Identifying Models of Environmental Systems' Behaviour

Beck, M.B. (1981). Identifying Models of Environmental Systems' Behaviour. IIASA Working Paper. IIASA, Laxenburg, Austria: WP-81-158

Preview

Text WP-81-158.pdf Download (894kB) | Preview

Abstract

This paper addresses the question of how theories are developed about the behavior of large, complex systems such as those typically encountered in managing environmental quality. The specific problem considered is that of model structure identification by reference to experimental, in situ field data. A conceptual definition of this problem is given in terms of the notion of testing model hypotheses to the point of failure. An approach to solving the problem is proposed in which the use of recursive model parameter estimation algorithms is a central feature. This approach is illustrated by a case study in developing a dynamic model of water quality in the Bedford Ouse river in central-eastern England. The results are organized around the two principles of attempting to falsify confident hypotheses and of speculating about relatively uncertain hypotheses in order to modify inadequate prior hypotheses. The essential difficulty demonstrated by the case study is one of absorbing and interpreting the diagnostic evidence of field data analysis and this is ultimately a difficulty associated with the complex and intrinsically indivisible nature of large-scale systems.

Actions (login required)

View Item