The structures of the links in the chain drought, haxvest failure, famine are examined. Possible interventions are feasible at the link between drought and harvest failure, to lessen or avert the impact. Such interventions, based upon knowledge of the climatology of drought-prone regions, may be discussed under three headings: short-term predictions, long-term predictions, and the analysis of spatial patterns of drought.
Basically, there are three climate-based methods of producing a shortterm (within growing-season} forecast of crop yield. The first is to use a water balance model to estimate soil moisture deficits. The second approach applies regression analysis techniques to the development of models of crop-climate relationships. The third is the possibility of long-range weather forecasts.
Long-term predictions rely on the fact that drought occurrence is a function of the balance between rainfall and evapotranspiration. It is possible to develop scenarios of future trends in these two variables, arising from both physical and anthropogenic forcing. Some potential forcing mechanisms axe discussed.
Finally, analysis of spatial patterns of drought at the national level will permit deeper insight into the phenomenon. This has potential benefits for the planning process, such as the identification of drought-prone axea~ and of spatial compensation zones (regions commonly drought-free when the principal grain-producing areas are affected).. Together, these three items form a comprehensive strategy for applied research into the physical and human factors underlying drought.