2010 Articles

Report on the forecast quality of seasonal predictions of malaria incidence in Botswana over the period 1982-2005

Connor, S. J.; Curtis, A.; Mason, S.; Thomson, M.; Tippett, M.

The ability of Stream 2 ENSEMBLES models to predict climate variability that is known to be related to annual variability in malaria over Botswana is tested. The ENSEMBLES multi-model is shown to have no clear difference in predictive skill compared to that for DEMETER models, but an ability to distinguish between high- and low-malaria years remains evident. More detailed analyses of the influence of observed climate variability on malaria incidence in Botswana were conducted. Evidence for district-scale climate signals is weak, as is an influence of seasonal temperature variability on malaria incidence, and so the ability of the ENSEMBLES models to predict these aspects was not considered. Real-time forecasts were communicated to the region during the duration of the project.