By Dr Neil McIntyre*, Nataliya Bulygina, Caroline Ballard

*Imperial College, London

Chair: Prof Paul Bates

Abstract:

Environmental modelling requires that a prior conceptualisation of the system is combined with evidence of behaviour. Most modelling exercises do this implicitly but without much thought about how to maximise the diversity and value of information included, or how to formally estimate uncertainty in the information or the model outputs. Using a Bayesian procedure, we combine three distinctly different types of information into a catchment scale conceptual rainfall-runoff model: 1) outputs from a small-scale physics-based model; 2) nationally regionalised flow indices; and 3) a few available flow measurements. The method is applied to a spatially distributed model of the Hodder catchment to illustrate the method's strengths and possible flaws.