Relative Population Density Gravity Model

Using open source data to model population density in remote areas

 

High fidelity population density data can be difficult to obtain in many parts of the world, if it exists at all. Global measures of population density include the Landscan dataset from Oak Ridge National Laboratory, but with a resolution of 1 kilometer, Landscan has limited utility. MDA encountered a need for a higher fidelity population density index for human geography analysis in the Philippines. In particular, our application required a relative (not absolute) measure of population density to use as an input to other geospatial models.


To achieve our goal of a high fidelity relative measure of population density, we combined MDA’s proprietary 30 meter land cover dataset and a 30 meter digital elevation model with analysis of open source data from Twitter, Foursquare and Wikimapia. We used a gravity model to combine the open source information with our land cover and elevation data to create a single layer depicting a population density for each 30 meter cell relative to the entire area of interest. This layer is a critical input to other human geography geospatial models such as mapping remote havens for bad actors or optimal distribution of resources for humanitarian aid.