FME produces different coordinates compared to ArcMap

Hi,

 

 

The FME Reprojector isn't always very clever about which "path" it uses from one projection to another, which can lead to discrepancies compared to other systems such as ArcGIS. This is why I'd recommend you use the CSMapReprojector in stead.

 

 

Or, in your case, you could use the EsriReprojector transformer, which ought to give exactly the same results as ArcGIS.

 

 

David

Apologies,

 

 

I mean the CSMapReprojector seems to give the exact same results as the regular Reprojector.

Found this description.

 

"If the coordinate system of the data source and data frame are not the same, you may get a different result if you calculate geometry using the data frame's coordinate system than when you calculate using the data source's coordinate system. It is recommended that an equal-area projection be used when calculating areas."

 

-- ArcGIS Resources > Calculating area, length, and other geometric properties (http://resources.arcgis.com/en/help/main/10.1/index.html#//005s00000027000000)

 

FYI.

Hi there,

 

 

Just want to find out if there is still no concrete solution to this issue.

 

Also, I still don;t understand the unavailability of the so called "ESRIReprojector" transformer.

 

I'm oon my knees here, as nothing I try seems to work...

 

 

Regards,

From Safe Support case:

 

 

The ArcGIS definition for Africa Lambert Conformal Conic seems to be different than that in FME.

 

 

ESRI:

 

Projection: Lambert Conformal Conic

 

Origin Latitude: 0

 

Origin Longitude: 25

 

Standard Parallel 1: 20

 

Standard Parallel 2:-23

 

 

FME:

 

Projection: Lambert Conformal Conic

 

Origin Latitude: 10

 

Origin Longitude: 20

 

Standard Parallel 1: 20

 

Standard Parallel 2:-10

 

 

When I used the ESRIReprojector with a custom coordinate system using the same parameters as FME's definition, the measured area of the ESRI reprojected 6' x 6' square was within a square millimeter of the FME reprojected copy.

 

 

It would appear that the difference in measured area is due to the difference in the coordinate system definitions. I haven't been able to find an official definition for Africa Lambert Conformal Conic, so I can't say which definition is correct between the ESRI or Autodesk (FME) versions.

 

 

Kind regards,

 

Dave Campanas

Just came across this post. I've found discrepancies with a similar reprojection task. In what instance would it be appropriate to ever use the reprojector transformer in FME? I don't have much confidence in the tool if "The FME Reprojector isn't always very clever about which "path" it uses".

I'm not sure quite what is meant by the "path" that it uses. I suspect it means the method of reprojection when reprojecting from one datum to another. If you choose Tools > FME Options in Workbench and click on the menu option for Coordinate Systems, then you'll see the different datum shifts available in FME. Click on the Edit button and you'll see the order in which grid files are applied.

 

So, I'm guessing the "path" means the order in which grid shift files are applied. For example in NAD27 to NAD83 you'll see that there are multiple grid files. You would definitely want to ensure that the ones you give priority to are top of that list. If not then the "path" could be wrong.

 

Additionally, there are multiple reprojector transformers because there are multiple methods of carrying out reprojection. Most will result in slight differences to each other because of the underlying algorithm. If you want FME to produce data that matches Esri reprojections, then use the EsriReprojector transformer. If you want a match to the CSMap library, use the CSMapReprojector. The CSMapReprojector, by the way, is the only one that handles 3D data with vertical handling, so if you are wanting that capability, that's the only transformer to use.

 

I hope this helps. Incidentally, you might find you get more answers if you post this as a new question. Adding it here it might not be visible since folk will have to scroll past a lot of old comments to see it.