SGER: Development and Assessment of Ecological Process Models

University Of Washington, Seattle WA

Investigators

Abstract

Process models of ecological system directly represent how a system functions. They can be used to generate new ecological theory, and sometimes, can be used to predict the consequences of environmental management. A new method, the Pareto Optimal Model Assessment Cycle (POMAC) for developing and assessing process models is explored and contrasted with the more traditional technique of least squares fitting. POMAC uses evolutionary computation to seek parameters of the model that can simulate different outputs of process models. The model that can be developed using POMAC will be compared using the standard method of least squares fitting and also the Akaike Information Criterion, designed to produce parsimonious models. Developing POMAC should lead to a more effective method for model development and assessment. Saying that a model accounts for a particular amount of the total sums of squares of the data does not provide an adequate assessment-we need to know how it has achieved that. POMAC permits many different types of data to be used simultaneously, giving greater potential use of ecological and environmental data.