Skip to content

Problem with spTransform en +towgs84 ?

3 messages · Javier Garcia-Pintado, Roger Bivand

Original
Javier Garcia-Pintado
# 
Hi,
Just to add that I've also detected some problems in this sense. I've have
exported a shapefile from ArcGIS to KML, and also imported it into R, and
from there to KML. The final KML files differ in some 70m of latitude,
while the longitudes match.

# import the shapefile
[1] " +proj=tmerc +lat_0=49 +lon_0=-2 +k=0.999601 +x_0=400000 +y_0=-100000
+ellps=airy +units=m +no_defs"

# Note, the above is OSGB 1936 / British National Grid, which  equals
# CRS("+init=epsg:27700")
# Then:

knapdale.supgeoLL <- spTransform(knapdale.supgeo, CRS("+proj=longlat +
datum=WGS84"))
kmlf <- "knapdale_supgeo.kml"
writeOGR(knapdale.supgeoLL, dsn=kmlf, layer="supgeo", driver="KML")

I'm not saying that the R part is responsible for the difference, but the
mismatch is there. Some details:

ArcGIS 9.3 (Windows XP Pro)
R 2.10.1
Loading required package: sp
Geospatial Data Abstraction Library extensions to R successfully loaded
Loaded GDAL runtime: GDAL 1.4.2.0, released 2007/06/27
Path to GDAL shared files: /usr/local/share/gdal
Loaded PROJ.4 runtime: Rel. 4.5.0, 22 Oct 2006
Path to PROJ.4 shared files: (autodetected)

Best regards,
Javier
Javier Garcia-Pintado
# 
Hi again,
However,in my case, if instead of the previously mentioned version in my
linux box, I use my windows version of R:
Loading required package: sp
Geospatial Data Abstraction Library extensions to R successfully loaded
Loaded GDAL runtime: GDAL 1.6.3, released 2009/11/19
Path to GDAL shared files: C:/ARCHIV~1/R/R-29~1.2/library/rgdal/gdal
Loaded PROJ.4 runtime: Rel. 4.6.1, 21 August 2008
Path to PROJ.4 shared files: C:/ARCHIV~1/R/R-29~1.2/library/rgdal/proj
The mismatch problem is solved. So, could this be, in this case, a problem
of the PROJ.4 Rel. 4.5.0, 22 Oct 2006, or GDAL 1.4.2.0 versions in the
linux computer?

Thanks,

Javier
---
Roger Bivand
#
On Fri, 24 Sep 2010, jgarcia at ija.csic.es wrote:

            

      
      
      
    

        
Older releases of the EPSG database for PROJ.4 and GDAL did not 
necessarily include +towgs84= parameters for known +datum= - this has 
improved with time. The burden is, though, on users to understand that 
both the datum and the ellipsoid must be known, and in the case of 
ambiguous "datum" definitions like ED50, what the +towgs84= should be.

See projInfo("datum") for known datum definitions.

Currently, the GDAL version is 1.7.2, and PROJ.4 is 4.7.0 (but declares 
itself as 4.7.1). These are included in CRAN binaries for Windows and CRAN 
extras binaries for OSX. If you install from source under Linux, these are 
also what you should have, and Debian/Fedora packagers for GIS are usually 
not far behind, and checking for updates in some fashion should be 
possible. In R, do make a habit of running sessionInfo() and recording the 
output if you need absolutely reproducible research; then 
update.packages() to pick up bug fixes and enhancements. If you need to 
step back, all older versions of packages are also available on CRAN.

It is known that datum specifications are a problem, which is why one 
needs to be particularly careful about correct +towgs84= parameters.

Hope this helps,

Roger