A Bayesian Framework for Assessing Dynamic Hydrological Systems

Hydrological models are a popular tool for simulating catchment processes in response to a rainfall event. The wide range of available models means that hydrologists are faced with the problem of determining which model is best applied to a catchment in any modeling exercise. An attractive alternative to selecting a single hydrological model is to combine the results from several models, thereby providing a model performance that is substantially better than any model alone. Methods based on Bayesian statistical techniques provide an ideal means to compare and combine competing models as they explicitly account for model uncertainty. Bayesian model averaging combines individual models by weighting model outputs proportional to their respective posterior probability of selection. However, the necessity of having fixed weights over the length of the simulation period means that the relative usefulness of different models at different times is not considered.