Managing Map Data for Change Detection Analysis

General
1

I recently digitized a map of the Fox River in Kendall County from 1967. I asked a previous question regarding how I can save my data using either the in-browser editor or the JOSM editor. While I received helpful answers, I have another question related to my previous post. I want to save this data so I can reuse it and update it for a 1986 map of the same area. How can I do this without altering the original data from the 1967 map? I’m conducting a change detection analysis between the two maps and would like to copy the 1967 data to create a separate dataset for the 1986 river boundary. How can I ensure that the data for the 1967 map remains intact in OHM while I work on the 1986 version in OHM? Is there way to do this without exporting the 1967 data? Is this possible?

2

It sounds like what you have as input and what you want as output are both a series of snapshots in time. What OHM stores is a little different: if a building is identical in both the 1967 and 1986 maps, then OHM would ideally have only a single copy of it, with start_date=* and end_date=* tags that encompass those two dates. After saving it in that form, you can query Overpass for all the buildings or all the buildings at a given point in time. In other words, change detection would be part of the process of mapping in OHM.

That said, OHM is iterative. You can digitize the contents of the 1967 map first, tagging the buildings with start_date=1967 start_date:edtf=/1967 and end_date=1967 end_date:edtf=1967/ tags that leave open the possibility of it existing before or after. Upload that, then move on to the 1986 map:

  • If anything has changed in this map, retag the existing feature as end_date:edtf=1967/1986 and map a new feature reflecting the state of things after the change.
  • If anything hasn’t changed in this map, retag the existing feature as end_date=1986 end_date:edtf=1986/, because you can rule out an earlier end but leave open the possibility of a later end.

I used buildings as an example because they’re fairly isolated and easier to work with. Modeling rivers over time is a bit more advanced; see this topic for details:

1 Like