Speckle filtering for interferogram fomation


I am using sentinel 1 SLC images to generate a differential interferogram. However, due to the bad image coherence, I did a speckle filtering after the radiometric calibration. The problem is that the interferogram formation cannot be performed after applying the speckle filter and the following error appears : Input should be a single look complex SLC product.
Thus, I don’t understand if speckle filtering could be performed before an interferogram formation as a pre-processing step to improve the coherence.
BTW, the radiometrically calibrated product was saved in complex format which makes the situation even more confusing ??

Thanks in advance for replying.

Input should be a single look complex SLC product. is self explanatory. You need products that are Single Look Complex (SLC). I guess you are using GRD (Ground Range Detected) products?

Usually InSAR is done with SLC products.

To better answer you could you detail the steps you took?

Hi elmerehbi and thanks for replying.

I am using the following steps :

  • Radiometric calibration.
  • Co-registration with ESD.
  • Multi-temporal speckle filter.
  • Interferogram formation.

The strange thing is that I’m using SLC products which works just fine without speckle filtering. However once you perform speckle filtering this error appears. Does the speckle filter change the product type ? This is my question.

you need the i and q bands of the SLC products. They are removed after radiometric calibration.
As InSAR is based on the phase information, speckle reduction has no effect either because it is only affecting the intensity part.

Hi ABraun

On the countrary, it is the speckle filter that causes the loss of the i and q bands. When I go to parameters and try to select the three bands (i,q and intensity), the software doesn’t allow that and I get a product with only the intensity band.
Any Ideas on a way to walk around this anf perform a interferogram formation with speckle filtered images ??

The speckle filter removes the i+q bands because it is applied at the intensity band which consists of i and q, any modification.

Speckle filtering the intensity does not improve the interferogram quality at all. Instead, you can directly compute the interferogram after the coregistration and then apply the goldstein filter or azimuth/range phase filtering (found in the interferometric menu).

Ok thank you for the information !