A two-stage cluster sampling method using gridded population data, a GIS, and Google EarthTM imagery in a population-based mortality survey in Iraq

Table 1 Overview of gridded population datasets currently available

Dataset	Provider(website)	Spatial resolution	Input population data source	Interpolation method	Ancillary data	Year(s)
GPWv3.0	CIESIN (http://sedac.ciesin.columbia.edu/gpw/)	2.5’(~5 km²)	UNPD census data	Areal weighting ¹	-None	1990,1995, 2000,2005 (projection),2010 (projection), 2015 (projection)
GRUMPv1	CIESIN (http://sedac.ciesin.columbia.edu/gpw/)	.5’(~1 km²)	UNPD census data	Dasymetric mapping ²	-Night-time light imagery-Populated places	2000
LandScan^TM	ORNL (http://www.ornl.gov/sci/landscan/)	.5’(~1 km²)	Population Division of the U.S. Census Bureau	Smart interpolation ³	-Land cover-Road networks-Digital elevation models-Slope-Satellite imagery	2008

Adapted from [27].
¹ Areal weighting overlays a grid onto sub national administrative unit population data and distributes the population across space according to the proportion of the administrative unit area that is contained within the grid cell [28].
² Dasymetric mapping disaggregates sub national population estimates into grid units using ancillary data such as road networks [28, 29].
³ Smart interpolation disaggregates sub national population estimates to grid cells according to likelihood co-efficients of population occurrence derived from ancillary data such as proximity to roads, slope, land cover [30].

ISSN: 1476-072X