A three-tier overlapping prediction scheme: Tools for strategic and tactical decisions in the developing world

During the last decade our understanding of processes that determine the variability of the atmosphere and the climate system have improved to the extent that predictability of some phenomena has become established. The predictability, at least in the short-and long-term timescales, has on rare occasions been translated into useful predictions. However, the value of forecasts (i.e. how the forecasts are interpreted and their value to a user group) has not improved at the same rate. Arguably, the problem lies with the psychological and physical separation of the scientist or technician who makes the forecasts, and decision-makers and user communities who utilise the forecasts. Furthermore, the separation is exaggerated by the fact that for any one forecast there are many potential applications for different user communities each with unique needs. This added complexity makes it impossible in a practical sense for a forecaster to communicate with all users. Instead, for a given probabilistic forecast sets of decision tools need to be constructed for each user class. Clearly, this cannot be accomplished by a forecasting centre but requires input of intermediaries (organisational or personnel) and the user communities themselves. We describe one attempt to provide probabilistic forecasts over an overlapping set of timescales with decision models to bridge the forecaster-user separation. Such a model allows user communities to make interrelated longer-term strategic and shorter-term tactical decisions and to hedge in an iterative sense against uncertainty.