Message-ID: <dc22b2570911200914h59ed06b2l934ae255db4f01b2@mail.gmail.com>
Date: 2009-11-20T17:14:33Z
From: Robert J. Hijmans
Subject: changing the data type of a gdal dataset
In-Reply-To: <4B062112.3080809@umontana.edu>
Alan,
You can have a look at the code in the raster package on R-forge (see
the writeGDAL function). Or perhaps use this package: one of the
primary reasons for developing it was to automate (and hide) block by
block raster processing --- and quite successfully, it is being used
to process raster files that are too large for the leading commercial
gis software. See raster::predict to apply a model to a set or raster
predictors.
Robert
On Thu, Nov 19, 2009 at 8:54 PM, Alan Swanson <alan.swanson at umontana.edu> wrote:
> Dear R gurus,
> I have a function that applies various model prediction functions over a set
> of large image files, producing a single output file with the same spatial
> extent. ?Due to memory issues, I'm breaking the input and output files into
> tiles. ?I have this working except for one small issue regarding data types.
> I create a new gdal transient dataset by copying an existing one using:
> handle <- GDAL.open(fullnames[1],read.only=T)
> tds <-
> copyDataset(handle,driver=new('GDALDriver','GTiff'),strict=F,options=NULL)
> ...
> putRasterData(tds,t(preds), offset= c(strt[1], 0))
> ...
> saveDataset(tds,outfile.p)
> GDAL.close(tds)
> ? ? Which works great, except that my output always needs to be floating
> point, but the input may be byte or integer, in which case the output
> dataset retains the format of the input file. ?So I either need to change
> the data type of the new file, or create the new file using:
> tds <- new("GDALTransientDataset",driver,dims[1],dims[2], type="Float32")
>
> and then copy the spatial reference information from an existing dataset. ?I
> can't figure out how to do either of these. ?Your help would be much
> appreciated.
> Cheers,
> Alan
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>