Brief description

The research project attempts to address the issues of uncertainty estimations in rainfall runoff modelling (i.e. fully distributed catchment models), especially the uncertainty estimation of the input data, model structures, parameter estimations, and their combinations. This research is intended to find optimum combinations of model complexity, spatial resolution, and data uncertainty yielding a minimum prediction uncertainty (Rode, 2006).

This research will examine the effects on predictive uncertainty of various uncertainty sources:
a) the input data uncertainty (e.g. their measurement accuracy or data interpolations),
b) model structure/parameter uncertainty (e.g. their complexity of hydrological processes, complexity of the model structures, the scale of the simulation, and the spatial resolution of the modelling), and
c) combinations of uncertainties in input data and model structure/parameters. Based on the individual and combined uncertainty estimation results, a stochastic modelling tool will be developed, allowing investigation of optimal and efficient representation of uncertainties. The full potential of uncertainty estimation in rainfall-runoff model will be achieved by combination of stochastic rainfall generation model and fully distributed model. The stochastic rainfall model will be developed based on the various scenarios of input uncertainties and DATA Engine. The stochastic fully distributed catchment model will be investigated by using the Monte Carlo methods, GLUE method, and Bayesian approaches, on the basis of model complexity, the scale of the simulation, and the spatial resolution of the models. Furthermore the fully stochastic rainfall runoff model will be developed in combination with two stochastic models. Their uncertainty prediction in the applications will be suggested with confidence.