Useful GDAL commands for working with hyperspectral data

The Geospatial Data Abstraction Library (GDAL) contains a number of utilities for working with remote sensing data. The following are some commands which are particularly useful when working with airborne hyperspectral data.

Create subset and change format

To create a band and spatial subset of a larger hyperspectral dataset the following command can be used:

gdal_translate -of GTiff -co COMPRESS=LZW \
 -projwin 481500 766000 482500 765000 \
 -b 190 -b 40 -b 15  \
 outputs/f281013b_utm_wgs84N50_mapped_subsection.bil \
  • -of: Output Format – we are using GeoTiff file. For all available options see
  • -co: Creation Option – specifies the creation options, these are specific to the format. For GeoTiff we use compression to reduce the file size. For ENVI we can use the creation options to specify the Interleave (BSQ or BIL)
  • -projwin: A rectangle delineating an area to be extracted, in the format min X min Y max X max Y
  • -b: A given band to extract
  • final two arguments: input and output filenames

Create mosaic of multiple flightlines

Often a mosaic of all individual flightlines is required. Due to the large sizes of the mapped files this can often be time consuming. One option is to first make a ‘Virtial Raster’ using the ‘gdalbuildvrt’ command:

gdalbuildvrt -srcnodata 0 outputs/f2810_mapped_mosaic.vrt \
 outputs/f281013b_utm_wgs84N50_mapped_3band.bil \
  • -srcnodata: No data value of input lines
  • first argument: output VRT file
  • final arguments: list of input files

Rather than creating a new file a small .vrt text file is created which contains links to the individual lines.

Many programs which use GDAL (e.g., TuiView) can open this file directly. To make opening the file easier we can do some optimisation. First pre-calculate statistics using:

gdalinfo -stats outputs/f2810_mapped_mosaic.vrt
  • -stats: Calculate statistics and save to .aux.xml file
  • final argument: file name

The gdalinfo command can also be used to print information about a raster. Note for the .vrt file it shows the individual files which are included.

After creating statistics generate overviews (lower resolution copies at different resolutions) using gdaladdo:

gdaladdo -r average outputs/f2810_mapped_mosaic.vrt 2 4 8 16
  • -r average: Use average for resampling overviews
  • first argument: file name
  • numbers: overview levels to build (2x, 4x, 8x and 16x)

We can also make a real raster using gdal_translate (as shown previously).

Extract pixel values for a given location

gdallocationinfo -valonly -geoloc \
 outputs/f281013b_utm_wgs84N50_mapped_subsection.bil \
 481801 765289
  • -valonly: Only print values to the screen
  • -geoloc: Interprerate x and y values as coordinates rather than pixel and line
  • first argument: file name
  • x y: Easting and Northing of pixel to extract statistics from

This will print the values to the screen. To save as a text file we can redirect the output using ‘>’

gdallocationinfo -valonly -geoloc \
 outputs/f281013b_utm_wgs84N50_mapped_subsection.bil \
 481801 765289 > outputs/extracted_values.txt

This text file can be opened in other programs for further analysis.

This post is based on the hyperspectral practical session from the NERC Airborne Survey and Research Facility workshop held at Plymouth Marine Laboratory, March 2016. For more information about the ARSF Data Analysis Node (ARSF-DAN) and details about future training opportunities follow @ARSF_DAN on twitter. Hyperspectral data acquired by ARSF are available to download from NEODC.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s