Thanks for this. It worked fine without resampling Could you please tell if there could be any correction(s) in any of the steps. Thanks in advance.
At the moment your steps are depending on the approach of your research, nothing wrong, as I mentioned in my previous post and @mengdahl also gave very nice comment concerning the resample step, now your BG, looks like this,
One thing you could do is, to apply speckle filtering before multilooking, as an trial step, and check up your result, I point to this issue according to this article,
Comparison of several multi-look processing procedures in INSAR processing for ERS-1&2 tandem mode
However, in your case the interferogram formation is not included in your steps, that’s why, it’s worth trying up. to be your BG as below,
Hey really thank you for this so valuable information which is difficult to find in literature. I have fe more doubts. First one is during calibration, I have to save as complex output or just sigma0 band. And what if I do not process each sub-swath individually and just directly do:
- Speckle filter
- Polarimetric Decomposition
- Terrain Correction
Will it take care of each sub-swath?
It depends on the goal of your study and the approach you’d like to follow, this issue is well explained by our colleague @ABraun in the following posts,
- Radiometric & Geometric Correction Workflow on the options of calibration
- Radiometric & Geometric Correction Workflow on Beta0, Sigma0 and Gamma0
- S1 radiometric correction on the effect of Terrain Flattening on images
It’s very easy to process each subswath individually and merge them later on,
I have gone through the posts and got ample of information Still I am not cleared with the difference between sigma0_VV and Intensity_VV ??
Please have a look at the following post, I mentioned, what is sgma and sigma0
Intensity is well defined in the following ,
That was very nice information thanks. Now I have cleared majority of my doubt. I started implementing the graph using gpt but the problem is it is taking a lot of time to compute even I took subset. Is it usual or something wrong with graph??
It depends on your GB and your machine, generally it should not take long time, one suggestions at the moment , don’t use the last graph you built, and create a new G. save the *.xml in different directory, use simple call by navigating to the directory where your data is available using command console and then call the gpt as this, “gpt mygraph.xml” and check up the results, let me know if it helps,
With all the help and typical tips, I performed the H-A decomposition. When I plotted the H-A plane I observed no surface scattering (low entropy low alpha) . My study area is some parts of Spain whose NDVI is varying from 0.2-0.6 (moderately vegetated). I cross checked with a region of THAR desert, there also a same curve was observed with no surface scattering mechanism. Any reason behind this??
I brought you some answers from the help of SNAP and from the STEP FORUM,
First take a look at the In H-A-Alpha decomposition calculations,
Second the eigenvalues of coherency plying an important role of the result,
most of the decompositions require 4 polarizations: HH, HV, VH and VV.
However, most of the Sentinel-1 data in IW mode is in only one or two polarizations:
So is it not possible to compute H-A decomposition for dual pol data? The post you referred was also facing the same problem with H-A decomposition. Is there any alternative we can use to compute H-A decomposition?
In the calculation of H-Alpha, there they mentioned eigenvectors u1, u2, u3. Do you know how to get this beta (orientation angle), gamma and delta(phase angles) used in the calculation of engine vector u. It will be a great help. I can code them separately then.
it is possible, but the outputs are not very good. I tried to explain it here: H-Alpha Plane Problems
In the named example, HH/VV is used and the data was simply not suitable to represent all scattering mechanisms. So VV/VH has even less information because it lacks horizontally transmitted signals.
Source : Analysis of circular polarization backscattering and target decomposition using gb-sar, https://www.researchgate.net/publication/315959521_Analysis_of_circular_polarization_backscattering_and_target_decomposition_using_gb-sar
Also following you could find good example,
Source : Scattering Mechanisms and Within-Field Variation, https://www.nrcan.gc.ca/earth-sciences/geomatics/satellite-imagery-air-photos/satellite-imagery-products/educational-resources/9585
Since sentinel-1 is monostatic as mention here Sentinel-1 : Monostatic or Bistatic? , VH and HV should not matter right? Also in my case nothing is coming as surface scattering even for a desert area
I even didn’t get how to compute beta(i), gamma(i), delta(i). Can we compute using SNAP?
Sorry I don’t know how to calculate it in SNAP, but I found the above mentioned article for you,
But in SNAP, the only orientation angle θ could be calculated,
I’ve attempted this with but the merge won’t accept H-alpha input. Am I doing this wholly wrong?
Have you managed to re-merge the sub-swaths after the polarimetric decomposition? I’ve tried with TOPS Merge but it doesn’t accept H-a inputs.
How can I solve this problem? I also encountered this problem, thank you.
Please describe your data and the steps you applied to it.