The State of US Highway Classification
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
Title |
| |
Title of Series | ||
Number of Parts | 41 | |
Author | ||
License | CC Attribution 3.0 Unported: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
Identifiers | 10.5446/58202 (DOI) | |
Publisher | ||
Release Date | ||
Language |
Content Metadata
Subject Area | ||
Genre | ||
Abstract |
|
State of the Map US 202236 / 41
1
4
7
8
14
18
21
22
24
27
28
30
37
39
00:00
Structural loadDefault (computer science)Cartesian coordinate systemInformationNumberLevel (video gaming)Router (computing)Decision theoryConstructor (object-oriented programming)Electronic program guideVisualization (computer graphics)Process (computing)Open setWikiLink (knot theory)EmailData analysisElectronic mailing listMountain passSoftwareDistanceMessage passingMappingKeyboard shortcutKey (cryptography)Context awarenessWordStatement (computer science)BitTranslation (relic)AreaFormal languageMultiplication signPopulation densityPoint (geometry)Social classThread (computing)Scaling (geometry)CASE <Informatik>State of matterAxiom of choiceZoom lensExpected valueArithmetic meanDirection (geometry)Presentation of a groupInterpreter (computing)HierarchyLimit (category theory)Connected spacePerspective (visual)VolumenvisualisierungCodeTerm (mathematics)Latent heatPhysicalismVery-high-bit-rate digital subscriber lineProjective planeTransportation theory (mathematics)Musical ensembleDifferent (Kate Ryan album)Spectrum (functional analysis)Arithmetic progressionLattice (order)QuicksortConsistencyAddress spaceRoutingRootBuildingData managementLocal ringDialectService (economics)Universe (mathematics)Contrast (vision)Physical systemWeightForm (programming)Radical (chemistry)Mathematical analysisPearson product-moment correlation coefficientLecture/Conference
07:28
Standard deviationLevel (video gaming)Open setRoutingPoint (geometry)BuildingMappingConnected spaceProjective planeArtistic renderingState of matterLevel (video gaming)VolumenvisualisierungPrime idealArithmetic progressionSet (mathematics)CodeMathematical analysisAsynchronous Transfer ModeWikiSoftwareDialectElectronic mailing listConstructor (object-oriented programming)Latent heatPhysicalismPopulation densityContrast (vision)AreaMultiplication signService (economics)HierarchyDirection (geometry)Local ringFormal languageForm (programming)CASE <Informatik>Exception handlingDifferent (Kate Ryan album)Functional (mathematics)BitWeightProcess (computing)Physical system
Transcript: English(auto-generated)
00:04
All right, well hello everyone. This is the State of US Highway Classification and my name is Adam Franco. I am a hobbyist mapper from Vermont. I will be your guide for the next 20 minutes. So why do we need highway classification anyway?
00:25
There are many users of OSM data, everything from data analysis and research, but two of the most common uses are for map visualizations and routing applications. And
00:41
both of these have some important needs that need to be accounted for. So a map visualization, when you're looking at a map, that needs to manage information density. So it's simply impractical at a regional or national scale map to show all roads. In that case cities and urban areas would just be a smear of roads.
01:02
And if you take the opposite choice and just show motorways, well, rural areas may not have any roads. And similarly, when a user is viewing a map, they need to have their expectations set. Just because there is a road through a mountain pass doesn't mean that that's a good road that users should necessarily take, especially in the winter months.
01:26
And for routing applications, there's no visual presentation necessarily, but when building directions, routers need to collect people from their points of origin on some small roads, and then get them to the bigger, higher throughput roads that can
01:44
efficiently get them to travel long distances. And without a systematic classification of highways, the router doesn't really know to keep them on a main road versus when to jump off and take a shortcut. That becomes a very confusing
02:03
situation. And so when we talk about highway classifications, we're talking generally about the OSM highway key. So these turns, motorway, trunk, primary, secondary, tertiary,
02:21
unclassified is actually a classification in OpenStreetMap. This is a Britishism and residential. How we translate those to our American context is actually a kind of tricky thing, because these words are not naturally used in how we talk about highways. So a bit of history.
02:43
Before, about a year ago, the highways in OpenStreetMap were very inconsistently tagged. So there was very little translation between a secondary road in one state or area and a secondary road in another, or trunk was used inconsistently around the country. What is primary? It's all very vague. And
03:07
compounding that, partially due to that British language translation into American English, the there became this convention of using the highway trunk tag to indicate
03:21
what we call expressways, or roads that are divided, high-speed, but not quite motorway or interstate construction level. So they might still have an intersection here or there, but they're a high-speed road. And at the opposite end of the spectrum, the Tiger import of a decade ago
03:42
classified all small roads as highway residential, just compressing the number of levels in the road hierarchy. Really confusing things when routers are trying to make decisions of which one to take. So as you can see from this screenshot of the Midwest,
04:04
when roads are tagged with the classifications that were being used with this trunk-meeting expressway rather than most important roads, it was really impossible to render a map that people looking for a highway map could actually
04:23
expect. You see all these little stubs and there is a way to get from a road from Springfield to Pine Bluff or from Fayetteville to Jonesburg. But those aren't shown because
04:44
those don't have that high level of construction. And then other places we see little islands of motorways here and there. Not a great situation. So yeah, as I alluded to, there's no way to render a coherent highway network at low zooms without showing way too many roads in urban areas.
05:06
And from the routing perspective, I've been trying to use OSM data for all of my personal driving directions just to make sure it works, and it didn't. My router was diving me off through city centers rather than sticking to the main highway as it should have. And
05:24
when we don't have these consistent classification, we end up getting into churn and edit wars because people, mappers, have different interpretations of what road class to apply. So there have been lots of mailing list discussions about this over the years.
05:43
It's been a frustrating problem until this past May, 11 months ago, now, as of I guess today is April 1st, maybe, a thread on the TalkUS mailing list came up.
06:04
Evan Fairchild started this question about the trunk road tagging yet again, kind of an annual topic coming up. And after 17 participants, since almost 70 messages, we actually dug into this again without the hopelessness
06:23
that had been there in the past and really figured out a problem statement and a way forward of how we could make the highway tagging in open street map more seen. And so those discussions moved out of the mailing list over time into the OSMUS Slack channel, Slack's
06:43
highway classification channel, where we've now had over 4,000 messages by 51 active participants and almost a hundred other, a hundred folks total reading the messages going by, and then even more in state-by-state level channels. So this is really getting a lot of buy-in from the active US mapping community.
07:06
And as we figured this out, we've documented our process in the open street map wiki. There's the 2021 Highway Classification Guidance wiki page, which has the full overview and everything you need to know, as well as links back into
07:24
our Slack discussions and mailing list discussions. So what is the new approach? So when we look at these highway tag guidelines that, with their British terms, I find it helpful to really look at these as a
07:46
just a labels for a seven-level hierarchy. So the highway motorway tag comes with some very specific physical construction limitations, so it has to be high-speed, it has to be
08:02
divided with controlled interchanges and a bunch of other guidelines. So trunk, highway trunk, is what is used for any roads that are those most important connectors between places that don't meet those guidelines. So they're really a 1A, 1B, they're the same thing from a connectivity perspective.
08:24
With motorway having that little physical piece augmented to it. And from there we go on down the hierarchy to the smaller and smaller roads that have connectivity importance more and more and more local.
08:41
So the other thing that came out of these discussions was that we really, as Americans, we really do need some tagging for those big improved infrastructure roads. And this is where we get to the expressway equals yes tag, which has been around for a decade but not heavily used because there weren't
09:04
renderers that used it for visualization, and it didn't really do anything. But we decided, you know what, we really need to fix this problem in the US. We need to break apart the physical from the connective importance of roads.
09:20
So as I mentioned, this is a connectivity-based approach, and here we have New York with, looking in New York State, has this lovely, so far the only data set we've seen from a state. They have an arterial classification code where the State Department of Transportation has actually
09:43
done that work of the connectivity analysis of their road network. So their top, their ACC equals one, maps directly to highway equals motorway. ACC equals two is for the non-motorway important connectivity modes. So those are shown in this
10:01
rendering. So those two form this top network, and as you can see over the top of New York State, there are not many people, so there's not much traffic, but there is still an important highway connection there. And so as we go down the hierarchy from that top motorway and trunk level, each additional level of
10:22
the network is still going to be a connected network with itself, trying to avoid routing islands or spurs, except in cases where there's terminal geography, like a blind valley going up the mountains to a town in the mountains or a peninsula where geography really dictates how roads appear.
10:49
So as we go looking for data, how can we apply, what data can we apply? So I mentioned New York has this nice arterial classification code, but most states don't. So State Departments of
11:03
Transportation will publish a functional classification, which is useful for figuring out which roads are more important than others. There's also the National Highway System, which is another designation applied by Departments of Transportation of the most important roads, and these
11:23
get us close, but they are not sufficient for what we need in OpenStreetMap. The functional classifications will, or both of these actually, really drive where highway funding is going, and so State Departments of Transportation and legislatures will give optimistic
11:47
or extra weight to roads that they need to do a lot of fixes on. And so when we're talking about how we do connectivity and actually making maps and routing systems that are useful, we can't just rely on these because the reason that these classifications are in place are not
12:06
the same reasons that we have. So there have been other suggestions, ideas proposed for how we might identify what are the primary and trunk
12:22
roads. U.S. versus state networks have been brought up, but that doesn't always work because these aren't consistently applied that U.S. routes are always more important than state roads. I believe California has actually gotten rid of all of its U.S. routes and made them all state routes. And then some states have county routes or divide their state routes into
12:45
primary state routes, secondary state routes, and then what about Texas, which also has farm to market routes and ranch to market routes, which are yet something else. How do we map all that in? So the process we develop is that all of these data are useful, but we can't rely on
13:10
them. And so we need to do something a little bit more human-centric, which is to determine what are the most important regional population centers, and then what are the best ways to get
13:21
between them. And that can really define our top-level, over-weighted trunk network, and then from there we can work down. So what are the most important regional population centers? We can look at raw populations, but that doesn't always work when we have such a high contrast
13:42
between urban density and rural density. So in an urban area there may be a bedroom community with no town center and very minimal services that has a population of 75,000 people versus in say rural Wyoming, the county seat that has an airport, hospital, a university,
14:08
and all of the services that people need for like several hundred miles in any direction may only have 15 to 20,000 populations. So we really need to have local knowledge and people
14:21
in each state looking at what are those most important regional centers that need to be connected by that big top-level highway map roads. So another big consideration in this is that we
14:41
need to document and discuss this. So this is happening in the OSM Wiki. As folks in each state are building out their lists of regional population centers and the routes between them, documenting this gives other mappers the opportunity to really reflect on what makes
15:03
these actually important. And so it's much more easy for someone coming in new to this to look at, oh these are the list of the top most important communities, what about Community X? Okay should this be in there or should it not? And there can be a discussion on what makes
15:24
the community important and actually get that list to be pretty solid. Same thing with the routings between them. There may be two parallel routes. You can have a debate on which one is better when you are looking at the documentation rather than mappers just going in
15:41
changing the open street map data somewhat arbitrarily and then churning back and forth. So another big advantage that we've seen is that this documentation and research project figuring out what are the most important communities, what are the best ways to connect them is really a great way to do outreach. And with that getting more eyes on the guidelines
16:12
is good and it also invites people who may have been editing on their own, adding buildings or an excuse to come in and start participating in some of the larger community discussions.
16:24
So at this point the trunk and motorway networks, the top level, has been completed in seven states and five more states are very close to complete. They have their drafts in progress, the mapping is happening, they're almost there, and another 12 states have drafts started. So
16:46
almost half of our states are making some progress on getting these state-level classification guidelines in place. So you remember this disconnected highway network? Well
17:00
in places where we've done this work, shout out to Moira Prime I believe did most of this work in Alabama and Mississippi, we now actually have the classifications in place to render maps that people expect, which opens up all sorts of opportunities. This is the American map style
17:21
that Brian presented on earlier. We can now actually do some consistent visual language to represent what those highway levels mean when it's not totally arbitrary. And similarly we can now render expressways in a style that is familiar to users of American
17:41
paper maps with this case with a white fill. And in this case you see both trunk-level roads that are expressways as well as primary roads that are expressways. So different levels of connectivity importance but they're both built as big infrastructure. So what's next? This afternoon we have a highways birds of a feather session which will not be
18:08
at 320 I should say, which is not going to be specifically highway classification, it'll be all things highways. And then Brian and I tomorrow will be leading a workshop on drafting state-level highway classification guidelines. So if you are in a state that
18:24
has a draft in progress or needs a draft started, come to that workshop and we'll get you going and figuring out what your next steps are and getting those guidelines off the ground. Also join us in the slack or the talk us mailing list and on the wiki.
18:46
So I think that is about time.
Recommendations
Series of 70 media
Series of 41 media