[Update 2009-09-10: Aruchanal Pradesh’s placenames are now back to normal in Google Maps.]

China and India are two of the globe’s most touchy countries when it comes to how their borders are depicted on unofficial maps — in fact, in both countries it is illegal to publish maps where the borders deviate from official “guidance”. This makes the border between India and China quite interesting, especially as large chunks of it are disputed.

There is a pleasing symmetry to these disputes: In the west, Aksai Chin is claimed by India but administered by China. To the east, most of Arunachal Pradesh is claimed by China but administered by India. These are huge territories, so you might think there is plenty of room for a negotiated compromise — and you’d be right; just this Friday China and India sat down for yet another round of talks — their thirteenth, in fact. Differences are “narrowing”, according to Reuters.

In the past few days, Ogle Earth began receiving a surge of Google Search-referred visits from Indian IP addresses to older posts about Arunachal Pradesh, mostly on variations of searches for “Arunachal Pradesh Google”. Google had previously made an error regarding that region’s disputed border in Google Earth, but had fixed it quickly. Has Google now messed up again, at a particularly sensitive moment in time?

In a word, Yes. Indian bloggers and media this past Thursday noticed that many place names in Arunachal Pradesh had turned Chinese in Google Maps, depicted both in Chinese characters and their Anglicized equivalent. For example: Around Tawang, home to the famous Buddhist monastery near the Bhutanese border, place names are now written like this:

All place names north of China’s claim line are depicted thus. Place names to the South are mostly in Hindi, with an Anglicized equivalent:

Those are indeed aberrant depictions of place names, if Google intends to be neutral about a region administered by India but claimed by China. (Place names in Aksai Chin are in Chinese, for example.) And sure enough, Indian bloggers, reporters and even politicians were not particularly inclined to give Google the benefit of the doubt, not 24 hours before the start of those “high-stakes” border talks. Times of India quotes one MP:

”This is shocking. How could Google change the names of places of a sovereign country without the country’s knowledge?” fumed Congress’s West Arunachal MP Takam Sanjoy.

Etc.

But then Google reacted. To their credit, they quickly admitted to a mistake in a statement released to the media and republished by many Indian media sites on Saturday:

Earlier this week, as part of a routine update to Google Earth, we published new data for the Arunachal Pradesh region that changed the depiction of certain place names in the product. The change was a result of a mistake in our processing of new map data. We are in the process of reverting the data to its previous state, and expect the change to be visible in the product shortly. We would like to clarify that this issue did not impact our depiction of international borders.

(A quick fact check: It wasn’t place names in Google Earth that were affected — they still depict only Anglicized versions of Hindi names. It’s in Google Maps that the error occurred. Google Earth does show some user-generated Panoramio photo icon labels in Chinese in Arunachal Pradesh, but there is nothing wrong with Panoramio users writing in their own language.)

Someone identifying themselves as an employee of Perfect Relations, India’s self-described largest PR firm, even posted the above statement “on behalf” of Google in the comments section of one of the Ogle Earth posts currently receiving all that Google Search traffic. This seems to be a new strategy, aimed at quickly quenching any budding outrage in the (Indian) blogosphere, by taking the conversation to the blogs themselves.

Now it is just a matter of waiting until the names are corrected. But will that end the outrage? The problem is that when those Indian bloggers and journalists went looking on Google Maps, they also noticed that both the Indian and Chinese claim lines for Arunachal Pradesh are drawn in dotted lines. Neither is given the prominence of a national border, and Google does not play favorites, just as it doesn’t with Aksai Chin, where borders are also drawn in dotted lines.

In Google Earth, disputed national boundaries are perhaps even more clearly demarcated, as a red line instead of a yellow line, and annotated as well:

This is, of course, as it should be. National borders in Google Earth that are not bilaterally agreed to are in red, no matter how tenuous the claim. Good luck telling that to the rabid fringe among India’s bloggers though — and I’m pretty sure that is why Google’s statement contains the preëmptively plaintive line, “We would like to clarify that this issue did not impact our depiction of international borders.”

Google’s policy regarding disputed borders has, however, progressed since late 2007, when it last made a cartographical error in the region. Here is how Arunachal Pradesh’s border looked when Google made that mistake in early November 2007:

And here it is after it was corrected, in November 2007:

For at least the last few months, however, Google has been showing both claim lines:

Additionally, “lines of control”, such as the one between India and Pakistan in Kashmir, are in orange. I approve, because I am for more information, not less. Even if you don’t agree with the validity of a certain claim line or de facto control line, you should at least want to know where that line is. Maps that don’t depict such information for political reasons do their users a great disservice.

And Google can’t be faulted for not trying to bring as much nuance and context as possible to its maps. Here is the popup box for the main place name of Arunachal Pradesh in Google Earth:

One final question: How might an error like this have happened? Google in fact maintains two sets of Google Maps databases: One for the proper internet, and one for users in China. This latter map does depict Arunachal Pradesh as being integral to China, and with identical Chinese names as the one that accidentally populated the global Google Maps. (For example, Tawang on ditu.google.cn vs Tawang on maps.google.com).

The reason for the Chinese Google Maps is simple — Google serves them from Chinese servers, which are subject to Chinese law, and it would be illegal for Google’s Chinese business unit to depict Arunachal Pradesh in any other way. And somehow, data intended for the China map must have ended up in the global map. (It would be interesting to know how:-)

Could India demand a Google Maps for Indian users that depicts the world’s borders as India’s government decrees them? Google Maps isn’t served from inside India, because India doesn’t indulge in anything as offensive as censorious China’s great firewall, so the short answer is No, unless India’s politicians were to change the law and start making life difficult for Google’s Indian business units. [Update 14:47 UTC: Scrap that. maps.google.co.in shows both Aksai Chin and Arunachal Pradesh as being integral to India. I guess that map is served from inside India and/or Indian politicians have already gotten their way. But on the Indian Google Map, too, placenames are Chinese both in Arunachal Pradesh and Aksai Chin.]

(Just to be clear: You can currently access maps.google.com and Google Earth from China, but only at the pleasure of China’s censors, and with a performance hit due to the great bottleneck that is the great firewall.)

• Google LatLong: New data in Google Earth show wider swath of destruction in Darfur
  Google LatLong: US Holocaust Memorial Museum's Michael Graham writes about the updated Darfur data in Google Earth

Regarding the probable killing of Taliban leader Baitullah Mehsud: The BBC reports that:

The missile fired by the US drone hit the home of the Taliban chief’s father-in-law, Malik Ikramuddin, in the Zangarha area, 15km (9 miles) north-east of Ladha, at around 0100 on Wednesday (1900 GMT Tuesday).

Ladha was easily found via a default search in Google Earth (and the returned placemark comes courtesy of an entry in Google Earth Community); it is indeed in the South Waziristan region of Pakistan, and is the location of an old British fort. Then it was a matter of finding Zanghara. Here Geonames.org once again proved to be an invaluable resource. A fuzzy search for “Zangarha” in Pakistan returned one clear answer exactly 15km northeast of Ladha: Zangarai Algad, where “algad” denotes that the feature name is a stream or shallow valley. This valley extends northeast-southwest for a few kilometers, and I think it is likely this is the place referred to by the article.

The location can further be confirmed by looking up the purported burial place of Baitullah Mehsud. Reuters reports:

“He was killed with his wife and he was buried in Nargosey,” the [Pakistani] officer said, referring to a tiny settlement about 1 km (half a mile), from the site of the attack, believed to have been carried out by a pilotless U.S. drone aircraft.

(Dawn and the BBC refer to the village as “Nardusai”, but such discrepancies are frequent when transposing names to English.)

Browsing with Geonames.org in the immediate vicinity of the Zangarai placemark, it only took seconds to chance upon “Nargosa” village, one of the very few inhabited places in the remote and hilly area. Such successful triangulation of place names makes it quite likely in my mind that this is the place referred to.

Serendipitously, this spot has high-resolution DigitalGlobe imagery, taken on January 7, 2007. Individual houses and fortified compounds in the valleys and villages are clearly visible, and invite perusal. Barring any help from the US or Pakistani military or GPS-equipped journalists escorted into the area, it is unlikely we’ll get a more precise pointer to the targeted location.

Here is the file with all the relevant POIs referred to in this post. If you use them with the Geonames auto-refreshing network link linked to above, you can browse the area amid a wealth of obscure place names called up from Geonames’ database on the fly. If you want to know what kind of terrain the Taliban have as their home turf, take a look.

• techblog86 » Blog Archive » China Launches Investigation Into Maps
  China to punish "illegal" online maps of China showing military places, independent Taiwan, etc. Google beware (again)

• GeoNames Fulltextsearch : Naung Laing
  As usual, Geonames.org was my friend in searching for "Naung Laing" (try #5) #Myanmar
• Google Maps
  Myanmar mystery building is here in Google Maps:
• ArmsControlWonk: Big Odd Myanmar Box
  Myanmar's mystery building found amid nuke link speculation to North Korea

• Ancient Roman City Rises Again — Curry 2009 (730): 1 — ScienceNOW
  ScienceNow: Aerial imagery shows detailed outline of Altinum, a Roman city near Venice

Altinum was a prosperous Roman city on the coastal lowlands of northern Italy. It was sacked by Attila the Hun in the 5th century AD, and conquered by the Lombards a century later. Historians now speculate that these events prompted inhabitants from Altinum and other towns in the vicinity to resettle on more easily defended islands in a nearby lagoon… thus founding Venice. (Wikipedia has a good concise history of the town.)

Now the journal Science has published the results of an aerial survey of Altinum by Andrea Ninfo, Alessandro Fontana, Paolo Mozzi, and Francesco Ferrarese at the University of Padua. (You have to pay $15 for the full-text version, but ScienceNOW has an informative (free) article based on the results, while a free PDF supplement contains methodology and original source imagery.) Their research provides us with the first proper map of Altinum.

Why is this out of the ordinary? A few reasons: Altinum was abandoned wholesale, making it is one of the few Roman settlements in Europe not subsequently subsumed by medieval towns. And while Altinum’s building stones were eventually carted off as construction material elsewhere, the foundations remained, and these became especially visible during a drought in 2007. This allowed for an unusually comprehensive reconstruction of the town. From ScienceNOW:

The researchers discovered that the crops planted on the land were in different stages of ripening, thanks to differences in the amount of water in the soil. Lighter crops traced the outlines of buildings — including a basilica, an amphitheater, a forum, and what may have been temples — buried at least 40 centimeters below the surface. To the south of the city center runs a wide strip of riper crops. They were growing above what clearly used to be a canal, an indication that Venice’s Roman forebears were already incorporating waterways into their urban fabric.

The BBC News carries a virtual flyover of the near-infrared aerial image that the researchers took. Italian news agency ANSA has quotes from one of the team members describing what they found:

”We see a walled city, a theatre, an amphitheatre outside the walls, the basilica, the forum with its market, then a principle road connected to the Via Annia (the Roman road through northern Italy),” said [team leader Paolo] Mozzi.

“You can also see a canal that divides the city in two and heads towards the lagoon. Considering the sea level in Roman times, that canal must have been connected to the lagoon as well as with nearby rivers,” he said, adding that the map also suggests a protected port that would have been used by merchant ships.

These results are just begging to be put inside the wider context of a geobrowser, so I overlaid both the near-infrared aerial image taken by the researchers and the resultant maps they produced in Google Earth. Here they are, as a downloadable KML file. Be sure to play with the opacity levels of the folder, as well as of the individual layers. It’s amazing how well the amphitheatre and basilica stand out. All source material comes from from the PDF supplement on sciencemag.org.

(My usual tangential rant at this point: When-oh-when will major science- and news publications start publishing geospatial results in geobrowser-friendly formats? In the meantime, we’ll do it ourselves:-)

But viewing these overlays in Google Earth is not the only way in which you can enhance the context of this content. This particular area of northern Italy is blessed with an unusually high number of historical satellite images taken over the years and made available in Google Earth via the historical imagery slider (quite possibly because researchers were requesting imagery of Altinum from DigitalGlobe.) Several images from 2006 show many of the features with remarkable precision, and it is also interesting to see how the changing seasons highlight different features. Here’s a quick overview of some of the imagery, as well as the one used by the researchers:

2004/09/11

2006/05/10

2006/07/01

2006/09/13

near-infrared image

Notice how some of the images clearly show the semicircular theatre and the odeon structures. (Do compare the base imagery to the overlays.)

There is also “extra credit” material: The Via Annia, built in 131 BC, skirted the northern shore of the Adriatic sea, connecting Roman settlements, including Altinum (map). In the 1st century AD, the first road to allow wheeled traffic across the Alps was built — the Via Claudia Augusta — and one spur of it began in Altinum. Remarkably, both roads are clearly visible on Google Earth for some distance outside Altinum. The Via Annia extends from the southwest of Altinum to the northeast, and is visible as markings in fields for about a kilometer in either direction. The Via Claudia Augusta, on the other hand, shows itself as a long straight line extending northwest, cutting through an otherwise chaotic set of field plots. Clearly, the road came first; it can be followed, without too much imagination, for some 10km out of Altinum, at least until the town of Olmi.