Problems with coherence from 2023 SLC products

Microwave Toolbox Interferometry
#1

Hello,
I am researching the use of coherence for agriculture.
Using the 2021 and 2022 SLC products, I performed the processing which looks like this:

  1. Split
  2. Apply orbit file
  3. Back geocoding
  4. Coherence estimation with 20x5 window size
  5. Deburst
  6. Terrain correction

The results I got for the 2021 and 2022 data were very promising, but when I processed the 2023 SLC products, the coherence either remained very low throughout the entire period, or it was completely random.
I’ve already double-checked for any possible processing errors. All processing for previous years was done before November 2023. I’ve noticed that there have been some changes to SNAP recently because I couldn’t download the orbit file, but the update fixed the issue.
Have there been any significant changes to any of the processing steps recently?
Or have there been any changes to the SLC products in 2023? To be honest, I’m running out of ideas, I know that precipitation can lower the coherence, but I don’t think that’s the problem here, it’s also the problem for all 4 tracks I’ve created coherence maps for.

Best regards

#2

@djagula can you help ?

#3

I don’t see a problem. Using a pair in Nov 2023 and following those steps, I get a reasonable result.
Could you be more specific on how to repeat the problem? What are the product names of a pair resulting in random coherence?

#4

Ok, I think the problem was that the study area in 2023 was very dry and then very wet which affected the coherence a lot and that is why the results diverged from previous years.
Do you have any ideas how the effect of precipitation on the interferometry can be negated but not by discarding the imagery? I am interested in using coherence to monitor crops over time.

#5

I don’t know exactly what might explain the weird coherence, but the fact is that SLC pixels carry both amplitude and phase info, which is super important for interferometry. So, any tiny change in product format or processing steps can mess with your coherence measurements.

Also, coherence can vary with image resolution and how much you’re multi-looking. Maybe that’s playing a part in the differences you’re seeing.