TOPSAR-Split: half of the result missing

Hi,

as already mentioned here Back-geocoding returns always zero for slave , I am processing data over Finland for the years 2017-2018.

After solving the previous issue related to the DEM, I’m now facing another strange issue.
For a particular date 2018-02-04 (YYYY-MM-DD) the output of TOPSAR-Split misses half of the data. Look at the screenshots for a comparison between another date (2018-02-28), where the data is complete.


The graphs to reproduce this issues are attached below:

20180517_20180228.xml (5.9 KB)
20180517_20180204.xml (5.9 KB)

To verify that the problem is not in the source zip file, I’ve also run the process up to the calibration step for both dates, and the output has the same size. The process graphs to verify this are also attached below:

20180204_calib.xml (2.3 KB)
20180228_calib.xml (2.3 KB)

@jun_lu could you please help me here as well? Thanks in advance!

Kind regards,
Michele

I have looked at the graphs but I am not sure if I understand them. Could you provide more details on what you are trying to do? For example, which product is master, which is slave and we want to coregister which subswath/bursts from master with which subswath/bursts from slave etc. Thanks

I am co-registering one year of data to a single master date, which is 20180517.

In the provided example process graphs (20180517_20180228.xml 20180517_20180204.xml) I am running the following steps:

  1. Read (of master and slave)
  2. Slice assembly of the slave data
  3. Calibration
  4. TOPSAR-Split using bot polarizations (VV and VH) and a polygon for selecting the area of interest. I’ve set the bursts as
<firstBurstIndex>1</firstBurstIndex>
<lastBurstIndex>999</lastBurstIndex>

because I want snap to keep the data intersecting with the provided AOI from all the available bursts, since I do not know in advance which of them overlap with the provided polygon. Is it wrong doing like this?
The selected subswath is IW2.
5. Apply-Orbit-File (on both, master and slave)
6. Back-Geocoding using Copernicus 30m Global DEM
7. Write

The issue is with the 20180517_20180204.xml graph: as you can see from the screenshot the output is missing half of the data, which instead is present if you run the 20180517_20180228.xml graph. The graphs are the same except of the input data.

The other graphs were just a test to see if the output of calibration contained the full data coming from the slice-assembly step.

Please let me know if it is still not clear enough.

First I am not sure why you chose SLV images instead of GRD images. If there is no particular reason, I’d suggest you use GRD images to avoid a lot unnecessary processing.
Second, if you need to use SLC images, you should not use slice assembly. You should use something like the graph below:
image
We have a lot of SNAP tutorials on ESA’s website (Tutorials – STEP) and you will find that they are very helpful in guiding you how to use the operators.

Using GRD data is not an option, we need SLC data for our project.

I am using slice assembly because the area of interest is covered by two slices for Sentinel-1A and hence I need to merge them first.

Could you please try out the process graph I’ve sent you? I have not included the Deburst step, since the error happens before. It does not differ much from the one you are suggesting me to use. The visualized version of mine is the following:

Anyway, this is a bug report, since the provided process graph worked for 60+ images and fails on this one only.

Thanks in advance for the support.

@jun_lu are there any improvement about this bug? Were you able to reproduce it?

Please let me know if additional information to reproduce the behavior are missing.