Potential Predictability of Malaria in Africa Using ECMWF Monthly and Seasonal Climate Forecasts

Abstract

Idealized model experiments investigate the advance warning for malaria that may be presently possible using temperature and rainfall predictions from state-of-the-art operational monthly and seasonal weatherprediction systems. The climate forecasts drive a dynamical malaria model for all of Africa, and the predictions are evaluated using reanalysis data. The regions and months for which climate is responsible for significant interannual malaria transmission variability are first identified. In addition to epidemic-prone zones these also include hyperendemic regions subject to high variability during specific months of the year, often associated with the monsoon onset. In many of these areas, temperature anomalies are predictable from 1 to 2 months ahead, and reliable precipitation forecasts are available in eastern and southern Africa 1 month ahead. The inherent lag between the rainy seasons and malaria transmission results in potential predictability in malaria transmission 3-4 months in advance, extending the early warning available from environmental monitoring by 1-2 months, although the realizable forecast skill will be less than this because of an imperfect malaria model. A preliminary examination of the forecasts for the highlands of Uganda and Kenya shows that the system is able to predict the years during the last two decades in which documented highland outbreaks occurred, in particular the major event of 1998, but that the timing of outbreaks was often imprecise and inconsistent across lead times. In addition to country-level evaluation with district health data, issues that need addressing to integrate such a climate-based prediction system into health-decision processes are briefly discussed.

Assessing the effects of air temperature and rainfall on malaria incidence: an epidemiological study across Rwanda and Uganda

“…It can be observed from the figure that four out the seven Rwanda districts where weather was unable to improve the model's predictive ability correspond to endemic districts where transmission is highest. This finding is expected because, as suggested by Lowe et al (2013) and Tompkins and Di Giuseppe (2015), climate variability may be less likely to impact malaria transmission in endemic districts where climate is conducive most of the year and variations in incidence may be mainly due to control interventions or data errors.…”

“…In other words, transmission is always underway by January in this zone, while transmission in December only occurs in years in which the rains begin unusually early in October and early November. Thus, even in areas where malaria is mesoendemic or hyperendemic, malaria transmission can still be highly variable in certain months of the year, implying potential usefulness of malaria early warning systems (Tompkins and Di Giuseppe, 2015). A conspicuous downward trend in malaria incidence is observed over the period 2006-2010 in some areas of Rwanda.…”

Evaluation of the ECMWF Sub-seasonal to Seasonal Precipitation Forecasts during the Peak of West Africa Monsoon in Nigeria

“…However, in recent years, only few studies have considered the abilities of these S2S models to reproduce basic atmospheric variables. Among these few studies are Tompkins and Feudale (2010), Lynch et al (2014), Tompkins and Giuseppe (2015), and White et al (2015). For example, in Australia, a version of S2S model has been used in disaster risk reduction (DRR) activities as well as emergency management and response in Australia.…”

“…Nevertheless, they showed that the model skill at predicting and or simulating West African summer monsoon rainfall anomalies has increased in recent years, thereby indicating improvements since the 1990s. Another type of S2S model, an ECMWF-S2S, was utilized by Tompkins and Giuseppe (2015) to investigate the use of temperature and rainfall predictions for advanced warning on malaria in an idealized experiment. Tompkins and Giuseppe (2015) found that ECMWF-S2S is capable of predicting the years, during the last two decades, in which documented Ugandan and Kenyan highland malaria outbreaks occurred.…”

“…Another type of S2S model, an ECMWF-S2S, was utilized by Tompkins and Giuseppe (2015) to investigate the use of temperature and rainfall predictions for advanced warning on malaria in an idealized experiment. Tompkins and Giuseppe (2015) found that ECMWF-S2S is capable of predicting the years, during the last two decades, in which documented Ugandan and Kenyan highland malaria outbreaks occurred. Nigerian NWHC is on the verge of employing ECMWF-S2S model for its operational sub-seasonal forecast activities.…”