[R-sig-Geo] Processing HDF4 data files with R / GeospatialData Abstraction Library (GDAL)

[Rick Reeves]

Greetings:

Recently I augmented my installation of R ver 2.4.0 on Dapper Drake Linux
to read data files in Hierarchical Data Format Version 4 (HDF4) format. I
thought that the steps that I took might be useful to other R users who have
not yet had to modify one of the R packages to include a new capability.

Why did I need to have HDF4 support in R?
The rgdal package provides HDF image I/O support is however, the default
rgdal version downloaded from the archives supports only the HDF5 format,
which is not backwards-compatible with HDF4. Many spatial datasets
(for example, SeaWIFS digital ocean color imagery) distributed by
US government agencies are stored in HDF4 format.

Here is the procedure that I worked out:

To include HDF4, you have to download the HDF4 library
and then build a version of the  GDAL library that links
in the HDF4 support.

1) I downloaded and unpacked the hdf4,jpeg6 and szip libraries from:

 ftp.ncsa.uiuc.edu archives: 
/HDF: 4.2.rl-linux.tar.gz
/szip :/szip2.0-linux.noenc.tar.gz
/jpeg6b-linux-gcc3.4.3.tar.gz

  I did not need to build these libraries from source; instead, I just 
extracted
 the archive (.a) files  included with each distribution, and put them 
in a folder
 accessible to the GDAL library  build process. Call this library 
/opt/libs4GDAL
 for this discussion I put the following  files into this folder:
 libdf.a, libjpeg.a, libmfhdf.a, libsz.a,and libz.a

2) I built the GDAL library from source, linkin in the HDF file support.
    to do this: I ran the GDAL 'configure' script included with the distribution
   Note: To get a successful compilation, I had to modify the  configure script
   to include the correct location of the HDF source code include files:
   I added a line : HDF4_INCLUDE="-I/usr/include/hdf" at the place
   where the script tests for the existence of the include/hdf folder.Command
   used to run configure: ./configure --with-hdf4=/opt/libs4GDAL
   This built GDAL 1.4.0 library with HDF4 support

   To test the installation, I ran the GDAL gdalinfo and gdal_translate commands 
   from the command line to read several HDF4 SeaWIFSfiles. This succeeded

3) I started R from the command line, and installed GDAL support using the
    install.packages('rgdal')  command, which builds the library from 
source on
    linux machines.

4)I tested HDF support within R:

   library("rgdal")
   GDALinfo("an.hdf.file")
   hdfImage = readGDAL("an.hdf.file) // hdfImage is SpatialGridDataFrame
   image(hdfImage)   // displays the bitmap.

  These commands correctly read HDF files containing a single spectral
  band; however,  The rgdal routines do not appear to correctly read 
  HDF files containing multiple bands of imagery.
 
 The workaround here is to use the GDAL routines at the command 
 line  to extract the 'included' image planes into individual HDF4 files; 
 as noted above, rgdal routines correctly read the single band images.

Has anyone addressed the issue of processing multiple-band HDF4 files
with R / rgdal?

Cheers,

-- 
Rick Reeves	
Scientific Programmer / Analyst	
National Center for Ecological Analysis and Synthesis
UC Santa Barbara
reeves at nceas.ucsb.edu
www.nceas.ucsb.edu
805 892 2533

-------------- next part --------------
A non-text attachment was scrubbed...
Name: reeves.vcf
Type: text/x-vcard
Size: 339 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20070126/1ccb77ba/attachment.vcf>