I am working on a paper related to the land cover changes between 1984/85 and 2010/11 in my country and I am searching for Landsat 5 images from December 1984(2010) to March 1985(2011). Then I average band-by-band using raster calculator to build a single image for each period to evaluate land cover changes. Before averaging, each band will be corrected (ToA correction) and the image averaged will be corrected geographically.

Is this procedure correct?


I think you are still missing the classification part. Landsat 5 MSS has: Green, Red and 2 NIR bands.The TM has 7 Bands: RGB, 2 NIR, SWIR & Theraml.

Yet to create a land-cover classification one should process this data, mostly with supervised learning techniques (Though K-means is also used).

The procedure you show misses the classification itself. It should be as following:

  • Atmospheric Correction
  • Classification
  • Geometric Correction

Qgis provides tools for classification through the semi automatic classification plugin. You will also find video tutorials by the developer in his youtube channel: Luca Congedo. The plugin provide many capabilities, including: search and download of Landsat Imagery, atmospheric correction, sampling and classifying, etc.

  • 1
    If so you can mark it as an answer. Good luck – dof1985 Aug 10 '16 at 13:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.