Japan: National Highways 一般国道 system

[nezinscot]

All of my driving in Japan has been on the national highway system and on prefectural roads, not the expressways. I wanted to add this mileage but did not find a highway system for the "common" national highways.

Is anyone working on a highway system for the 55000+ km of national highways?  If no one is and you all feel the system is needed, I would be willing to help with the task.

I speak, read, and write intermediate level Japanese and would be willing to use this painfully acquired skill to dig through the ministry websites and GIS systems for original sources.  I am also familiar with the problems related to Japanese street naming and addresses.  The file creation and manipulation part of the task shouldn't be a problem because I spent years working in science and engineering-related IT.

The work would also help me improve my Japanese. I was definitely never taught highway-related vocabulary in class...

[michih]

Great news si404 developed the existing systems. I think he will answer what he thinks about it.

General note:

The requirements for developing new systems are described here: https://travelmapping.net/participate.php?u=michih&#hwydatamanager

When you reach stage 5 (I guess you are currently at stage 3?), jpntk and jpne could be your candidates for the peer-review:
https://forum.travelmapping.net/index.php?topic=3004
https://forum.travelmapping.net/index.php?topic=3668

I started the jpne system peer-review last May but gave up quite soon and will not work on it again any time soon.

[nezinscot]

I'm at level 4.  I read the developer manual before I volunteered so that I could have an idea of what the work might entail.  I no doubt need to give it another close reading.

Japanese numbered highways have some interesting quirks --- most intersecting roads are nameless, roads can span multiple islands, mountain roads can use widely divergent paths for each direction, roads may have two options (one a new tunnel the other the old twisty mountain road), ridge roads often wind back and forth across prefectural borders, etc.  At least the GSV coverage is good.

There are 47 prefectures, so a complete Japanese system would be a mini-US.  And if we include the extensive train and metro systems, the task gets even bigger.

[nezinscot]

I have almost completed my review of the Japanese expressway systems and would like to work on the Japanese national highway system next. 

There are 400+ national highways in Japan, plus hundreds of related bypasses and older routes that are still considered part of the national highways.  In all there could be ~1000 separate entries in this new system.  I believe we are trying to create smaller systems, so if I started this work, I would like to create separate systems for each prefecture (province). There are 47 prefectures. 

Separate systems for each prefecture will also help deal with the abundant duplicate prefectural road names found on the longer national highways. For example N4, the longest national highway, intersects P8 and P25 in four separate prefectures.

There is information online about the newer parts of the national highway system, especially the bypasses around towns.  GSV coverage is very good in urban areas and adequate in rural areas.  Highways are usually well marked with route and mileage markers.  Plus there are enthusiast websites in Japanese that cover the routes in detail, for example http://hgf03030.a.la9.jp/index.htm.

I would like to know if you all consider this a project worth doing.

[michih]

I believe we are trying to create smaller systems, so if I started this work, I would like to create separate systems for each prefecture (province). There are 47 prefectures.

Separate systems for each prefecture will also help deal with the abundant duplicate prefectural road names found on the longer national highways.

I think that we can go with smaller systems through the development process as we did for gbna or deub but the systems should be merged on activation. However, we should not do it by prefectures since the routes (wpt files) would need to be changed. This would affect users.

For example N4, the longest national highway, intersects P8 and P25 in four separate prefectures.

There is a simple solution: https://travelmapping.net/devel/manual/wayptlabels.php#over2

I would like to know if you all consider this a project worth doing.

I think so! We have 7 users with travels in Japan so far. We have regions or systems with less travelers.

[nezinscot]

I'll follow your advice about using devel-only systems to split up the work.

I think jpnh should be the system name for the national highways.

Devel-only systems can be used for each island, or island section for Honshu. They can be swept into jpnh as they are completed.
   jpnhky - Kyushu and Okinawa
   jpnhsh - Shikoku
   jpnhhk - Hokkaido
   jpnhkt - Kanto (east Honshu / Tokyo)
   jpnhks - Kansai (west Honshu / Osaka)

 I was thinking that jpnh would be type 3, but I see that US national highways are type 2.  Would type 3 be the better choice?  Japan already has systems for type 1 and 2.

[michih]

I was thinking that jpnh would be type 3, but I see that US national highways are type 2.  Would type 3 be the better choice?  Japan already has systems for type 1 and 2.

It should be tier 4 like the other national highway systems in Asia (and Europe) but the Indian one is tier 3 - likely a mistake.

[michih]

Just a note on N295:
https://travelmapping.net/hb/showroute.php?r=jpn.n295
https://travelmapping.net/hb/showroute.php?r=jpn.e065

- E51/E65 must be a one-point-per-interchange with E65's N295 wp
- E51/E65 might even be a one-point-per-interchange with E65's exit 10 + N295 wp
- E51/E65 does not (directly) connect to E51. Wp should be called E65_something
- NarApt must be a one-pont-per-interchange with E65! I tend to adding a wp to E65 at NarApt location.

[yakra]

N481
While deep into some shell scripting, I found E71(4)/P29 which led me here.
This would want to have either the E71(4) or the P29, but both isn't really something we do.

Avoid using two designations and a parenthetical suffix in the same label.

For the coordinates, I'd recommend moving the point in E71's file to match.
Back to the label, what to do here... I guess I'd recommend E71(4), as this is where the multiplex would end anyway if Exit 3 weren't right up in the mix being all confusing. It fits in with with E71(5) and E71(3).

Reposition E71(3) to sync up with the point on E71 and get concurrency detection working.
E71(5) is fine.

E71(2):
Wow. This is like some of the crazier parts of urban Texas. I don't envy you having to sort this out right when joining the project.
When plotting out points on these complex frontage road systems, I find it helps to ask where the points would fall if the frontage roads weren't in the picture, if the connections were made via traditional diamond ramps or equivalents.
On freeways/expressways, this usually results in point placement at a major central crossroad (often at another route in the HB if there is one) for diamond equivalents.
Half-diamond equivalents will usually be the first crossroad. The first one with a bridge, unless such is really far away. If there are multiple grade-separated crossroads close together where the ramps touch down, and a major route (also in the HB) isn't the very first one, I might still pick that sometimes, and get a graph connection.
Thus IMO JPN E71 2 is placed appropriately, centered at N26.
So E71(2) and N26, we'd want to collapse into One Point Per Interchange. Putting it at the same coords as on E71 will give a graph connection with E71 now; I presume N26 will be drafted in the future. Right now N481's file has more precicse coordinates, so I'd recommend replacing the coords in E71's file with those.
As for whether to name this point E71(2) or N26, I'm agnostic; won't push for one over the other. This reminds me of TX US183 at TX21, though I won't necessarily say what I did there is The Right Thing To Do.

[nezinscot]

Thank you both for the reviews. 

I have mostly followed your advice and have modified both the national highways and the national expressways.  Here are the changes I plan to submit.  I'm not convinced what I did with E51(10) on N295 is right, but I wanted to include the connection to E51.

https://github.com/TravelMapping/HighwayData/compare/master...denishanson:master

Should I be the one to make the changes to the E65 and E71?

[nezinscot]

Is there an accepted process for promoting a system in multiple steps from devel to preview?

I have created jpnh plus five devel-only system for the Japanese national highways, splitting the country by islands.  Should I push jpnh now to preview and sweep the other systems in individually once they are completed in devel?

[michih]

Is there an accepted process for promoting a system in multiple steps from devel to preview?

Just check out systemupdates.csv, e.g. here, here, here, here or here.

[yakra]

I'm not convinced what I did with E51(10) on N295 is right, but I wanted to include the connection to E51.

https://github.com/TravelMapping/HighwayData/compare/master...denishanson:master

Oh man. Another screwy interchange. Looking at N295 alone, it seems to have a clear center. But getting a sensible graph connection at E65 seems to beg for a point right where you have it now.
E51(10) matches this sign, though there's no other E51 waypoint, so the exit number suffix should be dropped. ..."E51" offends my sensibilities, as it directly connects to E65. Disambiguating multiple "E65" points? That could get messy too.
The more I look at this the more 1PPI (labeled E65) seems the way to go.
Then it becomes a matter of balancing the competing goals of a central point location vs. having a sensible graph connection point.
There's the "central ramp triangle" approach the manual suggests (which I'm not always a fan of); may be possible to weasel one's way out of it (like maybe even here?) on "this ain't exactly a 3-way high-speed interchange" grounds?

Should I be the one to make the changes to the E65 and E71?

Check with si404; dunno whether he'd prefer to make the changes himself or be OK with you doing it.

[si404]

Should I be the one to make the changes to the E65 and E71?

Check with si404; dunno whether he'd prefer to make the changes himself or be OK with you doing it.

I'm fine with this.

There's not much left of the jpntk review to do, once that's done, I'll activate my systems in Japan and hand over maintenance.

[nezinscot]

All routes for the islands of Kyushu and Shikoku have been added.  Honshu is next.

What I've learned so far:
-- Routes are averaging about 2.5 entries per route number, so I anticipate the entire system will have 1000+ entries.
-- Japan is mountainous so roads curve a lot. Route lengths can be 20% less than the official route length.
-- To keep the waypoint count reasonable, I tend to only include expressways, national highways, and most but not all principal prefectural roads (主要地方道, usually they have numbers less than 100).
-- Routes are averaging a waypoint every 2 miles or so.
-- Determining whether a section of road is the main route, a bypass, or an old road (original route) can be difficult.  Maps cannot be trusted.  GSV can only be trusted if the data is very recent. 

For Honshu, I've decided to do the worst first, so I have starting working on N1 - the Tokaido.  It now bears no relationship to the Tokaido of woodblock prints.  Its a 500km version of US1 through northern New Jersey, but with more trucks and stoplights. 

I've found a lot of sections where N1 runs beneath an expressway, connecting with the overhead expressway at a series of ICs.  How do I prevent concurrencies between N1 and the expressway?  Add routing waypoints to N1 between the ICs?   Nudge the N1 waypoints over a little, causing NMPs?