When creating agent-based models, one question is how many agents to include and where are they located. Often we create synthetic individuals or households based on census data however, this can be a rather time consuming task. So a recent project from RTI has caught my attention. It is a U.S. wide geospatially explicit synthetic population funded under a grant from NIH/NIGMS.
The dataset provides a synthetic version of each household and person in the U.S. based on 2005-2009 ACS public use microdata and other sources. LandScan 90-meter night-time population distributions were used to place each household across the landscape (a total of 112,383,675).
The basic website location for information and to download the data is at: https://www.epimodels.org/midas/pubsyntdata1.do
Another short write up can be found here: http://www.openabm.org/forum/2852