Changing populations are often accompanied by changing built-settlement landscapes. Here, small area population data and a limited set of environmental covariates have been combined with machine learning methods and dynamically-limited growth curves to annually interpolate (from 2000 to 2014) and annually project (from 2015 to 2020) the presence of built-settlements across the globe at 100m resolution. These annual built-settlement maps were then used to inform the WorldPop "Global per country 2000-2020" population datasets. An overview of the built-settlement growth modeling framework can be found in Nieves et al.