Hi,
OK - I've followed Hamish's suggestion and used cs2cs from PROJ.4 (I'll get to Grass 5.3 soon) to check the location of a test point (0 m E, 7332956.368 m N) from a Behrmann projection back into geographic coordinates:
[doibook:/usr/local/bin] dorme% ./cs2cs +proj=cea +lat_ts=30 +datum=WGS84 +to +proj=latlong +datum=WGS84
0 7332956.368
0dE 86d18'12.382"N 0.000
This gives me the same geographic coordinate as I got from my original use of s.proj in Grass 5.0.2. To test ArcMAP again, I created a shapefile with a Behrmann projection containing a single point at the test location and used the project wizard in ArcToolbox to change to a Geographic projection (wgs84.prj). The resulting point has the following geographic coordinates:
0dE 87d 7' 58.42" N
The output of g.projinfo for my Behrmann location is below. I'm using Mac OS X (version 10.2.8) and the following version of cs2cs:
[doibook:/usr/local/bin] dorme% ./cs2cs
Rel. 4.4.7, 31 March 2003
usage: cs2cs [ -eEfIlrstvwW [args] ] [ +opts[=arg] ]
[+to [+opts[=arg] [ files ]
David.
-----------------------------------------------------------
PROJ_INFO file:
name: Equal Area Cylindrical
datum: wgs84
dx: 0.000000
dy: 0.000000
dz: 0.000000
proj: cea
ellps: wgs84
a: 6378137.0000000000
es: 0.0066943800
f: 298.2572235630
lat_0: 0.0000000000
lat_ts: 30.0000000000
lon_0: 0.0000000000
-----------------------------------------------------------
PROJ_UNITS file:
unit: meter
units: meters
meters: 1.000000
-----------------------------------------------------------
On Friday, October 10, 2003, at 10:50 am, Hamish wrote:
I e-mailed yesterday about a mismatch between Arcmap and Grass 5.0.2
transformations of sites from a global cylindrical equal area location
to a geographic. I've been playing around some more and it looks to me
as if either:1) s.proj has a bug in it,
2) arcmap has a bug in the transformation of sites or,
3) I've missed something elementary.I created a point shapefile in arcmap of the centre points of the grid
cells for the CEA location and read this into grass using s.in.shape.
I've checked that the values in the sites file are what I was
expecting. I used s.proj to project these sites into a geographic
(wgs84) location. I also used arcmap to project the original behrmann
points shapefile into wgs84 and then imported this into my grass wgs84
location using s.in.shape. The two sets of sites are not in the same
place - in extreme latitudes the discrepancy is about half a degree. I
won't attach the files to preserve bandwidth but if anyone wants them,
let me know.Any suggestions?
5.0.x does not do datum transforms.
Did you try GRASS 5.3? Is it still broken there?
Half a degree is pretty huge though.
Try with 5.3's m.proj2?
echo "easting northing" | m.proj2 -oTry with PROJ.4's cs2cs? (what grass 5.3 uses more or less)
http://www.remotesensing.org/projIf it persists after trying 5.3 can you file a bug at:
http://grass.ibiblio.org/bugtracking/bugreport.html
?
please include some specific test points with the results from both
GISs, output of g.projinfo, and any info you have on the projection
parameters.If you can't try 5.3, just post a bug report with the test point and
someone else can check.thanks,
Hamish