Some Tennessee county boundaries noted as erroneous by TDOT

stevea · 2024-08-29T17:27:01.250Z

I recently received a forwarded email between some planners in Nashville and TDOT (the state’s transportation department) that some county boundaries in OSM data (notably where DeKalb meets Wilson around Alexandria) may be erroneous. This is causing some conflict as state GIS users use both OSM and E-TRIMS data to such an extent that recommendations are being made that OSM county boundary data in Tennessee not be used.

If you are local to the area or this might affect you, “heads up” that OSM county boundary data in Tennessee might need some mapper love and to be checked against authoritative sources.

AntiCompositeNumber · 2024-08-29T17:34:40.921Z

The admin boundaries for most of the US were imported from the TIGER database. My understanding is that at the time of import the TIGER data was low resolution and/or there was an error converting the data from NAD83 to WGS84. The current TIGER data is generally better and can no be converted to WGS84 without error, but importing it and conflating with existing boundaries is difficult.

It would be a good idea to make a plan to fix this that can be repeated nationwide.

stevea · 2024-08-29T17:43:58.785Z

Last updated a couple years ago, our TIGER fixup/County borders - OpenStreetMap Wiki tracks this (perhaps related to the particular problem around Alexandria, perhaps not). But these efforts (and other slow-growth TIGER-improvement efforts) seem to be largely stalled or are on a rather “lagging gain in altitude.”

Mappers who (continue to) improve our TIGER data, we need you, we love you!

clay_c · 2024-08-29T17:49:38.359Z

AntiCompositeNumber:

It would be a good idea to make a plan to fix this that can be repeated nationwide.

Agreed, and we should include minor civil divisions in such a plan, as they’re still mostly absent from OSM in several states where they exist.

ElliottPlack · 2024-08-30T20:45:44.116Z

I have worked on redoing county boundaries, and it is a tough job with all the relationships to manage. Anyone here willing to take on this task: you have my respect.

Some of the challenges include:

Unsnapping all the random land use and other elements that people attached to the boundaries, which shouldn’t be there.
Cutting up the source data—likely imported from elsewhere—at every location where the current county boundary splits due to a municipal boundary, state boundary, river, or some other feature used in the relation.
For big counties, just getting all the data into JOSM is a chore.
Dealing with modifying state boundaries as well, especially when the adjacent state boundaries are newer, have a modern source, or are not in alignment with your data.

In my professional experience working for the State of Maryland, I have noticed and pointed out to others that our official state boundary GIS data does not align with the official State of Pennsylvania boundary data. Officially, you will get a response from officials that only the statute can define the state or county boundary. Therefore, it can be open to interpretation as to where the boundary really lies between those two lines. Although OSM is not meant for survey purposes or anything like that, it does end up in a lot of applications where the precise location of boundaries matters to people.

stevea · 2024-08-30T21:56:47.552Z

Awesome reply, Elliott. I copy-pasted to Kerry (Irons, of Adventure Cycling Association, who originally prompted me starting this topic with an email by the TDOT folks) so he can forward to TDOT and “give them some context.”

County boundaries in USA really DO improve (over time, usually years), but it can be slow going! Thanks to all who contribute good work to the cleanup and improvements.

whammo · 2024-09-06T10:29:49.278Z

I went through and updated the county boundaries you highlighted @stevea as well as any others that looked particularly egregious throughout Tennessee. Wholeheartedly agree that the challenges @ElliottPlack mentioned make it hard to do this in a more systematic way at this point, but I’m writing a diary post with some of the marginal tips and tricks I’ve come up with.

stevea · 2024-09-06T19:56:04.333Z

I am a single contributor in a big project. When “being on this awesome team” happens, because we are a community like this…well. Take a look around (at our data and our people / community): OSM is a pretty awesome project. Thanks to everybody who intersect at these junctions, we certainly do work together.

maxerickson · 2024-09-06T22:16:36.968Z

The connectivity of the imported borders was fixed a long time ago. That wiki page could be archived or whatever (or perhaps re-purposed into an example that didn’t imply it was an ongoing activity).

maxerickson · 2024-09-06T22:20:53.217Z

When I was working on townships in Michigan, I made a translator for ogr2osm that transforms the state township data into OSM style exploded relation areas, with shared border ways. The logic is generalized, splitting areas at points that are part of more than one area, and it could be integrated into the tool instead of being part of the translation file.

It doesn’t complete the process of aligning the source data with existing OSM data, but it gets it a whole lot closer than it is if the areas are translated over as ways.

github.com

maxerickson/michigantownships/blob/master/twptranslation.py

import geom

import csv
import os

__location__=os.path.dirname(os.path.realpath(__file__))

def filterTags(tags):
    if tags is None:
        return
    newtags = {}
    # append " Township"? could also use LABEL
    for t in ["name","county"]:
        if t in tags:
            newtags[t]=tags[t]
    if "border_typ" in tags:
        t=tags["border_typ"]
        newtags["border_type"]=t.lower()
        if t=="Township":
            newtags["admin_level"]="7"

This file has been truncated. show original

stevea · 2024-09-07T16:15:53.583Z

@clay_c, @AntiCompositeNumber, I am low on bandwidth to help on this, but how might MCDs initially get chiseled out (roughly)? Start with a wiki page “per state?” This could feed into a “national projects queue” (to do). I know that such “wishes and dreams” (lists) of would-be-nice items can fizzle, I also know that with a breath of quick effort, breezes can become strong winds. A lot of good work is done by “channeling consensus.” (Though, it can be like herding cats).

Our US admin_level and US Boundaries wiki pages are like “a couple of books in the library,” they might be a right place for documenting MCDs. Although as I think about it, and because those are so very detailed, maybe MCDs deserve their own (sub-USA) wiki?

@maxerickson, that Python translator code for ogr2osm (part of GDAL) could be very helpful to exactly the sort of people (in TDOT, for example) who were at the start of this topic: thank you! Kerry has forwarded to TDOT the message that the (initial to topic) county boundaries have been improved by @whammo, and that OSM has “stepped up” and is really “on it.” Again, OSM has been Team Awesome here!

clay_c · 2024-09-09T14:48:44.842Z

stevea:

but how might MCDs initially get chiseled out (roughly)? Start with a wiki page “per state?”

This is probably a good approach, since MCDs are defined differently (or not at all) state-by-state. State GIS offices may have more accurate, authoritative data than TIGER anyway.

Some states require that land be withdrawn from a township when a neighboring municipality annexes it, which may necessitate periodic (yearly?) updates to MCD boundaries.

whammo · 2024-09-19T11:17:51.597Z

whammo:

I’m writing a diary post

I created a Wiki page, rather than a diary entry, to describe my process in updating these TIGER county boundaries. No earth-shattering tricks, but hopefully helpful and encouraging for anyone considering taking a stab at it.

https://wiki.openstreetmap.org/wiki/TIGER_fixup/County_boundaries