I have had several questions lately regarding importing multiple Google Earth images into the same drawing and the resulting accuracy of said images. This seems to be a concern for a few users who are working with Google Earth, so I created a test to show exactly how Civil 3D works with Google Earth imagery.

To explain how Civil 3D places images from Google Earth, we need to understand the concept of georeferencing and rubbersheeting. An image is a flat object, and the earth is anything but flat - it is more spherical than anything. In order for locations to be correct on all parts of a image, the image needs to be "rubbersheeted" - that is, points must be correlated on multiple points on the image and the earth and the image is warped to match as a result.

The inherent issue that people are seeing is due to the fact that when an image is imported into Civil 3D from Google Earth, only the center point of that image is transformed to the desired coordinate system (georeferenced.) From there, the longitudinal width and latitudinal height of the image are transformed using an approximation. There is no rubbersheeting of the image to map it from a geographic coordinate system to the planar coordinate system.

Another issue is that the image in Google Earth is not orthorectified to any particular mapping plane, so distances measured on the image are not the true planar distances.

That's the short answer - this is behaving as designed. However, to see it in action, we need to look at an example. In this example, I have located a highway interchange in Google Earth. I chose this because it represents a large area that has sufficient detail to provide some graphical indications of what may be going on. When I import this image into Civil 3D at an eye height of 3189 meters, you get a very broad overview of the area, as shown in Exhibit 1 below:

If I hold the same center point and zoom in to the area of the flyover to get greater detail and import that image, I can see fairly consistent results. The following image (Exhibit 2) was inserted at an eye height of 329 meters:

As you can see, the two images line up fairly well. It's when we move off of that center point that we can see the real problem (Exhibit 3):

Moving to the north side of the interchange, we can see the problem in even more locations (Exhibit 4):

This issue repeats itself on the other two sides of the interchange. The largest error that I was able to find via rough spot checking was an 83 meter discrepancy.

Google Earth was meant to be used for conceptual purposes only - it was never intended to be used as a replacement for a ground-run survey or orthorectified and georeferenced aerial imagery. Use it, but use it wisely.

Noor Azhar  bin Mohd Nor

Consultant-AEC