Simple Discriminants
This procedure allows the remote sensing operator to pull out features that are visible on one or more bands, ratios, or principal components in the scene. On the left is the PC 3 for the clip of the local Landsat image. It shows the urban/developed, bare graound, water as a high value. Can we seperate water from bare ground and urban? Picking out all the cells with PC3 > ?11k first, we isolate just these whiter cells (open band 3 from the PC dataset separately so you can classify it in two levels, greater than and less than some threshold).
Looking at the rest of these PCs (or raw bands), how could we pull out city versus bare fields?