The climate change impacts community has long bemoaned the inadequate spatial scale of climate scenarios produced from coarse resolution GCM output (Gates, 1985; Lamb, 1987; Robinson and Finkelstein, 1989; Smith and Tirpak, 1989; Cohen, 1990). This dissatisfaction emanates from the perceived mismatch of scale between coarse resolution GCMs (hundreds of kilometres) and the scale of interest for regional impacts (an order or two orders of magnitude finer scale) (Hostetler, 1994; IPCC, 1994). For example, many mechanistic models used to simulate the ecological effects of climate change operate at spatial resolutions varying from a single plant to a few hectares. Their results may be highly sensitive to fine-scale climate variations that may be embedded in coarse-scale climate variations, especially in regions of complex topography, along coastlines, and in regions with highly heterogeneous land-surface covers.
Conventionally, regional "detail" in climate scenarios has been incorporated by applying changes in climate from the coarse-scale GCM grid points to observation points that are distributed at varying resolutions, but often at resolutions higher than that of the GCMs (e.g., see Box 13.1; Whetton et al., 1996; Arnell, 1999). Recently, high resolution gridded baseline climatologies have been developed with which coarse resolution GCM results have been combined (e.g., Saarikko and Carter, 1996; Kittel et al., 1997). Such relatively simple techniques, however, cannot overcome the limitations imposed by the fundamental spatial coarseness of the simulated climate change information itself.
Three major techniques (referred to as regionalisation techniques) have been developed to produce higher resolution climate scenarios: (1) regional climate modelling (Giorgi and Mearns, 1991; McGregor, 1997; Giorgi and Mearns, 1999); (2) statistical downscaling (Wilby and Wigley, 1997; Murphy, 1999); and (3) high resolution and variable resolution Atmospheric General Circulation Model (AGCM) time-slice techniques (Cubasch et al., 1995; Fox-Rabinovitz et al., 1997). The two former methods are dependent on the large-scale circulation variables from GCMs, and their value as a viable means of increasing the spatial resolution of climate change information thus partially depends on the quality of the GCM simulations. The variable resolution and high resolution time-slice methods use the AGCMs directly, run at high or variable resolutions. The high resolution time-slice technique is also dependent on the sea surface temperature simulated by a coarser resolution AOGCM. There have been few completed experiments using these AGCM techniques, which essentially are still under development (see Chapter 10, Section 10.4). Moreover, they have rarely been applied to explicit scenario formation for impacts purposes (see Jendritzky and Tinz, 2000, for an exception) and are not discussed further in this chapter. See Chapter 10 for further details on all techniques.
Other reports in this collection