National Highway Mapping

Let us not tag like this just for the rendering. In the custom India style on http://openstreetmap.in/ we can render both the ref and ref:old tags to give the same effect. Tracking this here: https://github.com/osm-in/openstreetmap.in/issues/8

Maintaining ref tags with both old and new numbers like NH66/17 will be a lot of headache.

I got a feedback from a customer saying that NH48(changed recently) is NH8. According to http://wiki.openstreetmap.org/wiki/National_Highways_%28statewise%29#Delhi, NH48 is the correct name, but according to MORTH NH8 is also correct. I’m confused with which one to follow.

Changing to the new ref may confuse users since the signs of the old ref still exist on the ground. We can follow this procedure to name the roads with multiple or changing names to render them on map. We can remove the old ref if new signs are installed.

I have changed the NH8 in Delhi back to NH48. Except old NH220; all NH numbers are changed back to new numbers now!

  1. The government has declared 2010 that the old NH8 is now called NH48.
  2. We cann’t have two different NHs with the same number. The NH8 is now in Tripura and Assam
  3. The new NH148,NH248… start according to the new numbering system from the NH48. NH8 makes here no sense. And MORTH continues to declare a lot of new NHs; all according to the new system.
  4. If we use NH8,NH48,NH8… according to the actual signboards, who will update this, if a new signbord appears on the road???

I don’t know why the government still uses a mix of old and new numbers.
We could do it like parambyte suggested.
But PlaneMad is right : I think, we should remain with the new numbers. And it’s a good solution if openstreetmap.in shows the old and new numbers. I hope this will be realized.

I have finished the Wiki-page with the statewise list of all Nhs.
http://wiki.openstreetmap.org/wiki/National_Highways_%28statewise%29#National_Highways_.28statewise.29

Now I try to find the reasons, why there are differences to the MORTH-list.
In Andhra Pradesh, I think, all NHs are completely mapped in OSM.
Nevertheless compared to MORTH 356km are missing!?
MORTH list NHs in Andhra Pradesh, which definitely are in other states (old NH222, old NH16, NH150, NH202, old NH222, NH161, old NH326, NH353)!!!

Superb Heinz :slight_smile: 97.7% mapped :slight_smile:

Superb work Heinz :slight_smile: . Coming to the issue of Andhra Pradesh, it is recently divided into two new states (Andhra Pradesh & Telengana) by Indian government. The government of India is still facing problems in differentiating the division of NH’s in these two states. This might be the reason for missing 356KM in Andhra Pradesh.

Hi Heinz, I have called and written to Maya Prakash, She is the director for planning in MORTH. They told they will check this. Hoping for right reply.

Dairy post on calculating road lengths from OSM: http://www.openstreetmap.org/user/PlaneMad/diary/37790

Heinz, would be great if you could share your process too. Also tips on how to work with such massive relations, even with JOSM I find this quite a mess.

The reason for all this relations is :
a) old and new numbers for the NHs
b) contradictory and unprecise publications about NH length.

The NHs now have the following relations:

  1. relation with the old number („network=IN:NH“)
  2. relation with the new number („network=IN:NH“ and a note “new number”)
  3. relation for the statewise length (with „network=IN:NH:XX“)
    parts, where two NHs share the same road have a separate relation (here „ref=NHXX;YY“)

The main problem with relations are:
Newbies delete a road and map it new; all memberships to relations are destroyed.
Better : divide the street in two parts ( one part at the beginning, short); delete the long part; continue mapping the short part. So all relations are still correct.

With the statewise relations we can exactly calculate the NH-lengths. I have controlled now upto Madhya Pradesh.

Results upto now : the Morth-list is full of errors; we have mapped 98-99% of all NHs.

PlaneMad : run your script once more, so comparing to my results is easier and maybe we find the script errors.

I have finished the control of statewise NHs.
So I can promise : no more relations!

We have mapped 97991,2km

Not mapped are about 800km (NH215,715A,333B,566,133B,137,148B,254, 108,731,702,306A,707A,709A,11, 334A).

So the total length of all NHs is about 98791km (without expressways) and 99.2% are mapped!!!

I tried to find reasons for the differences to MORTH in Gujarat :
a) in OSM all Nhs are mapped with high precision
b) Morth list the NH58, but the NH58 is in Rajasthan
c) I couldn’t find an explanation for the big differences.

In the other states the situation is similar.

Maybe there are additional NHs which are defined the last two years and are actually not in OSM and MORTH-list. The publications in gazette of India must be controlled.

Actually there is a little edit-war between me and another mapper. He insists that the new NH116 in West Bengal is mapped as NH41(old number). What can be done???
Is it possile that openstreetmap.in renders the old numbers?

Total length of all Nhs in OSM = 99048.8 km

Included are :
97824,3 km : Nhs with new numbers
1030,9 km : Nhs wthout a new number
193.6km : Nhs without any number
All defined Nhs are included ( even the lately defined).

Not included are:

about 740km : not mapped Nhs in OSM
about 80km : planned, yet not existing Nhs
?? : streets in big cities, which are mapped as trunks
trunk_ links at big NH crossings
expressways, which are mapped as motorways

I created for each state a gpx-file of all Nhs. If someone is interested, I can send it.
According to this data we have mapped about 99,3% of all Nhs.

I have updated the length in the list with new numbers.

Which is the longest NH in India??

NH27 = 3536.9km NH44 = 3536.0km

I have integrated the newl defined NHs (20.1.2016). Changes were necessary for NH9, 34,334,334C.
Now we have additional 57km NH without a new number (Ghaziabad to Meerut) ???

I have controlled the mapping of motorways in India. At the moment the following roads are mapped as motorways:

Dehli-Gurgaon
Yamuna Expressway (Dehli-Agra)
Allahabad Bypass Expressway
Kona Expressway (Kolkata)
Mumbai-Pune
Western Expressway(Mumbai)
Eastern Expressway(Mumbai)
Hyderabad: Outer Ring
Bengaluru : Outer Ring
Bengalura : part of NH48
Bengaluru : part of NH44 (Bellary Road)
Chennai Bypass Expressway
Chennai : Outer Ring Road
NE1= Ahmedabad-Vadodara

The Wiki of India says :“Note: use this highway classification for greenfield projects not for upgraded National Highways in India!”
Is this accepted??? Not all listed motorways fullfill this criteria!
We should find a consense, what are motorways and also what are trunks with motorroad=yes.

We have three new Nhs.

NH563 : I have extended the NH to Khammam. (MORTH still has problems where Telangana and Andhra Pradesh are!!)

NH717A and NH717B in Sikkim. Are there any mappers with local knowledge? I cann’t find these Nhs.

Dear Heinz,

in my oppinion are the following roads expressways:
Yamuna Expressway (Dehli-Agra)
Mumbai-Pune
Western Expressway(Mumbai)
Eastern Expressway(Mumbai)
Hyderabad: Outer Ring
Bengaluru : Outer Ring
Chennai Bypass Expressway
Chennai : Outer Ring Road
NE1= Ahmedabad-Vadodara
new Agra Lucknow Expressway

I had long discrussions with local mappers regarding this points. But most mapper dont understand the difference between trunk with motorroad=yes and a real motorway. In the advertisment of india is every new width road an expressway and this confuse sometimes new mapper.

I reverted the Allahabad Bypass while it is part of the golden quadrilateral Delhi - Kolkatta and this project is classified as trunk.
With best regards

Oberaffe: OK to your opinions about motorways.

I had a look which trunks are mapped as trunks with motorroad=yes.
After some corrections the following overpass-query shows the actual situation :
http://overpass-turbo.eu/s/eyy

It’s necessary to control this.

Amaroussi made in TALK-IN suggestions in the thread „National Highway relations are unacceptably huuuuge!“

Because I have created most of the mentioned relations I want to tell here my opinions. (I’ m not subscribed to TALK-IN.)
Amaroussi wants to install a system „system similar to the E-Road network“.
This system doesn’t exist in Europe. Some highways are mapped with such relations (E75 and some other), but only because it was the idea of the single mapper. There are still a lot of highways in european states with more than 2000 members.
Amaroussi wrote : „the main OpenStreetMap site is refusing to load the main NH44 relation“
That’s right, but that’s a problem of the actuall server-capabilities (causes time-out). After the next server upgrade this will surely be resolved.
Amaroussi wrote :“editing them is also slowing down JOSM significantly“
I loaded into JOSM more then 10 000km of NH-relations of India and I don’t notice a slowing down???

Planemad wrote :“ Have always found handling and modifying these relations painful.“
Mapping in Europe, where the map is much more complicated than in India, I find easily mapping situations where changements are difficult and where I don’t make edits because of missing knowledge. And I’m a quite expirienced mapper.

So I don’t agree with the suggestions of Amarousii and Planemad.
I made the two wiki-pages (NH with new number and Nhs statewise), controlled and updated them up to now.
The result is : We have a list of all mapped Nhs in India with a accuracy MORTH can only dream to have. Specially we have controll over the exact length of all mapped Nhs in each state.
Isn’t this a strong argument against the proposed nonsense bill. The „illegal“ OSM offers better results than the ministry!
The results in the statewise-page are only valid if the relations are mapped as described.
Now the relations of NH765 and NH44 in Telangana don’t show any longer the correct length.
Amarousii (not a Newbie) has added both directions!
If the local Indian mappers want the proposed system, I disagree but will accept it.
There are not so much rules in OSM. One is : „ You can map what you want“. Only vandalism will be reverted. And the proposal isn’t vandalism.
But you must understand that I won’t update this pages any longer. Then that’s the task of others with „better“ ideas.

Hi,

I am not sure how to interpret the attitude of Heinz_V (the phrase “I won’t update this pages any longer” seems a bit too knee-jerk), but according to https://wiki.openstreetmap.org/wiki/Relation:route#Members, it has been practice for road relations to contain all members of the highway, rather than in one direction only.

According to the wiki, the slip roads are given appropriate the role “link”.

What I will do for now though is review the NH765 relation and remeasure it as appropriate. I believe the average length, is the full length of the ordinary members plus half of the length of the forward/backward members, but I may have fallen behind with what the standard was since the mailing list is my main port of call.


Update: 16:04

Conditional on a consensus, we could adopt a route_master structure for the NHs and SHs. It would allow each direction to have its own relation and then we can easily get the average distance if measurement matters. However, I am not sure if we would then need the state-level relations any more.

Amaroussi : Thanks for reverting your edits.

„a route_master structure“ : I think thats not the right way. It’s for other purposes (bus lines…).

„It just seems odd that only one side is measured.“:

Our important references in India are the publications and decisions of MORTH (ministery). MORTH measures like I do in the statistic.
When I created the statewise relations, we didn’t know, to which grade the Nhs have been mapped. I added a lot of missing Nhs.
Now we have a good overview about what has been done and what is missing.
And we have the strange situation that our statistics are much better than MORTH.
For example in Madhya Pradesh : there are about 8100km of Nhs, but MORTH means there are about 5200km!!!
But there are still Nhs which are not mapped in OSM.
So for a longer while we need the statewise statistic.

The reason for the length problems is : In India Nhs are mapped as trunks. And a street is a NH if it is declared by the ministery.
In Germany there is no trunk without double carriageway or direct crossings with other streets. So here it easy to calculate the length by division by 2.
Approximately 70% of all Nhs in India don’t have double carriageways and furthermore there are a lot of way-segments which are shared by two or three Nhs. So length calculations are quite difficult.

The mapping density in India will increase. So the number of members of the nationalwide relations will increase also. The fact that too big relations can’t be displayed in the browser is not so important. In JOSM they can still be edited. In a few years we should examine if changes have to be made.

Besides the relation problem : Concerning the Nhs there is a lot of other work to be done :

  • bridges are missing
  • a lot of exits are missing
  • at trunk junctions trunk-links are not complete
  • trunk exits are mapped in many different ways (trunk-link, unclassified, secondary…)
  • the newly constructed second carriageway for many Nhs are not completely mapped
    and so on.
    So there is enough to do.

We have now about 125, 000 km of NHs mapped in India. That means an increase of 27,000 km in two years.
I have controlled all NH-relations. Using Osmose I added missing bridges, tunnels and connections.
Only one relation has been deleted. Thanks to all mappers!
I did not look at the uncountable ammount of errors beside the NHs!
A problem is still the NH183. Mappers in Kerala again renamed it to NH220.
The NH 220 is this : NH20 near Chaibasa, Gobindpur, Hata, Tiringidihi, Rairangpur, Jashipur, NH20 near Dhenkikot
The NH starts at NH83; the NH183A starts at NH183 and not at NH220!
Long ago I proposed to use the name-tag for the old numbers. So Mapnik shows both numbers.

By the way the longest NH is for sure NH44 (3663KM); NH27 has only 3538km