[GRASS-user] Confused when using i.fusion.brovey

[Markus Neteler]

On Mon, Jun 16, 2014 at 3:55 PM, Johan de Braak
<johan_debraak at hotmail.com> wrote:
> Dear all,
>
> I have used i.fusion.brovey to transform a Landsat 8 image (6 bands) to a
> cell size of 15x15. I have 3 questions regarding this tool:
>
> 1. Is the output reliable if I run this tool twice to convert all 6 bands?
> If not, what should I do to transform 6 bands with a reliable output? And
> why is it not possible to specify more than three bands?

The Brovey transform has been defined for using three multispectral
and the panchromatic bands:
http://grass.osgeo.org/grass64/manuals/i.fusion.brovey.html

That's how it was developed and suggested in Roller, N.E.G. and Cox, S., 1980.

> 2. When specifying the input, between brackets is mentioned: green tm2, NIR
> tm4, MIR tm5.

Yes, the *input* bands.

> The files created however are named red green and blue.

This is how they are commonly assigned to the R/G/B color
visualization tool, i.e. d.rgb.

> I have assumed that 'ms1' results in 'outputprefix.red', 'ms2' results in
> 'outputprefix.green' and 'ms3' resuls in 'outputprefix.blue'. Is this
> assumption true? Why are the output names different from what is mentioned
> between brackets in the input?

Because in the definition of this transform the blue band is not used.
In theory it is up to you which bands you input, in case you will
simply deviate from the original definition.

> 3. Is it recommended to use this tool before performing a supervised image
> classification with the output of i.fusion.brovey?

Personally I would not consider this since it may add correlation
between the bands.

Please also consider the updated version in GRASS GIS 7:
http://grass.osgeo.org/grass70/manuals/i.pansharpen.html

And perhaps this page:
http://grasswiki.osgeo.org/wiki/Image_classification

GRASS GIS 7 comes with new classification software:
http://grass.osgeo.org/grass70/manuals/g.gui.iclass.html
http://grasswiki.osgeo.org/wiki/Image_classification

Markus