# reprojection errors with google maps?

Ever since google maps came out, the standard for web maps has been in their “flawed” web Mercator projection. I am trying to figure out why this projection is not good from a geospatial analysis perspective. Here is what I am trying to understand:

Say I create a buffer around a point that is 1 mile wide. This calculation was done in ArcGIS using WGS84 as my coordinate system. So, I will have a lat/long for all of the points around my single point within a mile. If I then want to show this on a google map, it must be projected into web mercator, which uses meters as a coordinate system. If I trace this reprojected 1 mile buffer, how closely will it resemble my buffer that used a projection using lat/long instead of meters? Will it run over all of the same houses/trees/etc if I am looking at the imagery basemap?

Or, if I have “google coordinate system/projection” on Arcmap and perform my buffer process on this, will it be less accurate than if I had done it on WGS84? Does reprojecting between the two result in a noticeable difference? Can analysis be done in a meter coordinate system?

Sorry if my questions are a bit vague. I just want to figure out what types of problems I may incur performing analysis and having the results on a google map. Thanks a lot!

Replay

Sorry if these thoughts are a bit vague(!) - but maybe they will help...

One obvious point is; what is "1 mile wide" in Lat/Long?

Think about it. Some sort of reprojection/calculation must have happened already to turn a 1 mile buffer into lat/long! So your buffer has already gone through one transformation process. Does that affect the accuracy of the data?

Then you convert it to the Google coordinate system. Again I'm stating the obvious, but reprojecting data isn't going to make it more accurate. The only way is down.

If you care that the results look the same as the analysis, then the only sure way is to carry out the analysis in the same coordinate system that you are displaying the results in.

So, as far as I can see, the best solution is to start with the data in the Google coordinate system in ArcMap. Create the buffer and do the analysis there. Then there is no need to reproject the output and it will look the same.

The only question is will the analysis be as good? I'm thinking the results could be different; they won't necessarily be wrong, just different! Does that make sense?

Hope this is some use.

Category: projection conversions Time: 2012-05-29 Views: 0