High-resolution climate change scenarios for North America

Projections of future climate produced by General Circulation Models (GCMs) have a coarse spatial resolution. As such they often require further processing before being used for climate change impact studies. There are several approaches to generating these finer scale or higher resolution models. They include dy-namical downscaling (e.g. Regional Climate Models- RCMs), statistical downscaling and statistical inter-polation. RCMs are higher resolution climate change models driven by the conditions imposed by a “host” GCM (e.g., Laprise et al. 2003). Statistical downscaling methods correlate large-scale atmospheric processes to local scale meteorology and apply this information to GCM projections of future weather patterns to char-acterize future local climate (Wilby et al. 1998). Such approaches attempt to maintain physical consistency with the GCM representations of atmospheric proc-esses, and can generate distributions (e.g. means and extremes) of simulated meteorological data usually at point locations. However both RCM and statistical downscaling are highly computationally intensive. Statistical interpolation of GCM output is a simpler approach to obtaining higher resolution estimates from GCM models. Although lacking the physical detail of dynamical downscaling, these methods provide outputs that cover large regions and can capture the climate change signals simulated by the GCM. This approach can provide spatially and temporally consistent outputs useful for landscape-scale climate change studies (see also discussion in Houser et al., 2004). With this approach, the GCM data are normalized (i.e. changes are scaled relative to the GCM values for a baseline period, for example 1961-1990). This removes potential bias to make the scenario broadly consistent with actual historical observations. The overall approach is generally transparent and rapid. We have developed a process that implements a statis-tical interpolation approach to generating high-resolu-tion models (grids) from GCMs. We have applied the approach to several GCMs with results that cover all of North America. Our aim is to support landscape-scale climate change impact studies. This note briefly describes the process and products available