GCM models suggest an increase in global-mean precipitation, as one might expect from the associated increase in atmospheric temperature. However, because the spatial variability of rainfall is much greater than for surface temperature, regional and local details of changes are highly uncertain (IPCC, 1990a). Instrumental data from which long-term changes in precipitation can be determined are only available over land areas (Bradley et al., 1987; IPCC, 1990a), and this data suffers from the problems of incomplete coverage and inhomogeneity. With the likelihood that the precipitation signal-to-noise ratio is low, any meaningful comparison between observation and model, and therefore attribution, is precluded. In addition, since global precipitation is partially dependent upon global temperature, rises in global precipitation may be expected as a consequence of other causes of climate change.