Skip to main content

I did some deeper research into my accuracy problem.  I appears that FME is introducing error in the y coordinate (Northing) but not x (Easting.  When I use the FME data Inspector I get coordinates in meters like 3422549 m Northing by 541778 m Easting, but I checked three different coordinate conversion programs and they all agree showing 3422556 m Northing by 514779 m Easting, a differenct of about 7.5 meters, or 25 feet!  This must be happening in the Reprojector translator I used.  Is there a way to set the accuracy of this conversion?

I tried changing some of the parameters but to no avail so far.

Any help would be appreciated.

Below is my initial question I asked:

 

 

I have a translation that converts from Garmin GDB which has Lat and Lon in WGS84 and converts it to UTM83-15F.  I noticed, however, that when I compare the position of my waypoints in my GIS software to those in Google Earth (Garmin Mapsource s/w has the ability to view the waypoints in Google Earth), they were off anywhere between 20 and 45 feet!.  The waypoints in Google Earth matched the data from the Garmin GPS unit to 5 digits, but the FME translation did not.  For my purposes this is too much error.

I then tested my GIS software which has an import system for AutoCAD DXF and Garmin Mapsource also has a way to save the GDB file into and AutoCAD DXF.  This is a round about and manual way to get the waypoints into my GIS software, but they are much more accurate, between 2.5 and 5 feet.

Can anyone guess as to why the large error is ocurring in FME?  I do use a reprojector to get the waypoints to UTM83-15F and I suspect that is where the error is ocurring.  Anyway to improve the accuracy of this conversion?

Did you use LL84 as the source coordinate system? Because that is the one you need for GPS data.
Hello HTL

 

 

FME tells me, that UTM83-15F uses Datum: NAD83

 

LL84 uses Datum WGS84

 

 

Datum conversions usually need approximations and interpolation and cannot be calculated "exactly" in mathematical sense.

 

 

From FME setup I remember being asked which of the Grid files I prefer to use, something like Canadian or US-version ...

 

 

So I am sure you need some kind of NTv2 transformation (involving grid files) to get the transformation with better results.

 

 

How to setup these is not very simple and to understand the Log file messages involved in this ia also something for the very insider ...

 

 

Michael

 


 

Many software can make the conversion but all should be set depending on the location you are. The error can be more or less depending the postion on earth and wrong equation use.

 

EX: I had a few months ago, 3 users who took the same gps location in the artic circle and with different Datum. When they came back and try to overlay their obsevation, they were off from each other  more or less 1500 ft (not a mispell, 1500! because error is greater north)

 

 

Won't go in all the  issues in details but  once I changed all the long lat to the dame Datum and THEN bring it to a single software in UTM NAD 83, they were all right on top of each other...

 

 

You could try converting your LAT/LON lon in NAD 83 using the official conversion technique of your location and after, LL83 to UTM NAD 83...

 

 

 


Hi HTL,

 

 

This sounds like a problem we in Safe Support would like to have a look at. Would you mind submitting this problem, along with a copy of your source GDB file, to http://fmepedia.safe.com/knowledgeSubmitCase?

 

 

Thank you,

 

 

Dave Campanas

 

Product Specialist
I have seen similar behaviour before. You may need to define a custom datum/ellipsoid/coordinate system. FME uses specific references for coordinate system conversion and depending on the software you are using, the definitions may not 100% line up. I have see this with Geomedia coordinate system conversions as FME uses the ESRI datum and ellipsoid which is slightly different and can produce different results. 
Well, I will try to respond to all of your answers.  After digging into it even further, I found that using the UTM-83IF conversion, I was very accurate, close enough to determine that any error I was seeing was due to picture distortion in Google Earth (errors of less than 5 feet). 

 

 

So now the GPS waypoints converted into my Mapcom M4 GIS system is much more accurate and lines up much more closely with the GPs waypoints passed to Google Earth.

 

 

Based on this and conversations with technical people at Mapcom, I come to the conclusion that FME's UTM83-15F probably conforms to what M4 calls UTM-15 US Survey feet, and the UTM83-15IF conforms to what M4 calls UTM 15 Feet.

 

 

I was not directly involved in the M4 GIS system setup, and am still waiting on an answer from Mapcom.  The person with our company who worked on the conversion from Cadtel to M4 is no longer with our company.

 

 

Thanks for all of you input, it does shed some light into these techniques.
Dave:

 

 

I wanted to answer to you specifically, about my problem though.

 

 

I calculated two different conversion factors from meters to match FME's UTM83-15F and UTM83-15IF and determined that the first has a factor of about 3.2808339 and the second has a factor of about 3.2808399 and found that the second very closely matched the projection of our Mapcom M4 GIs system, about 1 foot.  The first, however, is about 24 feet off.

 

 

I don't know if this is all you need, but respond again if you would like to look at a copy of my workbench.  I don't mind sending it in for you guys to analyze.

 

 

I do think, however, that knowing what projection was used in the conversion from Cadtel to M4 would help, and as I said in the previous answer, I am waiting on a response from them.

 

 

Let me know.  Thanks again.

Reply