Writing a tif using rgdal

I seem to remember writing a wrapper for the gdal copy function,
although none of that has be extensively tested. If you copy an existing
dataset to a writable dataset, you could then retrieve the data, operate
on it and write back to the new dataset. It should (?) then have all the
metadata copied from the original. 

Sorry, I didn't see in your code
where you were overwriting gdal pointers.

THK

On Tue, 2006-02-07 at 20:19 +0100, Roger Bivand wrote:

On Tue, 7 Feb 2006, Andy Bunn wrote:

Howdy List:

I'm reading in a large number of satellite images and analyzing them in R. I
know how to read a tif into R using rgdal and manipulate it. Is there a good
way to write a new tif out using the projection data from the original?

So in the example below how can I write out z as a jpg using the metadata
and such from x?

-Andy


library(rgdal)
logo <- system.file("pictures/Rlogo.jpg", package="rgdal")[1]
x <- new("GDALReadOnlyDataset", logo)
displayDataset(x)
y <- getRasterData(x)
z <- rowMeans(y, dims=2)
dim(z)

Taking a deep breath and remembering to do save.image() frequently 
[overwriting GDAL handles can give seg.faults]:

library(rgdal)
logo <- system.file("pictures/Rlogo.jpg", package="rgdal")[1]
x <- new("GDALReadOnlyDataset", logo)
displayDataset(x)
y <- getRasterData(x)
z <- rowMeans(y, dims=2)
dim(z)
GDAL.close(x)

tif_driver <- new("GDALDriver", "GTiff")
getDriverLongName(tif_driver)
tif2 <- new("GDALTransientDataset", tif_driver, 77, 101, 1, 'Float64')
bnd1 <- putRasterData(tif2, z)
displayDataset(tif2)
tif_file <- tempfile()
saveDataset(tif2, tif_file)
GDAL.close(tif2)
GDAL.close(tif_driver)
tif3 <- GDAL.open(tif_file)
displayDataset(tif3)
GDAL.close(tif3)

$ gdalinfo zlogo.tif
Driver: GTiff/GeoTIFF
Size is 101, 77
Coordinate System is `'
Corner Coordinates:
Upper Left  (    0.0,    0.0)
Lower Left  (    0.0,   77.0)
Upper Right (  101.0,    0.0)
Lower Right (  101.0,   77.0)
Center      (   50.5,   38.5)
Band 1 Block=101x10 Type=Float64, ColorInterp=Gray

(I was playing with filename zlogo.tif, it reads into GRASS OK, but 
that's using the same GDAL drivers.)

Well, it works, though it's a bit too exciting dodging the seg.faults 
coming from over-writing live object handles while experimenting. For 
determined users, deleteDataset() works too (oh yes!). I'll try to wrap it 
up to be a bit smoother. No world file yet, but I guess that could just be 
copied? Suggestions welcome.

Roger

version

         _
platform i386-pc-mingw32
arch     i386
os       mingw32
system   i386, mingw32
status
major    2
minor    2.0
year     2005
month    10
day      06
svn rev  35749
language R