I divided one 30 Arc-Second (~1km) Basic Demographic Characteristics raster by another and ended up with very high values in the result.

The raster population count data were created using an areal-weighted distribution method where population is evenly spread across the entire administrative unit to which it belongs on the basis of land area. This may result in very small count estimates for a given pixel if the input administrative units were large (districts, provinces, etc…) and sparsely populated. Division by extremely small numbers may lead to artificially high results.

In order to control for this, we recommend that users consider lower spatial resolution versions of the data (2.5 Arc-Minute, 15 Arc-Minute, 30 Arc-Minute, or 1 Degree) in order to smooth away the artifact, or conversely impose a ceiling on map algebra operations at 30 Arc-Second resolution in order to snap extremely large artifacts to more realistic values.

Feedback and Knowledge Base