Hydrological Uncertainty in Practice: Reasons to be Cheerful

Abstract:

Assessments of hydrological uncertainty often show multiple, interacting uncertainty sources and worryingly large error bounds. Hydrological data such as rainfall and flow measurements, hydrological model structure and model parameters are all potential sources of uncertainty. However, the effects of hydrological uncertainty on our ability to make predictions or decisions can vary widely depending on the question asked, and the choice of question may in part mitigate the effects of high uncertainties.

This presentation will use real-world examples of uncertainty assessment in practice, taking an overview of several projects in the UK and New Zealand that involved estimation of uncertainty in predictions of flow regime, water quality trends, subsurface catchment boundaries and flood risk. I will discuss the different strategies used to quantify uncertainty in each project, how uncertainty was communicated, and the effect of uncertainty on the project conclusions and recommendations.

Of particular interest is whether hydrological methods can be designed to be robust to known sources of uncertainty. Potential strategies include changing the specification of hydrological indices to avoid highly uncertain subsets or analyses of data, or ensuring that only sources of uncertainty directly relevant to the specific management question are considered. I will discuss how these strategies were applied, or could have been improved, in each of the example applications.