Thailand-Myanmar boundary

Hi all,

Since taking up bird photography I’ve been spending a lot of time in the Doi Pha Hom Pok National Park in northern Thailand. There is a road there that is used by birders to access the high ridges that separate the two countries. Looking at the OSM map one might assume that this road, the Doi Lang-Doi San Ju Road, crosses back and forth between the two countries many times. IMO, such a situation is highly unlikely considering the mutual dislike and mistrust of both governments. My guess is that the boundary is either wildly misplaced or so imprecise as to be almost useless.

Conversations with locals suggest that the Doi Lang-Doi San Ju Road (OSM Way: 139216140 in the area of interest), may be the actual boundary between the countries. The boundary has OSM id:140845202. My thought was to place the boundary and the road in a boundary relation to consolidate them. It’s not a critical issue but I feel it’s worthy of some thought and perhaps some more research.

Does anybody have information or opinions to share?

Cheers,

Dave

I’ve done this many times because there just isn’t anything else you can do. Once I determine that a village is in a chenwat on the wrong side of a boarder, I move the boarder. I’m virtuallly certain it’s the right thing to do. Same with moving boarders to coinside with rivers.

As Jimmy Wales said about Wikipedia editing, “Be bold”.

Where is the spell checker? Border. Doh!

The borders were originally some imports from a CIA database or something like that, without high precision.
With better imagery available, sometimes “border structures” can be seen. I updated then some parts of the Thai-Malaysian border.
By the way, the border roads there are not directly on the border, but a few meters away, e.g.
https://www.openstreetmap.org/way/244715284#map=14/6.6837/100.2904
South-east from Padang Besar, border roads are on both sides.
Hence I would not use the road in the border relation.

There should be government official documents available describing the border. Probably in textual for like “in the middle of the river xxx” or similar.
Getting such non-physical boundaries right is extremely tricky. If it is reasonable to assume a border follows a river, then updating it is probably better than the old import from the old days where landsat was the typical imagery.

But where can we get such documents?

Thanks all for the responses.

I assume that in Thailand or Myanmar, bilateral state treaties are made public in some form.
Not certain whether they are available online.
You should be able to look them up in some government agency and probably also be able to make copies (potentially for a small fee).

This is currently out of scope for my language level of Thai. Others might be able to come up with documents. Maybe Mishari knows where to search, as he worked with some of the open-data before.

Here you can read up how other countries deal with border definitions, for example the longest land border:

https://en.wikipedia.org/wiki/Canada%E2%80%93United_States_border

As you can see, even they have disputed areas, which is then another obstacle you might face. There is a specific tagging on how to mark disputed borders.

I turned up a good source of Thailand shapefiles for all administrative levels just now. It is located here:
http://thaigis.net/thailand-gis-resources/. From that page I jumped to the first link which took me to the UN-OCHA Humanitarian Data Exchange Project main page

On that page, I downloaded this ZIP file (https://data.humdata.org/dataset/thailand-administrative-boundaries) which contains administrative level 0 (country), 1 (province), and 2 (district) boundaries, unzipped it and loaded the Admin level 0 SHP into JOSM. As I suspected, the boundary OSM has in the area of interest is poorly positioned in many places. In other places, it’s fairly accurate. Interestingly, the boundary runs right through the center of some of the Thai Army camps that guard the border. I did not see any instance of the boundary crossing the Doi Lang-Doi San Ju Road.

Unfortunately, the data is not without usage limitations. It is administered by the UN-OCHA Humanitarian Data Exchange Project, whatever that is. It is quite possible that we could get permission to use their data but I did not take that step.

The name UN-OCHA Humanitarian Data Exchange Project sounds as though exchanging data with them (they use OSM on the website) should be easy. But, maybe not. If I decide to spend some time researching this, I’ll let you know.

This would be a wonderful resource for Thailand OSM mappers. Shapefiles are available for landcover, parks and other man_made features. It’s a virtual treasure chest!

Cheers,

Dave

Edit: I might have been too enthusiastic about that website. I returned to it and couldn’t find parks or landcover on it. I must’ve seen those elsewhere. Still, accurate admin boundaries would be nice to have.

There’s a series of US Department of State documents from the 1960s called the International Boundary Studies. Most can be found on the web with a Google search, if one knows the keywords. A copy of the one describing the Thailand-Myanmar border can be found at https://fall.fsulawrc.com/collection/LimitsinSeas/IBS063.pdf . Mostly, the border will follow rivers or mountain ridges, so if the terrain is clearly visible enough in the imagery one can draw a reasonably accurate representation by following the zigzagging spines of the mountains. The level of detail, though, probably isn’t going to be much better than that of the LSIB data (the source of the latest import). We’re talking inaccuracies of mostly less than 100 m; this will require visible detailed features such as fences (as with the Malaysian border mentioned by Bernhard Hiller) or, as AlaskaDave noted in the OP, roads. (I wouldn’t actually expect to find entire villages on the wrong side with this data source, though.)

Personally, I’ve only manually adjusted borders where the visible features are very obvious, e.g. there’s a fence, or a pair of roads that clearly belong on different sides. A single road seems rather difficult to gauge—does it run exactly alongside the border, or how far away is it (or even whose side does it belong to)? In this case, the on-the-ground information would indicate that this road should be on the Thai side, so I guess it should be safe to adjust accordingly.

PS If anyone feels like it they could try checking the LSIB (Large Scale International Boundaries) to see if the latest version is more updated/accurate in the relevant areas (though I don’t think it’s likely).

Edit: Oh, just re-read the OP and saw that the suggestion was that the road exactly corresponds to the boundary. I’m not so sure about this; seems like a rather strange arrangement.

Getting the state border right is a sensitive topic.

Using some US source to define the border of a different foreign state doesn’t sound right.

The initial import into OSM only had very approximate borders. Most involved had been aware.

If updating them now, we should use a better source.

Quite right, Paul. It was only a starting place for my investigation. I thought it very unlikely that the boundary crisscrossed the Doi Lang-Doi San Ju Road and that started me looking for an explanation or some other way to verify my suspicion.

Subsequently, I added a comment that talks about the accuracy of the updated boundary I found online and indeed, the boundary lies directly alongside the Doi Lang-Doi San Ju Road in several places but it’s always on the Myanmar side, and it never crosses that road. The people that suggested the boundary follows the road were only approximately right.

I did a little more probing of the HDX site I found and it contains the following statement:

“Except where otherwise noted, content on this site is licensed under a Creative Commons Attribution 4.0 International license.”

What does this mean?

Hello,

I wrote a short email asking the HDX people about the use of their data and they in turn referred me to the organization that supplied the data to them. I contacted the United Nations Office for the Coordination of Humanitarian Affairs (OCHA) in Bangkok and they said the data was available to us. Here is the response.

Dear Dave,
Thank you for your interest in the data we posted on HDX.

The data originated with the Royal Thai Survey Department, and was acquired by the International Committee to the Red Cross (ICRC) Regional Office in Bangkok. It was then given to OCHA, polished up and posted on HDX.

As far as we are concerned, the data is freely available for humanitarian, or at least non-profit purposes. This would of course include OSM.

We hope you enjoy! And please do let us know if you come across national admin boundaries that are better than what we’ve made available.

Best,

John

John Marinos
Regional Information Management Officer
United Nations Office for the Coordination of Humanitarian Affairs (OCHA), Bangkok
Mobile: +66 081 912 9853 | E-mail: marinosj@un.org | Skype: john.marinos

So it appears that we can use their data to correct the boundary in the area of my current interest. Are we all okay with this permission or does it need to go higher in the OSM hierarchy for approval?

My reply to John’s email follows:

Thank you very much, John

I will present this email to our local group of OSM mappers to determine what our next step will be. OSM wants to take a very conservative approach to using data we did not produce ourselves. But I expect that we’ll start using this boundary data as soon as possible.

I want to mention that OSM isn’t able to restrict the use of its data in any way. That is to say, while OSM itself is a non-profit organization, it makes its data available through a very liberal Creative Commons License which makes that data available to any person or organization, non-profit or otherwise. That said, we certainly thank you for your cooperation and speedy reply.

Best,

Dave

Comments, questions, or concerns?

Hello,

you could send a note to the legal-talk mailing list:
https://lists.openstreetmap.org/listinfo/legal-talk

My judgement as a non-lawyer would be that it is not suitable, given they state " the data is freely available for humanitarian, or at least non-profit purposes."

OpenStreetMap is neither “humanitarian”, nor “non-commercial”. OSM data is used for all sorts of commercial purposes. Mapbox is selling the data as vector tiles. The OSM foundation is also registered as a commercial entity. No one can prevent me from extracting the Thai border and selling it for $$$ licensed as ODbL.

I also have the impression that the HDX people are not the owner of the data. They just forwarded data they received without clarifying the extent of the license.

So the right thing todo would IMHO be to contact the Royal Thai Survey Department and ask them for either a confirmation that their data can be included in OSM or even better asking them for updated data as well, maybe in colaboration to importing it. Potentially they have other boundaries available as well. Province boundaries or even smaller-scale.

Correct projection and other technical details to be considered before import is then a follow-up topic.

I reckon you’re right, Stephan.

I was afraid this sort of usage and rights controversy would be the result of my inquiry. I’m not anxious to communicate with the Thai government to try to secure rights to this data. My dealings with Thai Immigration over the years make me shy away from attempting to do anything as complicated as a request for data sharing. I hesitated before contacting the HDX people because I suspected that the permission, if granted, would not be unambiguous enough for the OSM high command or its legal department.

It’s a pity too because the data appears to be very good. You mentioned other boundaries and as it happens this dataset includes provincial and tambon boundaries along with the national boundaries. It would be very valuable data indeed.

Maybe I’ll give the legal department a shot, I dunno yet. I have no idea how to contact the Royal Thai Survey Department and no desire to get caught up in the can of worms that would likely ensue from any such contact.

Before spending extra time with license clarification: Can you confirm that the accuracy of the border polygons is better then what we have? Updating borders is a huge effort as we have to do it manually step by step to not come into conflict with existing features along it.

How would one do that? I mean, this is Thailand after all.

I am certain that the boundary does not jump back and forth between the two countries along the Doi Lang-Doi San Ju Road the way the present boundary in OSM does. Other than that, I have no way to guarantee the new boundary is more accurate than the old one.

Correcting the present boundaries is easy if you do it piece by piece. Note that the present boundary isn’t just one boundary. It contains many sub-boundaries in addition to the one for the two countries: districts, tambons, national park, wood, etc.

To make the operation easier I just selected a portion of the HDX boundary from the imported SHP layer, copied it, used “paste at original position” into my working data layer and then aligned the present boundary to it. I didn’t replace anything; I only used the HDX boundary as a guide after which I deleted it.

I didn’t upload my revised boundary because I knew the license situation had to be resolved first.

And that’s where the issue stands today. I’m not going to pursue getting permission from the Thai government. Having to deal with them on Immigration issues is enough to discourage me before I even start.

To summarize: I believe but cannot prove that the HDX boundaries are more accurate than what we have. The next step is to contact the appropriate Thai agency to ask for permission to use the data. But I’m not going to do that.

Cheers,

Dave

I think the validation is exactly like you did. You open it in an additional layer in the editor and check with existing geographic features we think indicating the boundary (like rivers or mountain ranges) to see whether it seems to follow it.

The actual update certainly has to be done manually, probably broken down to individual sections and existing features aligned to it.

As it seems useful to have this data in OSM: Can one of the native speakers take here the lead?
As Dave already pointed out: We want to avoid that someone is into trouble in case our hobby is suddenly re-classified as “work” and work permits regarding OSM are on the table. Communincation in native language makes is also easier to get the right “tone” in the conversation, as we are asking them to clearly state that the data can be used within OSM under ODbL. This would include the permission that someone takes the data out of OSM and sells is. Not that this would be most clever business model, as data is openly available under ODbL, but legally it is possible.

Administrative boundaries are indeed the type of data that we should find a way to get imported. Apart from the national borders (where the current data, even with the concerns found in this thread, isn’t that bad), there are the provincial boundaries, which are currently based on a very inaccurate old VMAP0 import, and those of amphoe and tambon, which are still almost entirely missing, except for the few cases where they’ve been manually traced by hand (and it’s a laborious, time-consuming process, probably not viable for the entire country).

I had earlier tried looking into this, and so has Mishari to a greater extent, but while the data is easily available enough, their legal status and possible licensing options are, well, pretty much obfuscated. As Mishari commented in another thread,

It’s not the brightest of prospects, but maybe that could change. There’s always room for hope.

a user recently added a disputed section of the boundary. The country polygon is no longer valid in postgis.

https://www.openstreetmap.org/changeset/101429571#map=11/9.6996/98.3313