PHD Discussions Logo

Ask, Learn and Accelerate in your PhD Research

Question Icon Post Your Answer

Question Icon

How can Landsat 8 image bands be used to distinguish different land covers in ENVI?

I'm new to remote sensing and tasked with creating a land cover map for a coastal district. I have Landsat 8 imagery loaded in ENVI, but I'm overwhelmed by the 11 bands. I need a clear, beginner-friendly strategy to choose which bands or indices to use as input for a supervised classification to reliably distinguish water, urban areas, and different vegetation types.

 

All Answers (1 Answers In All)

By Adi Answered 1 year ago

I teach this exact workflow. Start not with classification, but with visualization. Load a False Color Composite using Bands 5, 6, 4 (SWIR1, SWIR2, Red) - this brilliantly separates urban (cyan) from vegetation (dark green to brown) and water (black). Next, calculate indices in the Band Math tool: NDVI (B5-B4)/(B5+B4) for vegetation health, and a Modified Normalized Difference Water Index (MNDWI) using (B3-B6)/(B3+B6). For your classification input, I recommend stacking these indices along with key original bands like the Coastal Aerosol (B1) for water and the two SWIR bands. This composite will give your classifier much clearer spectral signatures to work with.

 

Your Answer