Combining global and local estimates for spatial distribution of mosquito larval habitats

Malaria, transmitted by mosquitoes, is the leading cause of mortality in many African countries. This study assesses the spatial distribution of mosquito larval habitats on the highlands of western Kenya. Environmental factors such as terrain, land use, and surface waters are evaluated for their influence on the habitats. Logistic regression is used to model the global trend in the variation of larval habitats, while kriging is used to model the spatially dependent local variation. Results show notable improvement in modeling accuracy by combining the regression and kriging estimates over using the regression estimate alone.