Seasonal and Subseasonal Forecast Applications on Climate and Malaria in West Africa

Overcoming the serious public health burden of malaria in sub-Saharan Africa requires a detailed understanding of the epidemiology of malaria in the region. In recent years, skill improvement of seasonal and sub-seasonal dynamical prediction has provided opportunities to use these forecasts to drive dynamical malaria models to issue early warnings of malaria epidemics in Africa. In this study, we are documenting the influence of climate on malaria outbreaks using in situ measurements, satellite observations, and model data from daily to seasonal time scales. Several steps are taken in this study. We first analyze the malaria cases from the National Malaria Control Programme of Senegal (NMCP) and from different malaria locations across African countries via the Malaria Atlas Project (MAP) to validate the models. Then, we run two malaria models, the 2010 version of the Liverpool Malaria Model (LMM2010) and the VECtor-borne disease community model of the international centre for theoretical physics, TRIeste (VECTRI). With these 2 malaria models, we simulate hindcasts of malaria incidence, using the CPC/NOAA daily rainfall and 2m temperature data as inputs. The goal is to determine the accuracy of both models in simulating seasonal malaria transmission in West Africa and Senegal in particular. Results suggest that both models agree with observations on the unimodal shape of malaria distribution. However, transmission peaks in the models tend to be delayed by one to two months in the study area. Models further agree that the seasonal malaria transmission contrast is closely linked to the latitudinal variation of rainfall. While the rainfall season is at its peak in July-August-September in West Africa, the peak of the malaria outbreak season occurs in September-October-November. These results are consistent with previous findings. However, the non-linear nature of the malaria models can lead to large differences between output simulations. These are discussed. Next, we study the predictability of malaria transmission. We employ the Canonical Correlation Analysis (CCA) and the Statistical Seasonal Forecast (S4CAST), where the predictand is rainfall or temperature extracted from meteorological stations, satellite, and reanalysis, or the malaria models’ outputs; and the predictors are the observed SST (ERSSTv4) and the North Americcan Multi-model Ensemble (NMME) predicted SST. Results are presented.

Keywords: Malaria, Climate, Forecast, Predictability, Senegal, West Africa