Parameter Uncertainty And Interactions In A Complex Catchment Modelling System
- Chemical Industry Press
- Publication Type:
- Conference Proceeding
- Proceedings of the 9th International Conference on Hydroinformatics 2010, 2010, pp. 1359 - 1366
- Issue Date:
Copyright Clearance Process
- Recently Added
- In Progress
- Closed Access
This item is closed access and not available.
The importance of estimating parameter uncertainty has been recognised in the literature with Bayesian methods being the most popular approaches for undertaking parameter uncertainty estimation. With these approaches, the parameter uncertainty is described by the posterior distribution. Application of these methods usually requires a priori knowledge about the proposal distribution of the values of the parameters. However, usually there is little a priori knowledge about the proposal distribution for the parameter values in many catchment models. Furthermore, the difficulties of exploring posterior distribution of parameter values increases as the dimensionality of the problem increases. An alternative approach is to use a GLUE methodology to develop a heuristic estimate of the parameter uncertainty. In this application of a GLUE framework, a real-value coding genetic algorithm was used to undertake an uncertainty analysis on spatially variable control parameters associated with implementation of the Storm Water Management Model (SWMM). The focus of the study was to investigate the values of spatially variable control parameters that are associated with behavioural parameter sets in a complex catchment modelling system. Of particular concern was the frequency that values within a defined range occurred. Interactions among the values of the spatially variable parameters were investigated also. Results obtained from an application of the Pearson Correlation method indicated no clear relationship between any two control parameter values. However, distinct relationships among a group of parameters were detected suggesting that the simulation performance was affected more by the combination of parameter values than by the values of the individual parameters.
Please use this identifier to cite or link to this item: