When deciding based on the future situation, standard methods for deciding under uncertainty such as CBA or CEA cannot be applied, as these require to compute expected utility or expected effects for each option. In contrast to deciding only on current extreme event risk, this is not possible in principle for long-term adaptation decisions as probability distributions cannot be attained across future emissions and hence across climate and impacts variables as these depend on a range of policy and social choices (such as, e.g., the future level of greenhouse gas emissions), which are not predictable in terms of quantifiable probabilities (Lempert and Schlesinger 2001; Hallegate 2009).

Alternative methods have been developed to address such decision making under deep uncertainty (i.e. without probabilities). These may be summarised under the label robust decision making (e.g., Lempert and Schlesinger 2000; Lempert and Collins 2007), although a clear cut terminology has not (yet) been established. Here, only the formal appraisal stages of these methods are described and not the overall decision making process that may also include participation.

Considering this, two types of methods can be distinguished based on whether they appraise options using the criterion of robustness alone or the criterion of robustness and flexibility. An option is flexible if it allows to switch to other options that might be preferable in the future once more is known about the changing climate. For example, in an aquifer under increasing water scarcity an adaptation option of demand management through water market credits is a flexible option, as this option can be abandoned and greater storage capacity infrastructure can be built at any point in the future. Building in storage capacity is less flexible because of large upfront investment costs meaning that abandoning it in the future will not significantly reduce the cost of the option.

The decision node for choosing between these two types of methods concerns whether the full set of options includes a flexible one.

When none of the options is flexible, then formal appraisal methods can focus on the criterion of robustness, and a one-shot robust decision making method is appropriate . An option is robust, if it is effective over the full or a large range of scenarios (Lempert and Schlesinger 2000; Lempert and Collins 2007). See Wilby and Dessai (2010) for an application to water management in the UK and Lempert et al. (2012) for an application to infrastructure investment decisions the Port of Los Angeles is facing in the context of future sea-level rise.

When at least one option is flexible, the criterion of flexibility may be considered as well. The general strategy thereby is to favour flexible options over non flexible ones and through this keep future options open (Hallegate 2009). The adaptation pathways method, for example, does so by characterizing options in terms of two attributes: i) adaptation turning points (ATP), which are points beyond which options are no longer effective (Kwadijk et al. 2010), and ii) what alternative options are available once a turning point has been reached (Haasnoot et al. 2012). Importantly, the exact time when an ATP is reached does not matter, it is rather the flexibility of having alternative options available. Prominent applications of this approach include the Thames Estuary 2100 Plan (Lowe et al. 2009; Penning-Roswell et al. 2011), the Dutch Delta Programme (Kabat et al. 2009) and the New York City Panel on Climate Change (Rosenzweig et al. 2011).

Table 2.8 supplements the analysis provided in the decision tree by presenting these characteristics and their indication on the critical task through several examples.Formal decision making methods and examples are described in the Toolbox section on Formal decision-analysis.

Table 2.8: Application of formal decision making methods to examples based on the criteria listed above.