Dasymetric population mapping based on U.S. Census data and 30-m gridded estimates of impervious surface

Authors: SWANWICK, RACHEL - National Socio-Environmental Synthesis Center; Read, Quentin; GUINN, STEVEN - University Of Maryland; WILLIAMSON, MATT - Boise State University; HONDULA, KELLY - National Ecological Observatory Network (NEON); ELMORE, ANDREW - University Of Maryland

Submitted to: Scientific Data
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/1/2022
Publication Date: 8/27/2022
Citation: Swanwick, R., Read, Q.D., Guinn, S., Williamson, M., Hondula, K.L., Elmore, A. 2022. Dasymetric population mapping based on U.S. Census data and 30-m gridded estimates of impervious surface. Scientific Data. 9:523. https://doi.org/10.1038/s41597-022-01603-z.
DOI: https://doi.org/10.1038/s41597-022-01603-z

Interpretive Summary: This is a summary of a data product we created that combines U.S. Census data with land cover data to give a high-resolution estimate of where people live in the continental U.S. We used Census data from block groups (the smallest unit of size that the Census provides full population data for) and data on paved surfaces from the National Land Cover Database to make a thirty-meter grid of population covering the U.S. This data product could be used to better estimate vulnerability of populations to environmental hazards, among other possible uses. We describe how the dataset was created and present a case study where we compare our data product to similar ones created by the U.S. EPA, Facebook, and Microsoft. We show that our dataset gives similar estimates for the proportion of people exposed to wildfire and flood risk, using a much simpler and easier to reproduce method than the other data products.

Technical Abstract: Assessment of socio-environmental problems and the search for solutions often require intersecting geospatial data on environmental factors and human population densities. In the United States, Census data is the most common source for information on population. However, timely acquisition of such data at sufficient spatial resolution can be problematic, especially in cases where the analysis area spans urban-rural gradients. With this data release, we provide a 30-m resolution population estimate for the entire contiguous United States. The workflow dasymetrically distributes Census block level population estimates across all non-transportation impervious surfaces within each Census block. The methodology is updatable using the most recent Census data and remote sensing-based observations of impervious surface area. The dataset compares favorably against other population data sources and provides a useful balance between resolution and complexity.