[Talk-us] proposed first principles for United States road tagging

[David ``Smith'']

On Thu, Mar 4, 2010 at 12:38 PM, McGuire, Matthew
<Matt.McGuire at metc.state.mn.us> wrote:
> I see three dimensions of road classification at play here.
>
> 1) System
> 2) Function
> 3) Observed Character
>
> System is the easy one. That is the road system(s) that that the road belongs to especially for signage, but also for road funding channels, and maintenance responsibility.  And I agree that in practice, Census Feature Class Codes have been used (incorrectly) to identify the system to which a road belongs.

Actually, the CFCCs are based almost entirely on the system to which a
road belongs.  The Census Bureau just botches it a lot because they're
not a highway department.

> Function describes the role a road plays in a road system and the types of trips (volume and length) it supports based on travel demand and trip generation. This is what the Highway Functional Classification System is designed for. It is used by Metropolitan Planning Organizations to distribute transportation funding.

This is what the highway tag should mostly be based on, because that
makes the most useful map.

> A road's Observed Character is what kind of road it appears to be to a person on the road. For general purpose maps, using observed character to classify the roads intends to match a person expectations to what they see on the ground. Character is highly correlated with function, but is not the same.

> I think Observed Character is what OSM is trying to achieve with the highway tag.

No, it's not!  Because character is highly correlated with function,
it's possible to have a close guess of function based on observed
character most of the time.  But if we give each highway a
classification based entirely on observed character, we'll have sloppy
maps that are difficult to understand.

> I think this because the OSM tag descriptions for highways have photos and describe how the road looks, and you cannot determine system or function from a photograph.

You've got it a bit backwards.  Because you cannot determine system or
function from a photograph, and the people who wrote the tag
descriptions were in a mindset of mapping from photos and physical
appearances only, the OSM tag descriptions are based on that.

> I also think it is what the Census Feature Class Code definitions describe.

The CFCCs, at least the ones that deal with roads, literally describe
a combination of system and observed character, though it can be
frequently wrong on either.  They translate to things like
"Interstate", "undivided US route", "state route in tunnel", and
"local street".

> I would like to see all three dimensions.

Where the OSM community is trying to go is to use the highway tag for
function, the ref tag and route relations for system, and various
other tags to describe physical characteristics.  Right now,
"highway=motorway" and in some places "highway=trunk" are tied to (or
strongly imply) certain physical characteristics, but I'm working on a
proposal to change that slightly.

-- 
David "Smith"
a.k.a. Vid the Kid
a.k.a. Bír'd'in

Does this font make me look fat?