- Mapping roads and paths in National Forests and Parks using aerial imagery, USGS and USFS/NPS Topo maps, USFS MVUMs, USFS/NPS GIS data, and the USFS Visitor Map.
- Cleveland National Forest - done!
- San Bernardino National Forest - done!
- Mojave National Preserve - done!
- Building and maintaining the RecoGNISer matching software with Matt Whilden which matches GNIS record data against existing features in OSM, identifies discrepancies, and recommends updates to OSM as MapRoulette tasks.
- Mapping BLM Ground Transportation Linear Features using aerial imagery, BLM GIS data, USGS Topo maps, county GIS data (where appropriate), and surveys
- San Diego County - done!
- Imperial County - done!
- Orange County - done!
- Riverside County - done!
- Ironwood Forest National Monument - done!
- Sonoran Desert National Monument - done!
- Mojave Trails National Monument - in progress
- San Bernardino County - in progress
- Cleaning up largely untouched TIGER roads in backcountry areas using aerial imagery, Bing Streetside, and county GIS data (where appropriate)
- Glossary of tags for landforms
- Published to Glossary of landforms
- Helping geography students in French Polynesia
- Adding all the missing Summit records from GNIS to OSM
- Documenting transformations from NAD83 to WGS84 for use with US government data sets
- ... with help from Ray Vanlandingham
- How to transform data from NAD83 to WGS84
- Documenting updates to municipal and CDP boundaries using the latest TIGER/Line data
- Updating OSM to reflect the removal of Sq___ names from features manged by the USGS Board of Geographic Names
- Joint effort with several other mappers - Go team!
- Also updated Wikidata and GeoNames to reflect the USGS BGN changes
Long-Term Maintenance of OSM Features
During the Sq___ rename, I noticed that many of the features we were working on were already mapped and had GNIS IDs that could be traced back to source records. We also came across features with Sq___ names that had been changed years ago by USGS but that hadn't been updated in OSM. For geographic features in the US, the USGS Board on Geographic Names is the definitive source for naming. If BGN changes a name, that's the official name everyone should use, including OSM (although we can and do keep old_name=* values and other name variants).
Another thing that was clear during the Sq___ rename is that USGS GNIS data is generally pretty good, but it's not perfect. It's not something you can import directly into OSM without manually reviewing all the data. So OSM has features that directly relate to GNIS records, but that often have better data than GNIS provides.
Like the Sq___ names that changed, things do change over time. Official names change. People build and demolish things. Landforms change naturally. And even when things don't change in the real world, sometimes data that's brought into OSM gets stale. Many geographic features in the US were originally imported from GNIS records, but those records weren't perfect and not all of the imports were manually reviewed. GNIS is continually improving their data to correct names, locations, and even the basic type of features (e.g., is it a spring or a well?). But OSM generally doesn't keep up with those updates.
This is what got me started on the RecoGNISer project to automatically match current GNIS records to features in OSM to identify discrepancies. That project found that about half of the features in GNIS are not yet present in OSM. And judging by some data sampling, about half of the GNIS features in OSM could be updated to better reflect current GNIS data. That's a huge task covering millions of features!
If we want OSM data to remain fresh over time, in general we need a few things:
- Unique references in OSM that allow us to link back to definitive external data sources
- E.g., the ref=* tag and its variants
- Automated processes that review features in OSM and check them against the original external data sources
- Manual review to ensure that high-quality edits and corrections of the original source data in OSM are preserved
This is also a potential path to quickly bring new features into OSM in a sustainable way that is not a bulk import of data.
|Kai Johnson (on osm, edits, contrib, heatmap, chngset com.)