Crop Removal Monitoring - Sentinel 1

The reviewer makes a fair point - speckle filtering should be done in radar geometry, in other words before Terrain Correction.

Does this mean that I need to do single speckle filter (e.g. Lee Sigma ) on each product before Terrain Correction?

If my understanding is right, the multi-temporal filtering is essentially the same with single speckle filter (Single or Multi-temporal speckle filter?). Does that mean I donā€™t need to do the multi-temporal filtering after the single filtering has already been done for each product?

The new processing steps would be:

  1. Read
  2. Apply Orbit File
  3. Radiometric Calibration
  4. Single Product Speckle Filter
  5. Terrain Correction
  6. Write
  7. Subset

Multitemporal filtering usually works better that single image filter - choose one or the other.

Ok, I will use the multi-temporal filtering. Just want to make sure I am doing it right. I list my steps below.

  1. Read
  2. Apply Orbit File
  3. Radiometric Calibration
  4. Subset
  5. Stack
  6. Multi-temporal filtering
  7. Terrain Correction

Please let me know if anything is wrong. Thank you.

You can only do multi-temporal speckle filtering BEFORE terrain correction if you use data from the same relative orbit. So, you canā€™t mix descending and ascending and neither overlapping orbits from different relative orbits.

The overall difference in doing (single-temporal) speckle filtering before or after terrain correction is somewhat trivial, esp. in flat (agricultural) terrain and depending on the filter (Refined Lee would be OKish). In general, you should use speckle filtering primarily for ā€œcosmeticā€ improvements of the pictures that go with your article. To compare time trends for a single field, better average for the whole field, and skip speckle filtering.

2 Likes

Guido, thanks for your answer and detailed explanation. I am only using the images from the same orbit with ascending pass in the multi-temporal speckle filtering. I will use the process steps I mentioned earlier in this work.

I have been reading many great materials about crop monitoring from you. Thanks for your contribution.

Best,
Yuting