Simple climate models are simplified global models that attempt to reproduce the large-scale behaviour of AOGCMs (see Chapter 9). While they are seldom able to represent the non-linearities of some processes that are captured by more complex models, they have the advantage that multiple simulations can be conducted very rapidly, enabling an exploration of the climatic effects of alternative scenarios of radiative forcing, climate sensitivity and other parametrized uncertainties (IPCC, 1997). Outputs from these models have been used in conjunction with GCM information to develop scenarios using pattern-scaling techniques (see Section 13.5). They have also been used to construct regional greenhouse gas stabilisation scenarios (e.g., Gyalistras and Fischlin, 1995). Simple climate models are used in climate scenario generators (see Section 13.5.2) and in some integrated assessment models (see Section 13.6).
Three additional types of climate scenarios have also been adopted in impact studies. The first type involves extrapolating ongoing trends in climate that have been observed in some regions and that appear to be consistent with model-based projections of climate change (e.g., Jones et al., 1999). There are obvious dangers in relying on extrapolated trends, and especially in assuming that recent trends are due to anthropogenic forcing rather than natural variability (see Chapters 2 and 12). However, if current trends in climate are pointing strongly in one direction, it may be difficult to defend the credibility of scenarios that posit a trend in the opposite direction, especially over a short projection period.
A second type of scenario, which has some resemblance to the first, uses empirical relationships between regional climate and global mean temperature from the instrumental record to extrapolate future regional climate on the basis of projected global or hemispheric mean temperature change (e.g. Vinnikov and Groisman, 1979; Anisimov and Poljakov, 1999). Again, this method relies on the assumption that past relationships between local and broad-scale climate are also applicable to future conditions.
A third type of scenario is based on expert judgement, whereby estimates of future climate change are solicited from climate scientists, and the results are sampled to obtain probability density functions of future change (NDU, 1978; Morgan and Keith, 1995; Titus and Narayanan, 1996; Kuikka and Varis, 1997; Tol and de Vos, 1998). The main criticism of expert judgement is its inherent subjectivity, including problems of the representativeness of the scientists sampled and likely biases in questionnaire design and analysis of the responses (Stewart and Glantz, 1985). Nevertheless, since uncertainties in estimates of future climate are inevitable, any moves towards expressing future climate in probabilistic terms will necessarily embrace some elements of subjective judgement (see Section 13.5).
Other reports in this collection