A forecasting model for desert locust presence in West and North Africa

The set of scripts provided in this dataverse contains most of the steps and data necessary to create and reproduce the models, as well as the software developped to produce automatic forecasting maps. Regarging model training, evaluation and selection, the scripts are written in R and are numbered from 01 to 07 (in LocustForecastCLCPRO4/RCodes). The software part depicting the automatic chain of : real-time downloading satellite image (steps 00 all in SatData-main), real-time prediction of locust distribution based on selected model (step 08 in PyhtonCodes), updating the forcasting map in a web interface (step 09 in mppcpro-main), correspond mostly to pyhton scripts. The folder SatData-main contains the pythons files for automated downloading satellite image from MODIS, VIIRS and Copernicus. The LocustForecastCLCPRO4/Rcodes folder contains all the scripts to build the database and models. All files inside the PythonCodes are meant for real-time model predictions. The mppcpro-main folder holds the R and Python scripts used for web interface.