I would like to classify Landsat 8 imagery so I can quantify Recently Harvested/Disturbed Forested Areas . I am interested in doing this for multiple scenes over large areas.
My idea is to create training classes primarily in recently harvested areas, (those which I can see on 2012 NAIP ortho imagery), but also in young forests, mature forests, urban areas, grasslands/shrublands and water. I plan on doing this separately for each scene using a 5,4,3 composite image to run the classification. I will likely have trouble distinguishing between classes like young and mature forests, but I'm mostly concerned with the areas that are recently harvested.
Once each scene is classified in this manner I will mosaic the resulting classes, as opposed to mosaicing the images before classification. Any criticisms, help or general suggestions would be greatly appreciated.