I've read a whole lot about the faultiness of using Google Earth for anything remotely related to surveying, but reading a few recent threads, prompts this question:
For any given location in the US today, is there a "mean deviation" for a single spot? That is, what is the likelihood that any Lat/Long is actually where it appears to be in Google Earth within "x" feet? Does it depend on the date of the photo? The people doing the geolocating?
Second, for a distance of, say less than a mile or so, what is the relative positional accuracy of the projection GE uses? If you confirm that one point is accurate to within "X" feet, is it a safe assumption that every measurable spot within that area is similarly accurate?
I've seen the obviously errant GE photos, but am curious about the "bigger picture" here.
It's a crap shoot for me usually. I have opened kml files derived from field data and had them "plop" right down in Google Earth on top of exactly where they're at....and I've had them miss by 30 or 40 feet just as often. Doesn't seem to be any reasoning. I avoid using GE for anything other than pretty pictures, although most of it is "pretty close"...
In my experience terrain plays a large factor in the "positional accuracy" of the photography on google earth. I've surveyed quite a few cell towers and plotted the lat/longs to google earth as a quick double check and have found that on flat ground they tend to be within 10-15' and that in mountainous terrain they can be more like 50' off. I did have one once right in the middle of the Salt Lake valley (high quality photography and flat ground) that the image was 75' off.
In my area it seems Google Earth has pretty accurate data. I never really used it for much. Sometimes the aerial photos are newer than the state GIS ones, so I've inserted them into a drawing to make the drafting process easier,but the resolution isn't as good.
We also used to use the surface data for concept plans for solar farms. They wanted to see ballpark grading issues on site before carrying out any field work - it was mostly for feasibility before getting too far in. There was a BIG disclaimer on those drawings about the source of the data and our lack of responsibility for the accuracy. The reason I mention it is because on the jobs we got where we actually went forward and did topographic surveys it wasn't that far off of the Google data. I couldn't quantify exactly off the top of my head, but I want to say that it was GENERALLY within a foot or two.
Google takes all the free data and compiles it. Ever notice that the imagery you get from your state's GIS looks identical to the imagery from the same date on GE? GE is a framework that is 'perfect', the data they pop on top is imperfect. The surface data is based on LandSat which is what, 30m resolution?
How good is it? It is just like an old person's underwear: Depends.
At least this much. Note how some things appear twice. Note how the street headed north suddenly gets torn off just before it hits a house. The street headed south disappears into some trees.
Dan Patterson, post: 328165, member: 1179 wrote: In my area it seems Google Earth has pretty accurate data. I never really used it for much. Sometimes the aerial photos are newer than the state GIS ones, so I've inserted them into a drawing to make the drafting process easier,but the resolution isn't as good.
We also used to use the surface data for concept plans for solar farms. They wanted to see ballpark grading issues on site before carrying out any field work - it was mostly for feasibility before getting too far in. There was a BIG disclaimer on those drawings about the source of the data and our lack of responsibility for the accuracy. The reason I mention it is because on the jobs we got where we actually went forward and did topographic surveys it wasn't that far off of the Google data. I couldn't quantify exactly off the top of my head, but I want to say that it was GENERALLY within a foot or two.
"I couldn't quantify exactly off the top of my head, but I want to say that it was GENERALLY within a foot or two."
That's at least an order of magnitude better than I see...
I just pulled up a job I worked on recently and the centerline of road elevations vary from within about 1' to 2' of my elevations.
Do they use the same elevation data everywhere? What do they use? It shows changes every foot, but I'm sure it's interpolating from 5 or 10 or 20 ft contours.
rfc, post: 328160, member: 8882 wrote: I've read a whole lot about the faultiness of using Google Earth for anything remotely related to surveying, but reading a few recent threads, prompts this question:
For any given location in the US today, is there a "mean deviation" for a single spot? That is, what is the likelihood that any Lat/Long is actually where it appears to be in Google Earth within "x" feet? Does it depend on the date of the photo? The people doing the geolocating?
Second, for a distance of, say less than a mile or so, what is the relative positional accuracy of the projection GE uses? If you confirm that one point is accurate to within "X" feet, is it a safe assumption that every measurable spot within that area is similarly accurate?
I've seen the obviously errant GE photos, but am curious about the "bigger picture" here.
I brought a GE image into Carlson the other day and it gave a message of about 57.5 feet accuracy. Had not seen that message before, not sure where it comes from. But it seems either Carlson or GE or both are attempting to deal with the metadata issue?
If you go to this location:
N32d03'42"
W095d18'47"
you will find a transmitter tower for a TV Station that you can see the side of it. Measuring the horizontal distance from the top of the tower to the bottom of the tower will give you about 320 feet.
I assure you that tower is not leaning 320 feet but it gives you an idea of how much distortion you can get from an extreme case of elevation change as you get away from the center of the photo. I don't know how much of that they adjust out at ground level but especially in hilly areas, there will always be some error.
James
NGA gives actual examples of "Web Mercator" errors in EXCESS of 36 Kilometers. Google on: NGA.SIG.0011_1.0_WEBMERC and see for yourselves. This is why it is an awful tool for military personnel to attempt to use. For civilian applications you may have good luck in some places, if not - it's no big deal. For the military, it can result in "Friendly Fire" and then people die.
I like Google Earth; it's OK for student use, but it's poison for military use.
As pointed out above, it depends ...
Wow! I've never seen it with an error of that magnitude with anything I have used it for. That said, I have not used it for much and never relied on it for information. I have noticed that it is far less consistent than the orthorectified photos I can download from NJDEP. Sometimes google pops right into the dwg and other times its off several feet. I don't think any decent surveyor would really rely on it for much other than to see what something looks like for help coming up with a price or to assist with the drafting process.
Cliff Mugnier, post: 328219, member: 505 wrote: NGA gives actual examples of "Web Mercator" errors in EXCESS of 36 Kilometers. Google on: NGA.SIG.0011_1.0_WEBMERC and see for yourselves. This is why it is an awful tool for military personnel to attempt to use. For civilian applications you may have good luck in some places, if not - it's no big deal. For the military, it can result in "Friendly Fire" and then people die.
I like Google Earth; it's OK for student use, but it's poison for military use.
As pointed out above, it depends ...
The 36 km error occurs if you mix-up "Web Mercator" with a Mercator-based coordinate system that's using an ellipsoid rather than a sphere. It's not an inherent fault of "Web Mercator".
I do still hate extremely dislike Web Mercator.
2nd Edit: Google Earth doesn't use Web Mercator at all, only the 2D Google Maps, Bing Maps, other 2D "basemaps" do. AFAIK, GE uses a simple cylindrical projection (Plate Carree).
Melita
mkennedy, post: 328239, member: 7183 wrote: The 36 km error occurs if you mix-up "Web Mercator" with a Mercator-based coordinate system that's using an ellipsoid rather than a sphere. It's not an inherent fault of "Web Mercator".
I do still
hateextremely dislike Web Mercator.2nd Edit: Google Earth doesn't use Web Mercator at all, only the 2D Google Maps, Bing Maps, other 2D "basemaps" do. AFAIK, GE uses a simple cylindrical projection (Plate Carree).
Melita
If I'm not mistaken, you have an option to choose several reference systems, unless this is purely how the information is displayed on the screen:
Ah, yes, those are display formats. Internally, the data layers for Google Maps are stored in "Web Mercator", and in Plate Carree/simple cylindrical for Google Earth. Normally, a user doesn't need to worry about the storage coordinate system, but if you're a developer planning to overlay aka "mash-up" data against GE or Google maps, it's faster to reproject data to match the storage coordinate system.
Melita
