Weather forecasters built a computer model of Botswana's rainfall patterns and factored in the lifecycle of the mosquito, which carries the parasite that causes malaria.

The model successfully predicted local epidemics of malaria from 1982 to 2002, with a lead time of up to five months.

Epidemics of malaria are typically triggered after heavy rainfall which forms puddles in which mosquitos breed.

These outbreaks account for only a small percentage for the estimated 500 million cases of malaria that occur each year, more than a million of which are fatal. Endemic malaria -- a term for a constant rate of incidence rather than a sudden rash of them -- accounts for the remainder.

However, epidemic malaria can inflict a inordinately high toll in some parts of Africa.

They also strain slender health budgets because of the sudden demand on hospitals and doctors and the rush to buy medicines, insecticides and mosquito nets.

Getting an early warning should save lives and money, for health watchdogs can time insecticide spraying and public-awareness campaigns more effectively.

The DEMETER model, devised by a team led by Tim Palmer of the European Centre for Medium Range Weather Forecasts, located in Reading, southern England, crunches data on atmospheric and sea temperatures and precipitation across southern Africa and the Indian and Atlantic Oceans.

Botswana was picked for the test as it is on the fringes of a desert and has a highly varied season of rainfall. Rain is concentrated from November to March but this can vary enormously from year to year.

Related Links
-