[Proj] Re: Problem overlaying georeferenced images in Google, Maps(projection problem!)

[Melita Kennedy]

-----Original Message-----

>Date: Mon, 02 Jun 2008 10:23:39 +0200
>From: Jorge <jorge.arevalo at gmail.com>
>Subject: [Proj] Problem overlaying georeferenced images in Google,
>	Maps(projection problem!)
>To: proj at lists.maptools.org, support.mn at elisanet.fi, cjmce at lsu.edu,
>	jagoncal at gmail.com
>
>Hi everybody
>
>To Janne and Cliff:
>
>> Hmmm....
>>
>> are you sure the "gdalwarp" you are using uses the latest proj-4 libraries?
>> Are you sure all programs are doing what thye should? If so, it is maybe the
>> input data (assumptions) then?
>The gdal tools and the proj.4 libs are the last version, yes. And I'm using them in the way the instructions say, I think. But is
>obvious that I have a mistake. Maybe there is an error converting the MIF/MID files in a raster file, but the process is so
>simple... I will review it, anyway.
>
>
>> More ideas...
>> 
>> The huge north shift, which is visible in your pictures,  is usually due
>> to the conversion between spherical and elliptical. There is a way to
>> go around that, but how it should be done in your case.... ???
>
>> I believe the easiest solution to that would be to correct that north shift by utilizing Rectifying Latitude for the 
>> International ellipsoid at the center of the area.  Don't know if that's in the PROJ4 Toolkit, though ...
>
>Mmm... thanks for the clue! Then, I should do this for every province's files (52, exactly), but maybe the shift isn't the same 
>in all... I'm going to see the rest
>
>To Jose:
>http://lists.maptools.org/pipermail/proj/attachments/20080602/4a49c655/before_and_after-0001.jpg
>
>The "before" shift isn't the same as mine, but this is VERY interesting... What projection is "before" and what is "after", What 
>are the input data? How did you do??
>
>Thanks a lot!

The Mercator-based coordinate system used by Google Maps can be thought of in 
two ways. 

1) It uses a sphere with radius=6378137.0 m. Sent to Mercator, hopefully, the 
code is smart enough to use the spherical equations.
2) It uses WGS84 with spherical Mercator equations. 

No ellipsoid-to-sphere conversion should take place. Data that's really on WGS84 
is assumed to be on the major auxiliary sphere (R=6378137.0m) and just projected 
using the spherical math. This means that you can't create a custom 3 or 7 
parameter datum transformation to convert directly to the sphere (because the 
latitude values are converted to the sphere).

So you might try transforming your data from ED50 (or other datum) to WSG84 
first. Then project those coords to the Google Mercator.

Melita
ESRI, Inc.