popRF: Random Forest-informed Disaggregative Population Modelling and Mapping

Here we introduce the popRF package in R that largely addresses these issues. This is done by functionalising the RF-informed dasymetric population modelling procedure (F. R. Stevens et al. 2015) in a single language that is completely free, open source, and environment agnostic. Further, the package has been parallelised where possible to achieve efficient prediction and geoprocessing over large extents, providing functions that have applied utility outside of simply performing disaggregative population modelling. This package was utilised already to predict population and inform the mapping of modelled human settlement (Nieves, Sorichetta, et al. 2020; Nieves, Bondarenko, et al. 2020; Nieves et al. 2021) at 100m resolution across 249 countries from 2000-2020, ingesting over 10TB of covariates (Lloyd et al. 2019) and producing another 70 TB of population and population related dataset