Structuring rainfall-landuse-runoff models for a large catchment in N.S.W.

Abstract

To retain the advantages of conceptual modelling while addressing the difficulty of parameter estimation, a procedure is proposed in which model structure is developed progressively, with model performance and associated parameter uncertainty being evaluated at each stage, to determine whether and which further processes should be added. When developing a structure to model runoff from a large catchment, a combination of global statistical performance measures and information concerning conditional error distribution was required; the former to enable the optimal performance of each structural variant to be compared, the latter to assess conformity of the errors to standard statistical assumptions and hence to indicate how the current variant should be augmented.