1st International Congress on Environmental Modelling and Software - Lugano, Switzerland - June 2002
Keywords
rainfall-runoff modelling, model identification, generic frameworks, uncertainty
Start Date
1-7-2002 12:00 AM
Abstract
A task which is often central to hydrological modelling is the identification of an appropriate model structure and a suitable parameter set for a specific case, i.e. a given set of modelling objectives, catchment characteristics and data. However, this identification process is difficult and will often result in a range of possible models, i.e. different parameter sets within a certain model structure, or different model structures. Two generic rainfall-runoff modelling and Monte Carlo analysis toolboxes have been developed to allow for the implementation and subsequent comparison of spatially lumped, metric and parametric model components in order to identify the model(s) and model structure(s) most suitable for a given application. These toolboxes include the use of multi-objective and novel dynamic approaches to performance and identifiability analysis. This enables a more objective analysis of the level of model complexity that is supported by the data. It also enables the modeller to test whether a given model structure is consistent with underlying assumptions, reducing model structural uncertainty. An application of these approaches to a catchment located in the South of England demonstrates the advantages of a flexible framework, combined with novel approaches to model identification.
A Generic Framework for the Identification of Parsimonious Rainfall-Runoff Models
A task which is often central to hydrological modelling is the identification of an appropriate model structure and a suitable parameter set for a specific case, i.e. a given set of modelling objectives, catchment characteristics and data. However, this identification process is difficult and will often result in a range of possible models, i.e. different parameter sets within a certain model structure, or different model structures. Two generic rainfall-runoff modelling and Monte Carlo analysis toolboxes have been developed to allow for the implementation and subsequent comparison of spatially lumped, metric and parametric model components in order to identify the model(s) and model structure(s) most suitable for a given application. These toolboxes include the use of multi-objective and novel dynamic approaches to performance and identifiability analysis. This enables a more objective analysis of the level of model complexity that is supported by the data. It also enables the modeller to test whether a given model structure is consistent with underlying assumptions, reducing model structural uncertainty. An application of these approaches to a catchment located in the South of England demonstrates the advantages of a flexible framework, combined with novel approaches to model identification.
