Data Problems > Censorship

The Compendium Table of Contents contains a single network link which enables you to download the latest version of the entire Compendium in a single file of about 1.1 megabytes.
____________________

A recurring theme of BBS discussion is possible censorship by Google. It has been suggested that this is not Google's responsibility, because the data is provided by other sources, and that users should take it up with such sources. We look to Google for the quality of the output, no matter how or where they obtain the data. As noted elsewhere, one does not receive bad food from a restaurant, and then permit the owner to tell you to contact the farmer if you are dissatisfied.

This is a conservative listing, because censorship can be a serious assertion. The censorship must be clear and obvious when compared with adjacent imagery. Generally, they are in three categories: blurring, pixelization, and "whiteout."

Obvious blurring or smudging: Canada and USA


Pixelization: Netherlands, New York State, Washington DC



[LIST][*]Canada
---- Ontario Hydro substation, censored by NY State GIS. Noted by C. Hays
---- Wager Bay area censored by blurring; NOT censored in Microsoft''s Virtual Earth, noted 28 July 2005, [color:"red"]Reasons unknown[/color]

[*]France
---- Dijon airfield, noted by "Jumble," 3 June 2007
---- Metz airfield, noted 3 June 2007
---- Reims airfield, noted by "sw," 22 March 2007
---- Toulouse munitions plant, La Soci�t� Nationale des Poudres et Explosifs, 4 June 2007
[*]Germany [color:"red"]Reasons unknown[/color]
---- Geilenkirken NATO HQ - the reason for this one is apparent. Caused by the new Netherlands imagery.
---- Heidelberg's "white-out" determined to be uncensored; Kleinglattbach, Monzelfeld and Oberthulba moved to "Missing Images."

[*]Israel
Jon Kyl (Republican of Arizona and co-chairman of the US-Israel Joint Parliamentary Committee on National Security) and Jeff Bingaman (Democrat of New Mexico) amended the 1996 Defense Appropriations Act:

In 1998, the US Commerce Department determined the routinely available resolution to be 2 meters. In June 2006, Google Earth improved most of Israel from the general 15 meter backgournd resolution.

[*]Netherlands (introduced with the new high-resolution images in March-April 2006 and greatly expanded in September).
---- Royal & official residences - we contrast 6 censored with 8 NOT censored
---- Military 2, NATO 11, Air Force 13, Army 25, Navy 8, Miscellaneous 4
---- Unknown 11 [color:"red"]Help is needed to identify these.[/color]

[*]Portugal: Lajes AFB, Azores

[*]Russia:
---- Chukotskiy censored by blurring, noted 23 July 2005 and 18 February 2006. Instead of blurring, Microsoft's Virtual Earth copied a nearby image and placed it on the area, which we show. [color:"red"]Reasons unknown[/color]

[*]Slovakia: Some military airports

[*]Spain: Several mainland coastal radar sites

[*]USA:
---- Washington, DC: The Naval Observatory and the US Vice President's house are uncensored since the January 2009 Inauguration. Microsoft's Virtual Earth has also removed their censorship from this location. However, Microsoft still blurs the 'Bird's eye" views of the Capitol buildings, the White House, and the Naval Observatory.
---- New York GIS censorship: Most of the high resolution imagery is supplied by the NY State GIS agency. The censorship appears to be selected by the counties, because they vary greatly in criteria. Westchester County continues to be the most highly censored, followed closely by Long Island. We have significantly expanded our analysis upstate and in Long Island after the September 2006 image upgrades. Many blurred Hudson Valley locations were noted 5 May 2006. Note the 7-by-7 mile size of the West Point Academy forest blurring in Orange County. Orange and Suffolk County have not blurred some locations - they simply "paint" them in white.

There apears to be significant problems with the terrain model in the South Island New Zealand. As an example: Sutherland falls
http://bbs.keyhole.com/ubb/showthreaded.php/Cat/0/Number/327642/page//vc/1
is approx vertical in real life, however it appears almost horizontal when viewed at an angle in Google earth. Adjacent Lakes etc are at odd angles, and rivers are flowing along ridges.

Mt Aspiring, a very impressive mountain has vanished, however the topography in the general area is correct.

Mt Cook, also a most impressive mountain, looks like a rounded hill in GE, and adjacent rivers are flowing along the side of mountains, rather than in the valleys.

I have checked the images against a series of maps, and they are very accurate - it is definitely a terrain problem. What is the underlying topo base?? SRTM??

It is so because the satellites measure the terrain with some time intervals. GE does not claim the terrain be accurate. It is only to give an idea. It was also one of the things I had enjoyed in the GE and, as you implied, it is in general correct. Nothing more is claimed upto today and I dont think there will be any better improvement, unfortunately, since the primary aim is the completion of the hi-res together with some layer info such as roads, etc.

Can you provide placemark pointers to the problem areas?

New to GE, but I must say I am very disappointed with the clear censorship of images on Long Island, NY. There is a patchwork of severely reduced resolution images in numerous downtown areas, trainstations, powerplants, etc, and of course, the reduced quality of the area is far larger than the area of the "protected" site.Certainly some may consider some of these areas vulnerable entities, yet these image areas are readily available on terraserver. So, not only did I subscribe to GE, but I was forced to subscribe to terraserver to fill in all the images I cant get with GE.

Just do a search for Brentwood, NY and set an eye level of about 35 miles to see just a sampling of the numerous censored areas in green.

I'm from LI and I unfortunately am all too familiar with this.

There's a bright side though. First of all, don't blame Google. It wasn't their choice to withhold those patches of imagery. In fact, in the previous database (still accessible for some subscribers as the "legacy database"), you can clearly see that all the "protected" patches are filled in with what looks like DigitalGlobe imagery. So all of LI is high-res.

When they switched over to the current database, for some reason, all of those fill-ins were lost. HOWEVER, if you check out central Suffolk, you can see that in one concentrated area (to the west/south of Riverhead) most of the gaps have been filled in. I noticed the size of this restored area grow gradually during the last several updates. I think it's safe to say they are on a mission to restore all of LI.

Why are there gaps in the base NYGIS imagery in the first place? I'm not sure, but you'd have to take that up with NYGIS, not Google.

I figured NYGIS was behind the reduced resolution for security reasons. Many of the reduced resolution areas contain small gas storage facilites, tank farms, train stations, State and County buildings, etc., yet I find it silly to restrict access when these images readily available elsewhere, especially the downtown areas that are restricted. Being that most of these "restricted" areas are readily accessible to the general public anyway. Hopefully the trend to replace the missing data will continue, and all the high res images will be restored. I'll cross my fingers.
Regards,
Crash

A brief tour of the French Alps revealed also several errors and omissions. The choice of named places is also arbitrary. Oddly, there seems to be more than one set of names because one mountain alternates between its correct name, Trainon, and the wrong name, Trenom. It depends when I look. Mont Blanc also alternates with its Italian name.

I suspect listing all the errors here for the world will cause a massive list. Google Earth is a truly wonderful product but I suspect it needs to rethink what places it names and improve the quality.

Hi
GE is new and the names may well be wrong. Members report the errors. However, I learned that Google is performing a work to acquire the local names thru local collaborative companies (just as they do for the roads layer).

Hello,

Clearly, you've looked into this in more detail than me!

I can imagine that image data might be misaligned when it's new but it it a bit odd that it comes in OK and then moves!

(I noticed that some of the errors in Luxembourg seem to be OK now.)

Cheers,

Simon.