Layer stacking S1 and s2 from different date of acquisition

Microwave Toolbox
#1

Hi,

I have a question here.
So I have several scenes from S1 since 2017-2019.
I also have one scene from S2 from 2019.
I would like to perform a classification.
Is that ok, if I simply stack all my scenes from my sentinel 1 and 1 single sentinel 2 together and perform classification afterward?
Or I should have scenes that have a relatively similar date of acquisition?

Thank you
Looking forward to hearing from you

#2

If you want to include seasonal variations in your classification (e.g. changing volume scattering of broad-leaved forests vs. stable scattering of needle forests) you can input all images into the stack. But many of the information of the S1 images might be redundant. You could reduce these redundancies by performing a PCA on the S1 stack first and only take the first 6-8 pricipal components.

1 Like
#3

YES, I did PCA and it resulted good. Thanks

#4

that is a good compromise between using multi-temporal information on the one side and having an acceptable amount of data on the other side.