Using response surfaces to estimate impacts of climate change on flood peaks: assessment of uncertainty

The potential impacts of climate change are an increasing focus of research and ever-larger climate projection ensembles are available, making standard impact assessments more onerous. An alternative way of estimating impacts involves response surfaces, which present the change in a given indicator for a large number of plausible climatic changes defined on a regular sensitivity domain. Sets of climate change projections can then be overlaid on the response surface and impacts estimated from the nearest corresponding points of the sensitivity domain, providing a powerful method for fast impact estimation for multiple projections and locations. However, the effect of assumptions necessary for initial response surface development must be assessed. This paper assesses the uncertainty introduced by use of a sensitivity framework for estimating changes in 20-year return period flood peaks in Britain. This sensitivity domain involves mean annual and seasonal precipitation changes, and a number of simplifications were necessary for consistency and to reduce dimensionality. The effect of these is investigated for nine catchments across Britain, representing nine typical response surfaces (response types), using three sets of climate projections. The results show that catchments can have different causes of uncertainty, and some catchments have an overall higher level of uncertainty than others. These differences are compatible with the underlying climatological and hydrological differences between the response types, giving confidence in generalisation of the results. This enables the development of uncertainty allowances by response type, to be used alongside the response surfaces to provide more robust impact estimates.