PORTRAYING CLIMATE SCENARIO UNCERTAINTIES IN RELATION TO TOLERABLE REGIONAL CLIMATE-CHANGE

Analyses of the impacts of future anthropogenic climate change on envi ronmental and social systems have been dominated by a 'top-down' appro ach. A climate change scenario is defined using output from one or mor e climate model experiments, the scenario is run through one or more e nvironmental simulation models, and the impacts of the prescribed clim ate change evaluated. This approach places a considerable burden on th e selection of which climate change scenarios should drive the impacts assessment. An alternative approach for assessing possible impacts of climate change follows a 'bottom-up' (or inverse modelling) approach. Here, the sequence of analysis steps is inverted. An assessment is ma de of what range of magnitudes and/or rates of regional climate change could be adapted to by an environmental or social exposure unit. The question is then asked of the climate scenario developer, how likely i s it that future regional climate change will exceed these limits, and by when? Under what scenario or modelling assumptions will these limi ts be exceeded? And how do these future changes relate to current clim ate variability? In this paper we present a systematic approach for co nsidering the effect of a set of scenario and modelling uncertainties on the likelihood of critical climate change being exceeded for partic ular exposure units. We present this assessment in the context of obse rved climate change over the last 100 yr and illustrate the approach f or the UK and for 2 thresholds of climate change. These are defined, v ery simplistically, in terms of summer mean temperature and rainfall a nd as such may nominally be regarded as relating to water resources in the UK. We argue that one of the strongest advantages of this approac h is that it disarms those who wield climate change scenarios as thoug h they were in some sense 'predictions' of future climate. By visualis ing the effects on realised future climate of different modelling assu mptions and scenario uncertainties, we make more transparent the judge ments that must be made in assessing the significance of climate chang e impacts on different regional exposure units.