Trouble with Census Data

[RachelSuhs]

I am doing project calculating displacement due to sea level rise (SLR). I am doing so by using intersecting the SLR polygons with census tracts for each state and then running zonal statistics as table with the intersect shapefile and a Gridded Population raster. However, when I run the zonal stats as table tool, I get higher rates of displacement than is reasonable. For example, one census tract has a total population of ~8,000 but the calculated displacement for that tract is ~12,000. Is there a reason for why I am getting higher rates of displacement than I should be? Or is there a better method I should use? Thanks

[JoeBorgione]

Gridded Population raster...

I would double check your population data;  For the tract that is giving you the problem, how is the 8,000 population derived?  The only population  data I've ever worked with is in polygon form, so the assumption  that the population is smoothly distributed across the polygon has to be made.  Does your gridded population  raster  have cell to cell values that would indicate density variability?

[RachelSuhs]

The 8,000 population is derived from current census data estimates and is in polygon form. The gridded population raster does have density variability if I am understanding your question right. The gridded population raster can be found here Data » Gridded Population of the World (GPW), v4 | SEDAC 

[JoeBorgione]

The polygon data shows 8,000; can you sum all your grid cells and see what that shows?