The detection and modelling of the impact of climate change on water resources has been the subject of study for some thirty years. Until recently, impact assessments were based on ad hoc sensitivity analyses or factorial experiments in which all possible combinations of global climate models (GCMs) and emissions scenarios were considered, together with a small number of downscaling methods and hydrological models. Given the large number of runs available from the latest generation of GCMs and increasing computer power the latter approach will lead to increased information overload. In turn, this could lead to “paralysis by analysis” (i.e. the deferment of decisions due to confusion and demoralisation) and so-called ‘maladaptation’ if the results are accepted without question, misinterpreted (i.e. treated as predictions or forecasts rather than projections) or used incorrectly.

An alternative way forward is the application of scenario planning and consequent identification of robust adaptive strategies. An adaptive strategy is robust if it works under a wide range of plausible climate futures. This makes it somewhat insensitive to model inaccuracies, the resolution of uncertainties and violated assumptions. Thus the focus of any impact, vulnerability and adaptation assessment should be preparedness, not prediction. The outcomes should include: quantification of planning, resourcing and construction lead times; means for the reduction of these lead times, and contingency plans that incorporate trigger points for alternative courses of action and their possible timing. To ensure successful adaptive planning, such assessments will need to be repeated into the future as new information becomes available.