For those interested, I am looking at a topo that was done in the field and viewed on Google through TBC. The northerly point of this topo is a centerline road shot and it checks well with Google which has a really nice 2013 photo in it. What surprised me was that the elevation was exactly the same as the topo shot (the Google display is 4690 and the topo point is 4690.
So the next point I checked is also "dead" on with Google, I got all excited and began checking other points. They varied from 2' low to 6' low:-(
Oh well, thought I had something for a bit. But it is better than it used to be, still not there for any kind of topo I do.
I've seen the same thing. Sometimes spots are right on and other times they're way out of whack. For what it's worth it's probably a little more accurate than a quad sheet.
We used the surface from Google imported into Civil 3D to do some conceptual layout for a solar site before. It wasn't good enough for a final design, but it gave us a start in the right direction. We compared it to our field locations on that one as well, and we had similar results.
I think it would be more fun to just start using them like they are good, then life could get more fun!!;-)
GE gets it's elevation data from the shuttle radar topography, with a post spacing of 30m. Generally speaking large flat areas will be accurate.. but it will truly be hit or miss.
http://en.wikipedia.org/wiki/Shuttle_Radar_Topography_Mission
I will point out that a few rare sites (such as the area around detroit metro airport) are clearly using something MUCH denser, such as LIDAR. You can see trees and buildings in that elevation data set and it is, frankly, amazing.