**PUB Working Group 7 (WG7): Uncertainty Estimation for Hydrological Modelling
**

**Objectives**

The main task of this WG is to stimulate progress and guidance on uncertainty estimation and model evaluation (diagnostics) theory and techniques for hydrological science and for operational use. The fundamental science questions are:

- How can we (explicitly) estimate and propagate all sources of uncertainty in hydrological modeling?

- What is an appropriate framework for (model/method) evaluation under uncertainty?

Background Uncertainty is an unavoidable element in any hydrologic modeling study. This uncertainty stems form the parameters, the model structure and me asurements of input and output data. This uncertainty is likely to be particularly severe in basins for which no (or only a few) measurements of their response (eg streamflow) are available, since these measurements are normally used to reduce (at least) the uncertainty in the parameters.

Whilst a commitment for the quantification of uncertainties in hydrologic flux predictions lies at the heart of the PUB initiative, methods to estimate and propagate this uncertainty have so far been limited in their ability to distinguish between different sources of uncertainty and in the use of the retrieved information to improve the model structure analyzed. Examples of these approaches are point methods, set theoretic techniques where the uncertainty is largely treated implicitly or Bayesian approaches which currently make strong assumptions about the nature of uncertainties involved.

However, a proper uncertainty framework is required for model/method evaluation/development and the analysis of model predictions. A framework is needed which allows for the explicit incorporation of different sources of uncertainties as well as for the incorporation of multiple sources and types of information. The approach should also allow for the recursive processing of information and provide feedback about structural discrepancies of the evaluated model. There is no clear consensus about how such an approach could be implemented. Also, there is little guidance about how information/uncertainties should be considered/evaluated when they are spatially distributed and/or at different scales.

The main task of this working group is to provide progress and guidance on uncertainty estimation and model evaluation (diagnostics) theory and techniques for hydrologic science and for operational use.

Methods investigated range from set theoretic methods, over Bayesian to simple point estimation techniques.

More information is available on the Group's website

**Key Participants**:

- Thorsten Wagener Pennsylvania State University