Improving intelligent dasymetric mapping population density estimates at 30 m resolution for the conterminous United States by excluding uninhabited areas

Population change impacts almost every aspect of global change from land use, to greenhouse gas emissions, to biodiversity conservation, to the spread of disease. Data on spatial patterns of population density help us understand patterns and drivers of human settlement and can help us quantify the exposure we face to natural disasters, pollution, and infectious disease. Human populations are typically recorded by national or regional units that can vary in shape and size. Using these irregularly sized units and ancillary data related to population dynamics, we can produce high-resolution gridded estimates of population density through intelligent dasymetric mapping (IDM). The gridded population density provides a more detailed estimate of how the population is distributed within larger units. Furthermore, we can refine our estimates of population density by specifying uninhabited areas which have impacts on the analysis of population density such as our estimates of human exposure. In this study, we used various geospatial datasets to expand the existing specification of uninhabited areas within the United States (US) Environmental Protection Agency's (EPA) EnviroAtlas Dasymetric Population Map for the conterminous United States (CONUS). When compared to the existing definition of uninhabited areas for the EnviroAtlas dasymetric population map, we found that IDM's population estimates for the US Census Bureau blocks improved across all states in the CONUS. We found that IDM performed better in states with larger urban areas than in states that are sparsely populated. We also updated the existing EnviroAtlas Intelligent Dasymetric Mapping toolbox and expanded its capabilities to accept uninhabited areas. The updated 30m population density for the CONUS is available via the EPA's Environmental Dataset Gateway