Developments in statistical downscaling since TAR (2001): North America

Since the TAR there have been numerous statistical downscaling (SD) studies but several important challenges remain largely unresolved (Leung et al., 2003). Foremost are questions surrounding the characterisation of predictor-predictand relationships, and the extent to which more sophisticated techniques can extract greater regional predictability. A significant fraction of articles were devoted to model intercomparison (principally statistical versus statistical methods); others highlighted the synergy between techniques used for statistical downscaling and those used for seasonal prediction. Although a few novel applications have emerged, regional climate change projections by SD methods continue to be most widely applied to the water resource, agricultural and conservation sectors. However, a handful of integrated assessments have begun to appear.