Climate change exaggerates current pressures in water management—adding to the debate on sound management strategies—and adds a new component. This new component relates to uncertainty in climate change: How can water management efficiently adapt to climate change, given that the magnitude (or possibly even the direction) of change is not known? Conventionally, water resource managers assume that the future resource base will be the same as that of the past and therefore that estimates of indices such as average reservoir yield or probable maximum flood that are based on past data will apply in the future. There are two issues: assessing alternatives in the face of uncertainty and making decisions on the basis of this assessment.

Techniques for assessing alternatives include scenario analysis and risk analysis. Scenario analysis is central to climate change impact assessment, but it is not widely used in water resource assessment (although there are some very important exceptions, such as at the federal level in the United States). Scenario analysis, as in climate change impact assessment, tends to involve simulation of the effects of different scenarios, although in water resources assessment these tend to be different demand and operational scenarios rather than different climate scenarios. Stakhiv (1998) argues that if water managers already adopt a scenario-based approach, as at the federal level in the United States (Lins and Stakhiv, 1998), climate change therefore does not cause any additional conceptual challenges to water management: Climate change can be regarded simply as an extra type of scenario. However, the uncertain nature of climate change and the potential for nonlinearities in impact mean not only that the range of scenarios conventionally considered may be too narrow but also that a larger number of scenarios must be evaluated. In practice, scenario-based approaches are used in few water management agencies, and adoption of scenario analysis would challenge conventional water management practices in many countries.

Risk analysis involves assessment of the risk of certain thresholds being crossed under different possible futures (Major, 1998). It generally involves stochastic simulation of hydrological data to develop a sampling distribution of possible futures. In principle, climate change can be incorporated into risk analysis by changing the underlying population from which data are generated according to climate change scenarios. Matalas (1997) discusses the role of stochastic simulation in the context of climate change and argues that given the wide range in futures that often is simulated by assuming a stationary climate, the operational assumption of stationarity may remain appropriate in the face of climate change in some regions. However, it is possible that climate change could generate futures outside those produced under stationarity, and it cannot be assumed that climate change can be ignored in all circumstances.

The second main issue is that of decisionmaking under uncertainty. This issue was widely investigated during the 1960s and 1970s, largely in the context of uncertainties about demands or the precise distribution of floods and droughts over the short and medium terms. Climate change has revived interest in decisionmaking under uncertainty, and several analyses of different techniques have been published (e.g., Fisher and Rubio, 1997; Frederick, 1997; Hobbs, 1997; Hobbs et al., 1997; Luo and Caselton, 1997; Chao et al., 1999). There still is considerable debate. Hobbs (1997), for example, concludes that Bayesian approaches involving allocation of probabilities to specific outcomes are more suitable than Dempster-Shafer reasoning (which requires the analyst to assign probabilities to ranges—perhaps overlapping—of outcomes), but Luo and Caselton (1997) conclude the reverse. Particularly significant is the issue of assigning probabilities to alternative possible futures. Hobbs et al. (1997) note unease among water planners in assigning subjective probabilities to different futures.

Table 4-14: Headroom “score” characterizing effect of climate change on resource zone yield: an approach used in UK (UKWIR, 1998).
Range in Resource Zone Yield between Four Defined Scenarios a	Case 1: Two Scenarios Above and Two Below Mean	Case 2: Three Scenarios Below and One Above Mean	Case 3: Three Scenarios Above and One Below Mean
<15%	2	3	1
15–25%	4	6	2
25–35%	6	9	3
>35%	8	10	4
a As percentage of “best estimate” of yield.

Planners of water resource and flood protection schemes conventionally cope with uncertainty by adding a safety factor to design estimates. This safety factor usually is defined arbitrarily. As part of a review of water resource design practices in the UK, a more formal approach to calculation of this safety factor, or “headroom,” has been developed (UKWIR, 1998). This procedure identifies eight sources of supply-side uncertainty and three sources of demand-side uncertainty, each of which is given a score. The total score is summed and converted into a percentage value for the headroom allowance (with a maximum of 20%). Climate change is included as one of the supply-side uncertainties; its score depends on the range of estimates of supply-yield under four defined climate change scenarios (Table 4-14). Although this approach has many arbitrary elements, it does represent a systematic approach to the treatment of climate change uncertainties in water resources assessment.

Different aspects of the water sector have different planning horizons and infrastructure lifetimes. The parts of the water sector with long horizons and lifetimes need to take a different approach to climate change than parts with shorter lead times; one assessment and decision methodology will not be suitable for all managers.

IPCC Homepage