3.6.5 Uncertainty and risk: decision-making under uncertainty
Climate change poses a major conceptual challenge to water managers, in addition to the challenges caused by population and land-use change. It is no longer appropriate to assume that past hydrological conditions will continue into the future (the traditional assumption) and, due to climate change uncertainty, managers can no longer have confidence in single projections of the future. It will also be difficult to detect a clear climate-change effect within the next couple of decades, even with an underlying trend (Wilby, 2006). This sub-section covers three issues: developments in the conceptual understanding of sources of uncertainty and how to characterise them; examples of how water managers, in practice, are making climate change decisions under uncertainty; and an assessment of different ways of managing resources under uncertainty.
The vast majority of published water resources impact assessments have used just a small number of scenarios. These have demonstrated that impacts vary among scenarios, although temperature-based impacts, such as changes in the timing and volume of ice-melt-related streamflows, tend to be more robust (Maurer and Duffy, 2005), and the use of a scenario-based approach to water management in the face of climate change is therefore widely recommended (Beuhler, 2003; Simonovic and Li, 2003). There are, however, two problems. First, the large range for different climate-model-based scenarios suggests that adaptive planning should not be based on only a few scenarios (Prudhomme et al., 2003; Nawaz and Adeloye, 2006): there is no guarantee that the range simulated represents the full range. Second, it is difficult to evaluate the credibility of individual scenarios. By making assumptions about the probability distributions of the different drivers of climate change, however, it is possible to construct probability distributions of hydrological outcomes (e.g., Wilby and Harris, 2006), although the resulting probability distributions will be influenced by the assumed initial probability distributions. Jones and Page (2001) constructed probability distributions for water storage, environmental flows and irrigation allocations in the Macquarie River catchment, Australia, showing that the estimated distributions were, in fact, little affected by assumptions about probability distributions of drivers of change.
Water managers in a few countries, including the Netherlands, Australia, the UK, and the USA, have begun to consider the implications of climate change explicitly in flood and water supply management. In the UK, for example, design flood magnitudes can be increased by 20% to reflect the possible effects of climate change (Richardson, 2002). The figure of 20% was based on early impact assessments, and methods are under review following the publication of new scenarios (Hawkes et al., 2003). Measures to cope with the increase of the design discharge for the Rhine in the Netherlands from 15,000 to 16,000 m3/s must be implemented by 2015, and it is planned to increase the design discharge to 18,000 m3/s in the longer term, due to climate change (Klijn et al., 2001). Water supply companies in England and Wales used four climate scenarios in their 2004 review of future resource requirements, using a formalised procedure developed by the environmental and economic regulators (Arnell and Delaney, 2006). This procedure basically involved the companies estimating when climate change might impact upon the reliability of supply and, depending on the implementation of different actions, when these impacts would be felt (in most cases estimated effects were too far into the future to cause any changes in practice now, but in some instances the impacts would be soon enough to necessitate undertaking more detailed investigations now). Dessai et al. (2005) describe an example where water supply managers in Australia were given information on the likelihood of drought conditions continuing, under different assumptions about the magnitude of climate change. They used this information to decide whether to invoke contingency plans to add temporary supplies or to tighten restrictions on water use.
A rather different way of coping with the uncertainty associated with estimates of future climate change is to adopt management measures that are robust to uncertainty (Stakhiv, 1998). Integrated Water Resources Management, for example, is based around the concepts of flexibility and adaptability, using measures which can be easily altered or are robust to changing conditions. These tools, including water conservation, reclamation, conjunctive use of surface and groundwater, and desalination of brackish water, have been advocated as a means of reacting to climate change threats to water supply in California (e.g., Beuhler, 2003). Similarly, resilient strategies for flood management, such as allowing rivers to temporarily flood and reducing exposure to flood damage, are preferable to traditional ‘resistance’ (protection) strategies in the face of uncertainty (Klijn et al., 2004; Olsen, 2006).