A high resolution population grid for the conterminous United States: The 2010 edition

Readily available high resolution data on population distribution is an important resource for monitoring human-environment interactions and for supporting planning and management decisions. Using a grid that approximates population density over the entire country seems like the most practical approach to exploring and distributing detailed population data but instead data based on census aggregation units is still the most widely used method. In this paper we describe the construction of 30 m resolution grid representing the distribution of population in 2010 over the entire conterminous United States. The grid is computed using 2010 U.S. Census block level population counts disaggregated by a dasymetric model that uses land cover (2011 NLCD) and land use (2010 NLUD) as ancillary data. Detailed descriptions of the ancillary data and dasymetric model are given. Methods of computing the grid are presented followed by an extensive assessment of model accuracy. Overall the expected value for relative error of the model is 44% which is at the lower limit of errors reported for other continental-sized, high resolution population grids. We also offer a more specific error estimate for areas with specified value of population density. Using two example areas, one highly urbanized and another rural, we demonstrate the advantages of using the gridded population data over the census block-based data. Our 30 m population grid is available for online exploration and for download from the custom-made GeoWeb application SocScape at http://sil.uc.edu.