I want to publish a mapservice in ArcGIS server, which will be overlayed on Google Maps or Bing Maps. For this, the Map Service needs to be in EPSG:3857/900913.

The Google Maps projection is supposed to assume a Spherical Earth Projection. So I went to

Coordinate Systems\Projected Coordinate Systems\World (Sphere-based)

There seems to be a Mercator (sphere).prj, but it does not seem to have the proper Projection parameters.

I then went to

Coordinate Systems\Projected Coordinate Systems\World

& there are 3 WGS84 based mercator projections:

Coordinate Sytem

I see that in a lot of ESRI literature, they use WGS 1984 Web Mercator (Auxiliary Sphere). Is that the Google Maps Projection? What is the WGS 1984 Web Mercator.prj then?

(As a bonus question, this image is from ArcGIS desktop 10.0, but does the answer again change at ArcGIS 10.1?)

  • Good question..I never think of this... I am using WGS 1984 Web Mercator (Auxiliary Sphere) projection and its working fine for Google map + if you want to use your layer as base map then you must use above projection.. ( Using 10.1 ArcGIS desktop +Server)
    – Sunil
    Feb 8, 2013 at 5:37

2 Answers 2


here is the epsg wiki on the subject.
With other links to esri and openstreetmap discussions.
My understanding is that in the epsg realm the 900913 is deprecated.
(900913 spells google in numeric so it was too cute to be official).
It is now officially 3857 in the epsg database.

and from esri:

ArcGIS Online will be transitioning to WGS 1984 Web Mercator (Auxillary Sphere)
What does this mean for you?

sterlingdq says:

@nnne – For a long time EPSG refused to assign a code to this coordinate system; therefore ESRI created the WKID codes 102113 and 102100.

When EPSG did assign a code, they used 3785, but later changed it to 3857. ArcGIS 10 will follow ESRI practice of using an EPSG code when one exists, and will advertise the coordinate system of the service as 3857. ArcGIS 10 and all Web APIs are being designed to recognize EPSG 3857, ESRI WKID 102113, and ESRI WKID 102100 as equivalent.

Sorry I was stil working on the answer.
WGS 1984 Web Mercator (Auxiliary Sphere) is what you want to select in arcmap but read the what does this mean for you if you are creating tiled services.
I am on 10.1 and this is the definition in it...
web mercator

and finally the WGS 1984 Web Mercator is just the old esri def pointing to epsg 3785


  • 1
    Ok, I know what the EPSG code for Web Mercator is. I am asking which projection should I select in ArcMap? Feb 8, 2013 at 16:09
  • After a Bit of testing I had found out that WGS 1984 Web Mercator (Auxiliary Sphere) appears as wkid 3857 in the rest end point, but the Blog article you have linked as well as ESRI KB article 37329 say that you should WGS 1984 Web Mercator. Hence the confusion. Feb 8, 2013 at 16:22
  • 1
    So It looks like either one could be used, while the Axillary Sphere looks more correct. As an aside It's Interesting that you can see the EPSG code in 10.1 I just have ArcGIS 10, and the text of these two projection files look very different Feb 8, 2013 at 16:56
  • 2
    WGS 1984 Web Mercator uses our standard Mercator algorithm which supports spheres and ellipsoids. To trigger the sphere code, the geographic CRS must use a sphere. This complicates workflows when your data isn't WGS84 (a transformation is needed). WGS 1984 Web Mercator (auxiliary sphere) uses an Esri-specific Mercator algorithm that is sphere-based but you can specify how to treat an ellipsoidal geo CRS, thus it can uses the standard WGS84 definition.
    – mkennedy
    Feb 8, 2013 at 18:32

The EPSG codes are not considered real projections according to the USGS. WGS84 (and NAD83) use the same Ellipsoid so they are essentially the same. All of the GIS systems I have seen in my state (Louisiana) are State Plane coordinate systems. I am a licensed Land Surveyor and worked with the State Plane systems exclusively. However my required training encompasses all of the projections. Surveyors and engineers change any Angular coordinates to some kind of Cartesian system, angular projections are useless to us because we work with real numbers calculated by the USGS which also include up to date satellite corrections of the shape of the earth. Of course we are working in sub centimeter environments.

  • 1
    I Don't see how this answers the question. Nov 21, 2017 at 5:45
  • All of these projections must be based on a specified longitudinal central meridian and the proper elipsoid (modern shape of the earth.. Simply saying stay with projections calculated by USGS keeps data from warping by switching to all of these different projections.
    – user35019
    Nov 21, 2017 at 13:23
  • USGS states that lat-long are not really coordinate systems in the strict sense (ref) but I am not aware of them claiming that all EPSG coordinate systems aren't 'real'. Such a claim would be bizarre as EPSG has codes covering the State Plane systems the OP references (and most other 'straight sided' coordinate systems besides). Further, to say that a spheroidal coordinate system is not strictly a coordinate system because it doesn't have straight sides is wrong. A coordinate system does not have to confined to cartesian space. Dec 4, 2018 at 18:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.