Earth observation for malaria modelling: a practical toolkit for satellite-based prediction of mosquito distributions using Google Earth Engine and R

This toolkit user-guide provides a user-friendly introduction to implementing the satellite-based analysis and random forest modelling to identify the key bio-geographical variables that influence mosquito distributions and abundance within the context of malaria studies. It is intended to be a resource for users with limited prior knowledge of analysis of this nature and presents step-by-step instructions for users to perform predictive modelling of mosquito distributions; sample datasets and analysis scripts are provided.

This guide uses three software packages, Google Earth Engine, R (with RStudio) and QGIS for data pre-processing, modelling and visualisation, all of which are cost-free for non-commercial use. Example scripts are provided to perform data processing and analysis in both Google Earth Engine (GEE) and R, and although these scripts are designed to automate the analysis to a large degree, they are currently optimised for the study area and time period used in the worked example for Lodja, Democratic Republic of the Congo. Users are free to adapt and build on this to suit their own purposes. By following the instructions, the user will learn to implement these methods. This will provide a basis for users to apply these methods to different datasets, areas and scenarios of their choosing.