Topic: Question about Google Earth

[FengK]

I imported an image from Google Earth with the scale legend as shown below, then I measured the length of the scale bar which is supposed to be 3958 feet, but i got 4053 feet. it is about 2.4% off. This seems to be consistent from a number of tries i did. Is this normal? Thanks.

[Matt__W]

I've questioned this before myself.  I've wondered if it's a straight up (estimated) horizontal distance or does it take into consideration the curvature of the Earth?   

[mjfarrell]

when and or if you insert this image into a drawing with survey data, how does the image fit the known points?

[FengK]

when and or if you insert this image into a drawing with survey data, how does the image fit the known points?

they don't fit exactly either.

[mjfarrell]

when and or if you insert this image into a drawing with survey data, how does the image fit the known points?

they don't fit exactly either.

In this case you will want to remedy by using Map>>Tools>>Rubber Sheeting
This tool will warp and distort the image to fit the known data.
It is always best to try to use a number of internal, and external points of correlation to assist the software in warping the image to fit as well as possible.

[CADaver]

Parallax

[Mark]

Parallax

http://en.wikipedia.org/wiki/Parallax

[Strucmad]

Interpretative photogrammetry

[FengK]

Parallax

http://en.wikipedia.org/wiki/Parallax

Thanks Randy and Mark. The reason why the distance from GE image scale legend is smaller is because it is measured using parallax? And this distance is usually smaller than the actual distance on the ground?

[mjfarrell]

Kelie,

I think the paralax issue is a probable cause, however it's most likely a false lead.
I would suggest a letter to the good folks at google, explain your process of how you imported what into C3D, version, etc.  And that a resultant plot wherein all known objects are the correct distances; that any attempts to correlate the graphical scale on the image yield results that cause you to doubt the accuracy of the data.
Perhaps they can also provide a definition, or formulae that explains how said scale is derived.

[CADaver]

The posted scale will only be accurate in one peculiar spot on the photo, ALL other will be out of scale.  The cause is parallax.

A single point lens at elevation takes a "flat" picture of the ground below.  Directly under the lens is a specific scale.  However as you move laterally along the ground, you move farther away from from the lens, and as you do the scale changes.  Rubbersheeting will help, around the points used for rubbersheeting, but the optical distortion will remain.

[FengK]

Kelie,

I think the paralax issue is a probable cause, however it's most likely a false lead.
I would suggest a letter to the good folks at google, explain your process of how you imported what into C3D, version, etc. 

Thanks Michael. I'm wondering would it be Google or Autodesk that I should be asking about this.

[FengK]

The posted scale will only be accurate in one peculiar spot on the photo, ALL other will be out of scale.  The cause is parallax.

A single point lens at elevation takes a "flat" picture of the ground below.  Directly under the lens is a specific scale.  However as you move laterally along the ground, you move farther away from from the lens, and as you do the scale changes.  Rubbersheeting will help, around the points used for rubbersheeting, but the optical distortion will remain.

Randy, Thanks for the explanation. One more question: This inaccuracy of imported image (in Civil 3D) from Google Earth, does it also apply to the surface imported from Google Earth? I'm guessing the answer is yes? Therefore if I want to import a surface from Google Earth and overlay it with a topo (with a bunch of streets) in Civil 3D, this is how I would do it: import both the image and surface Google Earth into the drawing: scale (maybe rotate as well if necessary) both the image and surface using two common points (as far away from each other as possible) as reference, say two intersections that are shown both in the image and topo. Is this an acceptable approach? TIA for your comments.

[mjfarrell]

I would not suggest scaling the surface data....

I also would not suggest using the GE data for anything other than PRELIMINARY design considerations.

I would suggest a topographical survey be performed, and then compared to GE be prepared to be [ ]

I would send identical queries to both Googl, and autodesk in regards this question.

[Dent Cermak]

I may be wrong, but I don't think Google Earth has ever made any great claims as to the accuracy of their images. They have certainly never said the stuff meets survey or engineering accuracy standards. Just using street address in Google Earth the closest fit I can get on cross hairs vs image is around 200 feet. For real accuracy their control network for their images would be cost prohibitive. Google's stuff is a mix of lidar, satellite, standard aerial photos, etc. From what I can see, it's not Orthophoto quality. It's for reference. Rubber sheeting can help, but there will still be error. You can probably get it as tight as you used to get with a Salztman vertical rectifying projector, but not much better than that. Cool tool for diplays and tax maps, but I would not take it any further.