usaak (Alaska State Highways)

[oscar]

I'm not sure there are any unsigned mainline U.S. routes, but certainly there have been unsigned U.S. and Interstate business routes, which we normally remove even if they're still technically on AASHTO's books.

I assume you mean US routes non-concurrent with interstates in places that don't post concurrencies well or at all....

I was talking about US routes that aren't signed at all, anywhere. For example, on one of the isolated non-US route segments in usaak, where every route marker "disappeared" as soon as it was posted, until the state DOT gave up (even the former US 666 never got to that point). If it's signage that's lousy by lower-48 standards but typical in the Arctic (like gaps of over 100 miles between reassurance markers, or no route signage on concurrencies except at each end), I'm still counting it as "signed".

[Duke87]

According to this map from the state DOT, AK 5 goes up to Eagle rather than to the YT border at Little Gold.

[oscar]

According to this map from the state DOT, AK 5 goes up to Eagle rather than to the YT border at Little Gold.

That looks like an old map.

Part 2 of the current Alaska Traffic Manual (2016 edition), at pages 55-56, has a map showing AK 5 turning east to the Canadian border rather than continuing north to Eagle. It also has a route definition table confirming that AK 5 includes the Top of the World Highway to the border, and part of the Taylor Highway but only between Top of the World and the Alaska Highway.

I had an e-mail exchange with someone at Alaska DOT&PF on the previously confusing definition of AK 5, which might've led to the clearer definition in the current manual.

OTOH, I'll have to follow up on an ambiguity in the current ATM on the definition of AK 10's northern segment, about whether it includes the McCarthy Road between Chitina and McCarthy as is now in the HB, or rather ends in Chitina. The 2012-2015 ATM supplements specified that AK 10 ended in McCarthy, much to my surprise (no AK 10 signage east of Chitina when I was last there in 2012, and the mileposts reset to zero in Chitina and change to a distinctive Chitina-McCarthy Road milepost style). The current ATM removes the specific reference to McCarthy, and suggests that AK 10 ends where the McCarthy Road begins in Chitina. I'm leaning to truncating the route to end in Chitina, but plan to re-contact Alaska DOT&PF to confirm that.

[oscar]

Alaska DOT&PF confirms that the northern (Chitina) segment of AK 10 ends in Chitina, where the Edgerton Highway ends and the McCarthy Road begins, rather than where the McCarthy Road ends in McCarthy as it was in the HB. I'll submit a pull request for this change tomorrow or Sunday. EDIT: AK 10 truncation is now in the HB.

With this loose end tied up, I think usaak is ready to be activated, after clearing a sharp angle Datacheck error and any temporary points in route files, unless someone would first like to take another look at this system or provide other comments. Otherwise, I'll activate usaak at the end of this month.

[oscar]

Last call for comments, or requests for more time to comment. Otherwise, I plan to activate usaak this Wednesday or Thursday.

[yakra]

Requesting a couple more days to comment, if it's not too late.

[Jim]

I'll hold off on the pull requests for the activation until I get the word that it's a go.

[yakra]

I gave AK1 a look-over, then remembered it's been largely active as part of usai for a long time.
I'm probably better off focusing on the more newly-drafted mileage.
More to come...

[yakra]

Skipping over the routes with large concurrencies that would require changes to active usai routes for now.

AK4:
CorLpDr: delete
+X723399 off-center
BluLakeRd: checking satellite and GMSV, I see no road junction in the area
WorGlaSRS: "State Recreation Site"?
+X880724: potential OldRicHwy point?
+X877209 off-center
+X243886 can be safely deleted
Is +X466714 waiting for a valid name, to become visible in the future? Otherwise, move north
+X355129 looks nicer here
+X259213 -> http://www.openstreetmap.org/?lat=62.545893&lon=-145.510826 ?
+X655172 -> http://www.openstreetmap.org/?lat=62.846138&lon=-145.470914 ?
PowTowRd: Aw gee whiz, they couldn't have called it Tower of Power Access Rd?

OldRicHwy points: Was there an effort to include most or all of these, was it largely as-needed for shaping, or did the age of the bypass (EG, _G & _H) factor into the decision?
The series skips from _E to _G, with no _F.

AK5:
Many points are a wee bit off-center, probably due to older less precise cartography when AK5 was first drafted. It may benefit to load into WPTedit and thaw & recenter points (I could GISplunge it if you're interested). In particular, a new JeepTr & +X840617 will get the center line more clearly within tolerance.

out of tolerance between +X669253 & +X378006
+X275517, +X819620, & ++X571860 can becomw two shapers if you're so inclined.
+X973259 can be deleted
+X284856 looks nicer here

Waypoint labels look good.

[oscar]

One overall comment: si404 did a peer review of usaak already, and some of the things you note made it past his earlier review. My suggestion for a "second look" of this system really was to draw comments on some of the more global issues raised by the system, including one (the indefinite but not necessarily permanent bridge closures in the middle of ak010cor) I hadn't brought up before Si did his review.

BluLakeRd: checking satellite and GMSV, I see no road junction in the area

Mapnik shows a junction to the east, to which I've moved the point. The old waypoint was carried over from CHM days, and the access road may've been relocated (or Mapnik may've improved its mapping) since then.

BTW, I also added ThoLakeRd on the other side of the switchback. Mapnik doesn't label that access road, but Alaska DOT&PF's route log (taken offline a few years ago) does. It also gets rid of the sharp angle error that had been at that switchback.

WorGlaSRS: "State Recreation Site"?

Yep. A "drive-up" glacier a short walk from the parking lot, which gives you a great view of all the crud the underside of a glacier picks up as it slowly flows downhill.

+X880724: potential OldRicHwy point?

Nope, no intersection in the area. +x230805 would be a more likely candidate, though it looks like the old highway was closed south of there and no longer connects to OldRicHwy_B.

+X877209 off-center

Off by about 0.05 mile, so I've gone ahead and shifted it on top of the highway as shown in Mapnik. That said, I'm not crazy about doing that for other routes, unless the point is far enough off-center that the Mapnik trace strays outside the lateral tolerance lines in the Waypoint Editor. I get pretty fussy about placement of labeled waypoints. But as I've said elsewhere, I don't view shaping point placement as a high-precision exercise. Especially in the Arctic, where Mapnik can sometimes de-center previously well-centered shaping points as it gradually brings its mapping up to normal lower-48 quality.

Is +X466714 waiting for a valid name, to become visible in the future? Otherwise, move north

It's the access road to an airstrip alongside the highway. But I can't find a name for that airstrip or the access road, not even in the DOT route log or the Milepost's more up-to-date and highly-detailed route log.

+X355129 looks nicer here
+X259213 -> http://www.openstreetmap.org/?lat=62.545893&lon=-145.510826 ?
+X655172 -> http://www.openstreetmap.org/?lat=62.846138&lon=-145.470914 ?

I made the third change. I made the first, then slid the shaping point a little farther north so I could remove another shaping point. The second change didn't do anything for me.

OldRicHwy points: Was there an effort to include most or all of these, was it largely as-needed for shaping, or did the age of the bypass (EG, _G & _H) factor into the decision?
The series skips from _E to _G, with no _F.

The points were included only as needed for shaping or to eliminate/reduce distance errors. Indeed, OldRicHwy_A isn't even the southernmost intersection with the old highway. A waypoint there would've been otherwise unneeded. Others were skipped for similar reasons, unless I know that a turnoff for the old highway is used by ordinary travelers to access local businesses, etc.

I could change _G and _H to _F and _G to fix the sequence, but that could screw someone who is using the _G point (no easy way to find that out, in a preview system). So I'm leaving those alone.

EDIT: Revised AK 4 file now in the HB.

[yakra]

One overall comment: si404 did a peer review of usaak already, and some of the things you note made it past his earlier review.

How did I miss the fact that that had happened? -_- Oh OK, AARoads, 2015...
I'f I'm not finding much more than minor shaping nitpicks that aren't of much importance, then usaak is probably, pardon the pun, in pretty good shape.
If Si already gave a full review, then I can call off my review if you'd prefer, and just let you activate the system. You've been waiting on usaak long enough.

My suggestion for a "second look" of this system really was to draw comments on some of the more global issues raised by the system, including one (the indefinite but not necessarily permanent bridge closures in the middle of ak010cor) I hadn't brought up before Si did his review.

"indefinite but not necessarily permanent bridge closures" -- would the IN912 precedent apply here?
I'll reread the AARoads thread, and this one, to catch up on global issues.

+X880724: potential OldRicHwy point?

Nope, no intersection in the area. +x230805 would be a more likely candidate,

Oops, I guess what I saw in satellite view was a utility or pipeline ROW.

though it looks like the old highway was closed south of there and no longer connects to OldRicHwy_B.

Since there's a point there anyway, unhide & optionally mark as closed?

That said, I'm not crazy about doing that for other routes, unless the point is far enough off-center that the Mapnik trace strays outside the lateral tolerance lines in the Waypoint Editor. I get pretty fussy about placement of labeled waypoints.

On AK5, WinTr_A is significantly off-center, as well as the aforementioned JeepTr.

I don't view shaping point placement as a high-precision exercise

Fair enough. I'll enter "silent mode" for off-center shapers in any future review.

I could change _G and _H to _F and _G to fix the sequence, but that could screw someone who is using the _G point (no easy way to find that out, in a preview system). So I'm leaving those alone.

Preview systems show up in pointsinuse.log, E.G. canmbw routes such as mb.rt042 .
No OldRicHwy points are in use. Same result if I search GitHub.

[oscar]

My suggestion for a "second look" of this system really was to draw comments on some of the more global issues raised by the system, including one (the indefinite but not necessarily permanent bridge closures in the middle of ak010cor) I hadn't brought up before Si did his review.

"indefinite but not necessarily permanent bridge closures" -- would the IN912 precedent apply here?
I'll reread the AARoads thread, and this one, to catch up on global issues.

I discuss the IN 912 precedent above. Basically, that closure was thought to be permanent, only that (assumed?) decision was reversed later, prompting some grumbling from people who'd clinched the route before what turned out to be a temporary closure.

On AK5, WinTr_A is significantly off-center, as well as the aforementioned JeepTr.

Hmmm, I updated labeled points on that file last month. Maybe Mapnik updated its mapping in the meantime. I'll go ahead and make those fixes, and while I'm at it check the shaping points.

Some labeled waypoints, either in this file or others, with both coordinates ending in 0 came out of my GPS receiver (which spits out only five digits after the decimal point). Unless Mapnik has since provided an intersection point of its own to replace my GPS reads, I'm reluctant to change those, since I'd be only guessing what the new coordinates should be.

[yakra]

si404 did a peer review some time ago. However, I think the system would benefit from at least a brief second look before activation. In particular, I would welcome comments on whether to truncate AK 10's Cordova segment, in light of the long-term bridge closures discussed in the preceding post;

I'd say truncate it. I don't like the idea of including inaccessible road segments, that cannot be clinched, in a route file to be marked as clinched.
If another segment farther east can be accessed by a (private?) ferry, then splitting the route is a potential option. However...

keeping two unsigned segments of otherwise signed routes (as explained in the thread on TM's temporary forum on the AARoads forum ... )

IE, AK 7 (Petersburg) and AK 10 (Cordova). I'm less convinced on this; I don't think the exception to the policy of excluding signed routes is justified. I fear the slippery slope...

and other notes in that thread.

Ferries: As far as I can see, there are no routes containing ferry segments, even short river crossings. Everything that should be chopped, is. Good.

"Pump Station _" points: Only used when needed for shaping or to break up long visible distance segments, I assume?

Any suggested fixes on those segments will need to be made to one or more other routes, too, so I'll be more reluctant to make such changes than for the non-Interstate routes.

Right, any changes here would involve edits to activated routes anyway, so I understand putting less priority there.
That said though, at least looking at AK1, there seemed to be some points for minor side roads, not necessary for shaping or distance, that could be trimmed. (Relics from early Interstate days, perhaps?) Do what thou wilt.

Any other notes from these threads I've missed?

[oscar]

In particular, I would welcome comments on whether to truncate AK 10's Cordova segment, in light of the long-term bridge closures discussed in the preceding post;

I'd say truncate it. I don't like the idea of including inaccessible road segments, that cannot be clinched, in a route file to be marked as clinched.

Even if the state is treating the closures as temporary?

In the usanp system, I pushed for and got re-inclusion of part of Hawaii's Crater Rim Drive that has been closed for several years (not before I and at least one other user clinched the entire loop), but may reopen once volcanic activity at that crater calms down and shifts to other parts of the volcano.

keeping two unsigned segments of otherwise signed routes (as explained in the thread on TM's temporary forum on the AARoads forum ... )

IE, AK 7 (Petersburg) and AK 10 (Cordova). I'm less convinced on this; I don't think the exception to the policy of excluding signed routes is justified. I fear the slippery slope...

I think there are lots of ways to avoid the slippery slope. One, for AK 10 (Cordova), is an exception for a route the state has tried to sign, only to be frustrated by persistent sign vandalism. Another, for both segments, is to limit exceptions to jurisdictions that don't assign or post route numbers for most of their highways (in Alaska, there are hundreds of state-maintained highways, but only about two dozen have route numbers). In my other jurisdictions, where route numbers are generally assigned to every state/provincial/territorial highway no matter how unimportant, I've not made or pressed for any exceptions, even though their exclusion caused me and other users some heartburn (e.g., CA 259). But Alaska is very different.

I think we've relaxed the unsigned routes rule in systems and jurisdictions outside North America where route signage practices are different than those in most of the U.S. and Canada. Right?

"Pump Station _" points: Only used when needed for shaping or to break up long visible distance segments, I assume?

No, I included points for all the pump stations that are on or near AK 11, AK 2, or AK 4. Only gaps in the number sequence are where the station is far from the highway, or was never built.

[oscar]

+X880724: potential OldRicHwy point?

Nope, no intersection in the area. +x230805 would be a more likely candidate, though it looks like the old highway was closed south of there and no longer connects to OldRicHwy_B.

OldRicHwy points:
The series skips from _E to _G, with no _F.

+X230805 => OldRicHwy_C (filling a gap in the sequence)

OldRicHwy_G and _H => OldRicHwy_F and _G.

These changes are in my local copy, to go in with my next pull request later this week.