Forest Non-Forest masking

From Open Foris Wiki

Jump to: navigation, search

This exercise provides an example of pixel based forest-non forest classification. The example uses a Landsat TM image as an input, extracts training observations from MODIS VCF -product and classifies the input image using the extracted training and Open Foris Toolkit Nearest Neighbour oft-nn classifier and estimator.


1. Download your Landsat scene using e.g. GloVis

2. Download the appropriate MODIS VCF tile

3. Subset the VCF image to the extension of the Landsat image landsat.tif VCF.tif clipped_VCF.tif 

4. Generalize the VCF product to forest/non-forest mask

oft-calc -ot Byte -um landsat.tif clipped_VCF.tif fornofor_vcf.tif 
#1 50 > 1 2 ?
5. Compute average Landsat response for both classes using:
 oft-stat fornofor_vcf.tif landsat.tif ls.stat
6. Clean the statistics file and keep only the averages.
awk '{print $1,$3,$4,$5,$6,$7,$8}' ls.stat > train.dat

7. Run Nearest Neighbour Classification to produce the Forest Non-Forest map with output FNF.tif

oft-nn -o FNF.tif -i landsat_t1.tif

Finally, have a look at the output with openev2/similar and try to figure out how to improve the output.

Back to Open Foris Toolkit Main Page

Back to Tools & Exercises

Personal tools