Cmod Political borders model?

[Hungry4info]

Something of concern, there isn't really a definite number of countries in the world. There's at least one oddity, Taiwan, for example, but it's an island-"nation", so you might be able to get away with it =P.

[granthutchison]

Folks:
I have current border data, as of the (disputed) independence of Kosovo this year, in the form of lat long pairs in text files. Much work was involved in extracting new national borders from old provincial borders to keep the data current. It's what I use to maintain my texture overlay, hosted by selden.
Anyone is welcome to the data if they want to convert them to CMOD.
They'll give you the Central Asian republics, the fragments of ex-Yugoslavia, and the united Germany, as well as East Timor and numerous small enclaves (Llivia, Indo-bangladeshi, central Asian republics).

Grant

[Cham]

Grant,

please, send me the data by email. I'll make a CMOD version of it. Is the data very precise ?

[granthutchison]

Is the data very precise ?

Same ultimate source as yours (the World Data Bank II), except with some finer detail taken from the World Vector Shoreline in order to fill in the new national borders.
I have it formatted in lat-long pairs, with one-line headers identifying each border segment. The headers should be easy enough to strip out or otherwise ignore: my own program to generate the overlay texture does that.

Grant

[Cham]

Grant,

then, can you send it to me by email (in chunks of less than 10 MB), or a download link ?

[granthutchison]

OK, here are zipped text files for download by anyone who's interested:

africa.zip (373 KB)
americas.zip (222 KB)
asia.zip (563 KB)
europe.zip (373KB)

The file names are self-explanatory. Data are presented as tab-delimited pairs of coordinates, longitude then latitude (decimal degrees; west and south are negative), with a line break after each coordinate pair.
Headers are short text strings I can explain to anyone who's interested, but which are always less than 10 characters in length, whereas the data lines are always greater than 10 characters: to filter the headers, reject all strings shorter than 10 characters.

Grant

[ElChristou]

Folks:
I have current border data, as of the (disputed) independence of Kosovo this year, in the form of lat long pairs in text files. Much work was involved in extracting new national borders from old provincial borders to keep the data current. It's what I use to maintain my texture overlay, hosted by selden.
Anyone is welcome to the data if they want to convert them to CMOD.

I knew some data were around the corner! Many Tx Grant!

[Cham]

Thanks Grant,

I'll convert all the data to CMOD tonight. Should be fast to do. However, the files are pretty small. I guess the borders resolution is much less high than with my previous data. I'll see this later.

[granthutchison]

I'll convert all the data to CMOD tonight. Should be fast to do. However, the files are pretty small. I guess the borders resolution is much less high than with my previous data. I'll see this later.

Then your files are much larger than they need to be, I'd suggest.
The original data show spurious precision (comparison with current SRTM topo data and between the WDB II and WVS datasets suggests an accuracy on the order of 100m), as well as containing many, many duplicated points, so I filtered them accordingly. I didn't bother trimming the trailing decimals, I confess, since that wouldn't alter the floating-point processing rate, and I didn't anticipate ever having to transfer the files electronically.

Grant

[Cham]

Here's a first quick model made with Grant's data. Of course, the color isn't final. This model is using dots only (instead of straight lines between each pair of dots) and isn't satisfying since there are many holes and gaps in the curves. I compared the curves with the model I made with the soviet era, and the match is very close (almost the same, except europe and a few cases) :

world.jpg

Next, I'll make a version with straight lines. It should be better.

[Cham]

Okay, the model with straight lines is superb. Almost finished.

But what are these small countries in Asia, in the north of Pakistan ?

strange.jpg

See the tiny red blobs near the center of that picture ? What the ???

[bdm]

Okay, the model with straight lines is superb. Almost finished.

But what are these small countries in Asia, in the north of Pakistan ?

It looks like an error to me, because those pieces of land look like they are in Kyrgyzstan.

EDIT - I found what these are. These are foreign enclaves in Batken province in Kyrgyzstan. The one of the left is the Tajik Voruh enclave, and the larger one on the right is the Uzbek Soh enclave.

Here is a map:
http://www.reliefweb.int/rw/fullMaps_Sa ... penElement

[BobHegwood]

It looks like an error to me, because those pieces of land look like they are in Kyrgyzstan.

Same thing I thought. See HERE.

[ElChristou]

EDIT - I found what these are. These are foreign enclaves in Batken province in Kyrgyzstan. The one of the left is the Tajik Voruh enclave, and the larger one on the right is the Uzbek Soh enclave.

Yep, you can also see them in Google map: http://maps.google.com/maps?hl=fr&num=10&as_epq=Kyrgyzstan+map&as_oq=&as_eq=&lr=&cr=&as_ft=i&as_filetype=&as_qdr=all&as_occt=any&as_dt=i&as_sitesearch=&as_rights=&safe=images&q=%22Kyrgyzstan+map%22&um=1&ie=UTF-8&sa=X&oi=geocode_result&resnum=1&ct=image

[granthutchison]

Yes, I have a special interest in enclaves. You'll find a number of them scattered around Europe and Asia, which I retrieved from the WVS datset.
Take a look along the Spanish-French border, and you'll find the Catalan town of Llivia, surrounded by French territory. There are others.

Grant

[Cham]

Isn't the world strange !?

I'll finish the world model today. I've found some small defects here and there, when three lines are joining (mostly in Europe). I'll try to correct them.

Please guys, suggest me a nice color for the borders.

[ElChristou]

...Take a look along the Spanish-French border, and you'll find the Catalan town of Llivia, surrounded by French territory...

First time I hear about this one!

Cham, beware not deleting any good data! (from what I saw they seems to be VERY good!)
Now, whatever color is fine and one can always edit the cmod to change it, no?

[Cham]

Cham, beware not deleting any good data! (from what I saw they seems to be VERY good!)

There are many very small defects, like these :

defects.jpg

The difficulty is to locate them in the numbers list (coordinates) and join the lines.

Lets not forget the lines accuracy is about 1 km, and the deffects shown here are at the same scale. So changing a small straight line to join the curves will not affect the accuracy of the map.

[Cham]

Without an adaptive planetographic grid, it's extremelly difficult to find the coordinates of the defects.

Please Chris !

[granthutchison]

Without an adaptive planetographic grid, it's extremelly difficult to find the coordinates of the defects.

In general, they should be easy, because they're at tripoints, and the segment headers code the countries on each side of each border. The mismatch comes from using two datasets, both of which have spurious precision, so that the end of one border doesn't precisely tie to two other ends. The "offending" coordinates are therefore in general at the start or finish of a border segment where WVS meets WDB II. The little loop in your fourth image is a surprising exception, though, and looks like a mismatched subseries of line segments. Which three countries are meeting at the central tripoint shown?
Personally, I'd just accept the mismatch at the tripoints, because the "correct" location of the tripoint is unknown to the precision the datasets use. But if it's causing you distress, let me know which tripoints are involved, and I'll see if I can get around to "fixing" the junctions. Might take me some time before I can fit it in, though.

Grant