predetermined acceptable performance level, (2) potential users select without properly assessing technical value. second presentation argues development novel methods for rigorously uncertainty underlying predictions should be top priority modeling community. Striving analysis tools, we introduce Bayesian calibration process-based as methodological advancement warrants consideration ecosystem research. This framework combines advantageous features both statistical approaches, is, mechanistic understanding remains within bounds data-based parameter estimation. incorporation mechanism improves confidence made variety conditions, whereas an empirical value selection allow realistic estimates predictive uncertainty. Other advantages approach include ability sequentially update beliefs new knowledge available, consistency with process progressive learning practice adaptive management. Finally, illustrate some anticipated benefits from framework, well suited stakeholders makers when making environmental management decisions, using Hamilton Harbour – eutrophic system Ontario, Canada case study.