3

This question concerns work-flow rather than specific methodological problems.

I will be using GRASS to classify Landsat 8 images to create a vegetation map for a study site that happens to fall over 2 separate Landsat scenes.

I would really appreciate some advice about which route to take and I'm not able to find any other information about this elsewhere:

1.) Should I FIRST find a way to combine the 2 scenes together (which presumably would require combining the 2 scenes of each channel separately - would that be right?), and then cut out the small area covered by my study site and THEN start the classification process.

OR

2.) Should I conduct image classification on the 2 adjacent Landsat scenes separately and then combine right at the end? (Although I'm worried this will take much longer, not to mention use up more memory - the study site I am working with is only 1400 km2)

N.B. I am using a Mac (Mavericks OSX) and have access to both QGIS 1.8.0 and ArcView 9.2 to conduct any manipulation on this Geotiff files.

Thank you very much for your time and advice!

  • If it was me I would be combine the two scenes into a single raster, this would simplify any downstream processing as you are dealing with just 1 raster. For example you've done your classification then want to do some zonal stats. Doing zonal stats on 1 raster is much more convenient than attempt to do it across separate rasters and you are less likely to introduce processing errors. – Hornbydd Nov 7 '13 at 12:41
  • Thank you very much for your advice. However, working with Landsat 8 imagery (and therefore 11 different channels) does that mean I should combine the 2 scenes for each channel and so in fact be working with 11 rasters? – Char_leopard Nov 8 '13 at 2:37
  • You would combine the bands you need for your study then merge those scenes (well that's what I would have done). I've not used Landsat 8 yet put there is advice on which bands you would want to combine here – Hornbydd Nov 8 '13 at 10:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.