Burned Area Reflectance Classification BARC, quick synopsis

barc_guide.md

Inspiration:

http://www.fs.fed.us/eng/rsac/baer/barc.html

http://www.mapbox.com/blog/putting-landsat-8-bands-to-work/

http://www.mapbox.com/blog/processing-landsat-8/

…quick [BARC] synopsis:

1.  Obtain pre and post imagery.
   

2.  Correct that imagery to Top of Atmosphere reflection (... for L8 see http://landsat7.usgs.gov/Landsat8_Using_Product.php)
   

3.  Calculate NBR for each image (NIR-SWIR)/(NIR+SWIR)
   

4.  Calculate dNBR where dNBR = pre NBR – post NBR
   

5.  Generate the BARC256 by shifting, compressing and truncating the dNBR to fit into a 8-bit image where BARC256 = (dNBR+275)/5
   

• Apply a gap mask if using L7 data.
  
  

6.  Manually threshold the BARC256 into burn severity classes, Unburned, Low, Moderate, High.
   

7.  Reclassify the BARC256 into a 4-class thematic raster (BARC4) based on the manually selected breakpoint.
   

8.  Generate metadata.
   

Process, with ImageMagick:

composite LC80410362013117LGN01_B5.TIF -compose minus_dst LC80410362013117LGN01_B7.TIF pre-nbr1.tif

composite LC80420362013124LGN02_B5.TIF -compose minus_dst LC80420362013124LGN02_B7.TIF post-nbr1.tif

composite LC80410362013117LGN01_B5.TIF -compose plus LC80410362013117LGN01_B7.TIF pre-nbr2.tif

composite LC80420362013124LGN02_B5.TIF -compose plus LC80420362013124LGN02_B7.TIF post-nbr2.tif

convert pre-nbr1.tif pre-nbr2.tif -compose Divide_Src -composite pre-nbr.tif

convert post-nbr1.tif post-nbr2.tif -compose Divide_Src -composite post-nbr.tif

composite pre-nbr.tif -compose minus_dst post-nbr.tif dnbr.tif

convert dnbr.tif -evaluate add 275 -evaluate divide 5 barc256.tif

Other to consider: PCA-derived burn index, https://twitter.com/vruba/status/347489714389000194 - is that https://en.wikipedia.org/wiki/Principal_component_analysis ? That's a new one to me.