This is the fourth and final in a series of posts about the size of the data in Google Earth. We already looked at:
- How much data 3D imagery requires
- How much data different types of 2D imagery requires
- How much historical imagery exists
Today we are putting it all together and trying to come up with an estimate for just how large the Google Earth database really is.
3D Imagery: 1024 TB
The total area of 3D imagery currently in Google Earth is approximately 524,000 sq km. We re-ran our tests for 3D imagery a number of times trying closer and closer views and every time we got closer, the figure got bigger. We eventually settled on an estimate of 2 GB to 1 sq km of 3D imagery although we believe it is an underestimate. This gives us a total of 1024 TB for 3D imagery.
Default Layer Aerial Imagery: 179 TB
For areas of countries we mostly relied upon this list on Wikipedia.
The continental US has more or less complete aerial imagery coverage. Total area approx: 7,663,000 sq km.
Based on our maps from this post, we estimate that the European countries Portugal, Spain, Italy, France, United Kingdom, Germany, Netherlands, Czech Republic, Austria, Montenegro, Greece, Denmark, Switzerland and Belgium have about 50% aerial imagery, as does Japan. Total area covered 1,506,000 sq km.
The US has approximately five times as much aerial imagery as the rest of the world combined.
Based on our previous results, we estimate aerial imagery requires approximately 2 GB per 100 sq km. This gives us a total of 179 TB for aerial imagery.
Satellite Imagery: 196 TB
Excluding Antarctica, the rest of the world’s land mass is approximately half covered with good quality satellite imagery. This is a total area of 63,638,000 sq km. Based on previous results, good quality satellite imagery requires about 0.3 GB per 100 sq km. Low quality satellite imagery requires significantly less space at about 18 MB per 100 sq km.
This gives us totals of 186 TB for good quality satellite imagery and only 10 TB for low quality satellite imagery.
Historical Imagery: 1,618 TB
This is the hardest to estimate. We will exclude the default layer in our calculations. The continental US has on average about seven aerial images for any location. This gives us a total of about 898 TB. Europe and Japan have approximately 50% coverage with five layers of aerial imagery. This comes to 118 TB. About a third of the world’s land area has five layers of good quality satellite imagery. This comes to 527 TB. About one percent of the world’s land area has twenty or more good quality satellite images. This comes to 75 TB.
So our final estimate for the total size of the Google Earth database is 3,017 TB or approximately 3 Petabytes!
Compare that to this post from 2006 at which time the estimate stood at 150 TB.
How accurate is our estimate? Given that the bulk of the size comes from historical imagery for which we simply do not have very accurate data, our estimate could easily be a long way from the true figure. In addition, the method used for determining how much data each imagery type requires was not particularly accurate. We have also completely ignored the old type of 3D buildings, and all the street and mapping data or layers. We have considered Street View to not be part of Google Earth as it should really be considered more of a Google Maps product or a product in its own right. We believe that overall it is a significant underestimate and the database is actually quite a bit bigger.
A significant proportion of the Google Earth database is US aerial imagery, both current and historical. For this we can largely thank the USGS and the USDA Farm Service Agency, although much of the most recent imagery is gathered by Google itself.
About Timothy Whitehead
Timothy has been using Google Earth since 2004 when it was still called Keyhole before it was renamed Google Earth in 2005 and has been a huge fan ever since. He is a programmer working for Red Wing Aerobatx and lives in Cape Town, South Africa.
Please fix the units in the pie chart.
Timothy Whitehead says
Thank you for letting me know. I have fixed it. You may need to refresh the page to see the new image.
It would be nice of Google if they could just tell if you are close or not. It shouldnt be that much of a secret for them – just one number.
Nice series of articles. I think you are close, but with some overhead it is probably 20% of so understated.
What about compression? Is this compressed or uncompressed data?
Timothy Whitehead says
The sizes are based on the space taken up in the Google Earth cache. Image data of this nature typically does not easily compress losslessly so it is likely it is largely uncompressed in both the cache and Google’s database. We also have not taken into consideration either the index or other database related issues for both the cache and Google’s database.
Chris Cardinal says
With that much historical imagery, it’d be great if Earth (at least on GMaps) offered the same time machine functionality street view offers.
Nice numbers, how’s backup? RAID?
Timothy Whitehead says
It will be in Google’s server farms which almost certainly use raid (it’s standard on most servers) as well as distributing the data all around the world. In 2006 it was stored using a Bigtable database and probably still is.