The physical basis for atmospheric predictability on seasonal time scales resides primarily on the notion that slowly varying anomalous lower boundary forcing can have significant impact on atmospheric development. Such external forcing is generally thought to be associated with sea surface temperature (SST) anomalies, though the effect of long-lived anomalies in soil moisture and snow cover may also play a significant role. It is clearly important to be able to access where on the global atmospheric variations are sufficiently affected by oceanic forcing to enable practical seasonal prediction. This requires measurements of atmospheric potential predictability. It is commonly believed that lower boundary conditions dominate interannual variability in the tropics and major monsoon systems. Therefore, the atmospheric variability may be highly predictable in those regions. But is there any regional differences? For example, would one expect the circulation over the equatorial eastern Pacific to be more predictable than the North Pacific warm pool. Further, one might also question whether the short-term climate predictability is the same throughout the year. In particular, how does the annual cycle influence the relative impact of anomalous lower-boundary forcing?

An ensemble of ten 45-yr (1956-2000) ECHAM4 GCM integrations, forced throughout by the observed SST and sea-ice datasets, are used together to provide a global and regional assessment of potential seasonal predictability and its annual-cycle dependency. For the precipitation, the predictability in the tropical Indian Ocean and western Pacific is lower than the eastern Pacific. Spring rainfall is more predictable than the other season for both the Tropical Atlantic and NINO 3.4 regions. On the other hand, the summer precipitation predictability in the maritime continent is higher than the other season. For the mean sea-level pressure, NINO 3.4 region is more predictable in summer and autumn while the North Pacific warm pool is more predictable in winter and spring. For the 500 hPa geopotential height, all northern hemisphere mid-latitude sections (Asia, North Pacific, North America, North Atlantic and Europe) have lowest predictability in autumn. North Pacific and North America regions are with largest potential predictability in winter while it is more predictable in summer for the Asia region.