SATELLITE IMAGERY, TSETSE AND TRYPANOSOMIASIS IN AFRICA

This paper describes the application of remote sensing to studies on the African trypanosomiases causing sleeping sickness in man and 'nagana' in domestic animals. After giving some biological background to the problem, an important relationship between the risk of infection of domestic animals, tsetse fly numbers and fly infection rates is presented. An understanding of the latter leads to a prediction of risk in cattle. The problems of analysing and interpreting the distribution and abundance patterns of flies are explored, and a mortality climogram approach is described in which the important meteorological variables appear to be temperature and saturation deficit. This approach, which can be applied to data collected at only very few sites, leads to extensive predictions of the distributional limits of the tsetse Glossina morsitans Westwood. These predictions are supported by the known distribution of this species throughout Africa. A similarity noted between published whole-Africa normalised difference vegetation indices (NDVI) and tsetse distribution leads to an exploration of the usefulness of NDVI for such vector studies. The likely information content of the images is identified through principal component analysis, and correlations are shown between NDVI values and annual temperature, saturation deficit and rainfall figures for more than 300 sites throughout the sub-Saharan region of the continent. NDVI values are highly correlated with both saturation deficit and rainfall figures. Significant correlations are shown between vector mortality rates and mean monthly NDVI values, and between the physical size of the vectors (known to reflect the mortality rates affecting the parental population) and NDVI along an approximately 700 km transect across a whole range of eco-climate conditions in West Africa. Reasons are suggested for the localisation of human sleeping sickness to just one of several regions sampled by the transect. Finally, seasonal changes in case numbers of human sleeping sickness in both Uganda and Kenya are correlated with mean monthly NDVI from the areas concerned. The first of these correlations is negative, the second positive. Explanations for this difference are given in terms of the different vector species and epidemiological situations in these long-standing foci in East Africa. It is concluded that the rather limited amount of information available in NDVI has already proved enormously useful, and a plea is made for a more complete exploration of National Oceanic and Atmospheric Administration (NOAA) data, especially at the highest available resolution, whilst not neglecting the vitally important ground-based studies that made possible the results presented in this paper.