[Gdal-dev] USGS vs. ESRI vs. OGR

[Frank Warmerdam]

Grabowski, Hank wrote:
> I followed your advise to compare to the USGS tool.  The conversion
> results are listed below.  As we can see, the ESRI and USGS tools match
> identically (to the precision of ESRI which was lower than the USGS
> solution).  There is a convergence parameter from the USGS calculation
> which is significantly larger than the error between OGR and ESRI/USGS.
> Being a novice at this reprojection business, I don't know if that is
> the bounds of the accuracy of the solution.  I would tend to believe not
> however, since a 40 minute minimum accuracy would be a multi-kilometer
> error in position.  On a hunch I looked at the two definitions used for
> WGS84 by both OGR and ESRI, and their coefficients match to reported
> accuracy.  I'll continue investigating this issue if no one has any
> ideas what could be causing it.
> 
> UTM X	597196.62	Meters		
> UTM Y	4116325.5	Meters		
> 				
> Product	Lat			Lon			Convergence
> Scale
> 		DD MM SS.sssss 	DDD MM SS.sssss 	DD MM SS.ss
> Unitless
> USGS		37 11 24.57282 	085 54 17.71610 	0 39 43.15
> 9.99716E-001
> ESRI		37 11 24.57 	085 54 17.72		
> OGR		37 11 24.76	085 54 17.62		

Hank,

I am a bit confused.  When I do UTM16 NAD27 to geographic NAD27 I get:

cs2cs +proj=utm +zone=16 +datum=NAD27 +to +proj=latlong +datum=NAD27
597196.62 4116325.5                               (input)
85d54'17.716"W  37d11'24.573"N 0.000              (output)

It would *appear* that your locations for the USGS and ESRI are the NAD27
lat/long location, not the NAD83 location unless the datum shift at the
location is less than 1/100 of a second.

I broke down the conversion into steps from UTM 16 NAD27 to Latlong NAD27,
and then used "nad2nad" explicitly with the conus file just to be sure
that my code for cs2cs wasn't mixing things up, and I get the same final
OGR output coordinates you did.  While the ESRI software uses a somewhat
different grid shift file than ESRI, I believe that the PROJ/OGR software
is doing the right thing given the values in the conus file.

In summary, I question whether your USGS/ESRI numbers really represent
applying a datum shift at all.

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent