Security through obscurity? Not post-Google Earth

The most recent imagery update is somewhat special in that many of the updated tiles are from DigitalGlobe imagery taken earlier this year. That’s a faster turn-around rate than we’ve seen, previously.

Do this: In the Layers pane, open the DigitalGlobe Coverage directory. Open the “DG Coverage – 2007” directory. Now turn on the layer for cloud cover 0-10%. This is the category that a lot of the new imagery belongs to. By surfing with this layer turned on, you can see where tiles correspond to the overlay. That’s where most of the newest imagery is.

If you do this, you can see how the Federation of American Scientists found a second Jin-Class submarine docked in a Chinese port, blogged on their Strategic Security Blog. That’s on an image taken on May 3, 2007.

Focusing on the middle east, notice that all of the Gaza Strip is now shown on imagery taken in June. As the tiles also contain Israeli imagery, they are artificially reduced in resolution to 2 meters per pixel, by US law for US satellite imaging companies. You can see pixellation artifacts in the resulting imagery.

Not that this is stopping some of the more alarmist voices in Israel from complaining, despite Israel being the only country in the world that can boast such blanket censorship. Editorial voices in the Israeli tabloid Yedioth Ahronot are apparently not happy, as quoted by Charles Levinson in Conflict Blotter:

[Yedioth Ahronot:] Sensitive installations, Air Force bases with their planes and helicopters, missile bases and even the nuclear reactor in Dimona have never been photographed better. A recent Google Earth update shows satellite pictures that make it possible to see clear, sharp pictures of military and civilian targets all across Israel.

Up until recently, the satellite pictures of Israel on Google Earth had a particularly low resolution: every pixel was equal to 10-20 meters. Now, the satellite maps of Israel show great parts of the country with a resolution close to two meters per pixel.

That’s not quite true. Large swathes of Israel have been visible at 2 meters per pixel since the summer of 2006. All that’s happened is that in some instances coverage has been updated, while in others it has been improved.

Meanwhile, some elements of the media in India are keen to find any support for their own jingoistic defense of censorship. India Daily has an article that explicitly makes the Israel-India analogy:

Google Earth uses civilian satellite imagery to portray the details on the earth. It is now easy to find the key Israeli and Indian nuclear reactors. It is also easy to find key military and air force bases.

The biggest question hover in the minds of Israelis and Indians — ‘Given the fact both the countries are surrounded by Islamic Jihadists and terrorists, should Google Earth not reveal secrets of these countries?’

Erm, yes it should? The Dimona imagery has been available on DigitalGlobe’s online store long before it hit Google Earth. Considering that Dimona and other places have been in public domain, and that those most determined to see the imagery have access to sources that are not Google Earth or even DigitalGlobe, I think it is a very good idea to democratize access to such information as much as we can — and Google is the leading light in this de facto campaign.

And to state the obvious, any army today that still believes in security through obscurity is in fact neither.

4 thoughts on “Security through obscurity? Not post-Google Earth”

  1. I’m not sure that using the Digital Globe coverage layer give correct results. I checked two areas that say they have 2007 imagery, but it isn’t; it’s at least 2 years old. One place in California, one in Iraq. This has been the case in the past when I tried that layer. Now I don’t refer to it at all. If there is a way of double-checking, I’d like to know about it.

  2. To add to my prev. comment, there are lots of 2007 rectangles that have no imagery in them, just the base. Also, if you’ve checked out Ayers Rock in Australia before, it now shows many 2007 rectangles (makes it easy to find), but the imagery has not changed from the last time I looked, say a month or so ago. So, what do those rectangles tell us? is it that there is captured imagery, but it may not have gotten into GE yet?

  3. Ernie,

    The DG imagery layer only represents the swath tiles available through DG from their library. I think Stefan is only attempting to point-out that one can determine a reasonable assumption in discerning what images are present based on the maximum 10% cloud coverage requirement for Google Earth, and through that DG layer.

    0%-10% cloud coverage tolerance is the typical standard for vis applications like Google Earth. In other simulation environments, we actually require a final output of 0% cloud cover — usually acheived by incorporating more than one image of the same area, cutting out the clouds, with additional techniques to ‘blend and match’ the imagery — to make it appear that it’s one image. Otherwise, tasking and costs would make obtaining a cloudless image non-feasible.

  4. Daniel answers a question I have had about the GE imagery. While the DG imagery layer overlay may show the date of satellite imagery coverage – from what is stated this DOES NOT necessarily indicate GE has placed that imagery frame itself in the data base. This could be because – insufficient cloud free coverage or GE has not purchased that particular frame (yet?). The image blending process described by Daniel could even more greatly blur the actual date of acquisition. So an assured determination of the imagery may only be available based on a known “ground truth” at x date.

Comments are closed.