In recent years, the development of high-speed computers and global meteorological monitoring by satellites has led to rapid improvements in forecast capability on synoptic to seasonal time scales. Short and medium range (1 to 14 day) forecasts are based almost entirely on what is termed dynamical predictability, i.e., prescription of atmospheric fluctuations based on 'internal' forcing through such mechanisms as flow instabilities, non-linear interactions, thermal and orographic forcing, fluctuating zonal winds and tropical/extratropical interactions. The limit of dynamical predictability appears to be about two weeks for synoptic scale flows and about one month for the longer planetary waves (wave numbers 0-4). Atmospheric variability is driven not only by internal dynamics but also by boundary forcing, which is related to external factors such as soil moisture, vegetation, sea-surface temperatures (SSTs), snow and ice cover. The surface is particularly influential in the tropics, where boundary forcing overshadows internal dynamics as a mechanism of interannual variability and the relatively stable major circulation features (Hadley and Walker cells) are prescribed to a large extent by land/sea contrast. Surface forcing is variable in time and space, but on time scales considerably longer than the forcing introduced by internal dynamics and its inclusion in prognostic schemes could improve predictability (Shukla, 1981;Miyakoda, 1982). The persistence (often six months or longer) of the relevant surface properties provides the forecast potential.This has long been recognized as far as the ocean is concerned and SSTs have been used to predict seasonal variations (see for example, the review by Namias and Cayan, 1981). Until recently, however, the role of the land surface has nearly been neglected. The failure to recognize its importance in weather prediction and climate variability can be partially attributed to the inadequate knowledge on the part of atmospheric scientists as to the physical functioning of the surface and also to the lack of adequate data bases for determining the appropriate surface boundary conditions at a given time and place (Dickinson, 1983;Miyakoda, 1982). Now, however, several major efforts are underway to promote research on the relationship between the land surface and climate and to incorporate surface and subsurface processes, including the role of vegetation, at TU Muenchen on July 11, 2015 ppg.sagepub.com Downloaded from