I have had very little experience with Landstat, and don't remember the bands, but try infrared wavelengths as they are good in differentiating uplands from wetlands in aerial photos, and I think deep water habitats can also be differentiated by the intensity of blue color. Cowardin et al developed a wetland and deep water habitat classification that has been used extensively in USA for map coverage Using remote sensing. I will try to send a digital copy, not sure what specifics you might find in methods, but I know some ground checks were required.

Hello,

The following links should help you in my opinion:

Regards,

Dr. Abhishek Mukherjee

Cowardin et al digital copy. You may need to contact the US Fish and Wildlife Service or perhaps look on internet if the remote sensing criteria used is not explained. Once you get the right aerial photos or remote sensing, much of the classification can be done with experience and some field checking. I do remember that forest clearcuts were sometimes confused with other land classes.