Crop classification using SENTINEL-1A SLC data

Microwave Toolbox
#41

Since the burst image is modulated in azimuth, the azimuth spectrum has a phase ramp. Generally if we want to perform operations between pixels, such as interpolation, we should apply deramping first to the signal and apply reramping later after the operation in order to preserve the phase information. Therefore, we should not apply multilook to the complex image unless the image is deramped.

1 Like
Deramping TOPSAR SLC data
#42

@jun_lu That’s totally true. However, for some reasons, I would like to deramp my images too. Does SNAP has any way to deramp my images?

In the back geocoding process, you can output the demod/deramped phase. But how to use it?

I tried the following :

  1. From the I/Q of the slave image, i recompute the amplitude and phase
  2. I remod the phase ramp and remove it from the slave phase image
  3. I recompute I/Q in order to have a deramped slave image

I still have this phase this phase ramp in azimuth. Did I do something wrong?

#43

Currently SNAP does not have an operator to deramp the image. I think in theory you can deramp the coregistered slave image using the demod/deramp phase output from backgeocoding. But I have never tried.

1 Like
#44

I tried (Several times actually) but never managed to deramp it. I will try again. If it works (or not), I will tag you with the procedures / results. It may interests other people too

1 Like
#45

Hi thanks for the steps given above . But for my data of SLC IW, the above steps worked out only upto Polarimetric Unsupervised classification… for the next step of decomposition it gives the exeception as “Input should be polarimetric”… Kindly clarify me …Thanks in advance

#46

Sentinel-1 has only two polarizations VV/VH, but for most decompositions you need quad-pol data.
Please see here: Sentinel-1 SLC Polarimetric Decomposition and here H-Alpha Plane Problems

#47

Hi,
Usually all these dual decomposition and classification are meant for polarimetric data with HH and HV or compact polarised data with RH and RV, but sentinel 1 data are available with VV and VH polarization. There must be some diffrence between the two.
Thanks