How much data is in Google Earth’s 3D imagery?

Google continues to add new 3D imagery on a regular basis. Most of the major population centres in the US and Europe will soon be covered and many other parts of the world have significant areas covered too. So, just how much data does it constitute and how does it compare to plain overhead imagery? And when you browse an area in Google Earth, how much data must your computer download?

To answer the above questions we used the Google Earth cache to monitor how much data is being downloaded. We first deleted the entire cache, which can be done by first signing out of the Google Earth servers (File->Server Sign Out) and then going to Tools->Options->Cache and clicking the ‘Delete cache file’ button. Next, we created a zigzag path that covered a square region of 3D imagery and ran it as a tour as described in this post. We initially set the camera height for the tour to 1 km. At this height the camera can see the next row of the zigzag so that as it proceeds, all imagery in the area covered is fully loaded to the resolution Google Earth chooses for that view height. We did this for several different patches of 3D imagery. We also discovered after running most of our tests that if we get closer, then even higher resolution imagery is loaded, so we did another test for a smaller area from a height of just 300 m.

Our main tests covered an area of approximately 100 square kilometres each. As you can see below, this is actually only a tiny amount in relation to some of the larger areas of 3D imagery, but we could not do much larger areas as we were concerned about exceeding the size limitations of the Google Earth cache. As it was, we discovered that although the maximum size you can set for the cache in Google Earth’s options is 2 GB we were, in fact, able to grow it much bigger. At one point it even exceeded 4.6 GB and then later shrunk to just over 2 GB.

The area we tested near Milan, Italy (white zigzag). The red outlines show areas with 3D imagery.

The zigzag shape we used. (This one is Tokyo, Japan).

A corner of the area covered showing that only the area around the zigzag was fully loaded.

We also ran each test with 3D buildings turned on, and 3D buildings turned off.

Here are the results for 100 square kilometres with a camera height of 1 km.

	3D buildings enabled	3D buildings disabled
Milan, Italy	3,097 MB	294 MB
Atlanta, Georgia	1,468 MB	296 MB
Abilene, Texas	803 MB	250 MB
Tokyo, Japan	1,317 MB	273 MB

The surprise is the wide variation from location to location. The area near Milan, Italy, uses four times as much data as the area near Abilene, Texas. The figures for plain imagery are much more consistent. All the areas tested have aerial imagery. In a later post we will have a look at whether or not there is a difference between aerial and satellite imagery.

The second surprise was just how much data 3D buildings require. We ran a test from a height of just 300 m over an area of just 20 square kilometres and it filled up 2,230 MB of cache, so the figures above are actually much smaller than if the best resolution imagery possible is loaded. The single area around Milan, Italy that has 3D imagery is approximately 2,500 square kilometres so it requires approximately 278 GB of data. That’s about the same as a small hard disk drive.

To see the zigzags we used in our tests, download this KML file, and to see what parts of the world have 3D imagery download our KML map.

This is the first in a series of posts trying to work out just how much data is in various aspects of Google Earth. If all goes according to plan, we should, at the end, be able to make wild guess at just how big the Google Earth database really is.