Hi
I have a pre-processed output of Sentinel-1 images GRDH
A GLCM output of GRDH images
As well as a coherence output of SLC images
I want to layer these three results together to classify these three outputs together.
But Layerstack is not done
Thanks for the reason why Layerstack is not available?
And I also Layerstack the intensity and texture images together and apply the classification, which I got an error with * no product reader for band ‘null’*
What is the reason for this error?
are the SLC data deburst and multi-looked already?
The easiest solution is to apply terrain correction to all products before you apply layer stacking (or collocation)
Hi
Yes, the steps are as follows
To perform coherence using SLC images:
Split - Ord - back Geocoding - coherence - Deburst - ML - TC
To perform GLCM using GRDH images:
Split - Ord - Cal - Deburst - ML - TC - GLCM
And for pre-processing images and extracting the intensity of Sentinel-1 images using GRDH images:
Split - Ord - Cal - Deburst - ML - TC - dB
Now I want to layer these three results together
But it is not possible
Have you tried Collocation instead?
No, I have not tried, I do not know how to proceed
The collocation module is explained here: Synergetic use of S1 (SAR) and S2 (optical) data and use of analysis tools
Thank you
I also Layerstack the intensity and texture images together and apply the classification, which I got an error with * no product reader for band ‘null’*
What is the reason for this error?
Looks like one of the products was not correctly written. Can you please visually check all bands you want to input into the step?
Also, did you save all products in the BEAM DIMAP format?
Hi
Visually, the image output intensity is 10 meters
But the texture output I feel is different due to the extraction of pixel texture and the 7 in 7 window is used
And I did not save the image output, it is the same output stored inside the software
the texture contains empty pixels which are defined as NaN (not a number). Maybe this is the reason for the error. You can disable the NN value in the band properties (don’t forget to save the product after changes) of all texture images.
It also looks kind of blurry and has a systematic offset. Something looks wrong to me in the second image.
How can I disable NN images?
open the band properties and change the settings here
Alternatively, you can set a custom NoData value directly in the GLCM operator
Thank you
I will check
Does the RF classifier work with the original product (without textures) as input?
Yes, I did the classification once for GLCM and once for intensity separately, it was done correctly and there was no problem.
Have you tried renaming all bands of the collocated stack before classification?
You can save it as a separate product. An example is given here: Landcover classification with Sentinel-1 GRD
No, I did not change the names before classification
Please give it a try.
I did the classification based on the tutorial file, that’s right.
There is no problem when we enter the GLCM results and the intensity results each in a separate file in the classification section
The problem arises when we combine the intensity band with the GLCM band
When Layerstack is done, no classification is done after this operation