So, how did I find the 58 likely Phantoms as described in the Splash/Info Page? I investigated several different parameters to "weed" Phantoms out from all the data sources (see Future Study below). Things became complicated. In the end, the methodology I used was to determine if, within a quarter mile radius of their lat/long point locations, any of the possible Phantoms (as accessed from the GNIS) existed on land that was NOT residential. This was a proxy for areas where there is no remaining residential population and therefore no existant community. Here's how I did it using ArcGIS Pro:
Future Study:
- Ground truth, satellite imagery, street-view (lower hanging fruit)
- Remove those unincs that are close (how close??) to more urban areas
- Churches
- Cemeteries
- Fire Stations
- What proportion of area of residential parcel compared to the area of the buffer?
Created by Michael Hasinoff for Geography 778: Practicum in GIS Development, UW-Madison, Fall 2022