sarproz-group

A new (and major) update for Sentinel S1A

Tagged: 

Viewing 12 reply threads
  • Author
    Posts
    • #1649
      Yuxiao
      Spectator

        Hi all,

        Sorry for the slow updating progress for Sentinel S1A module during the past semester. In the next new update of sarproz there will be a major update for Sentinel S1A module. The new one includes the following major updates:

        (1) Sarproz will automatically read precise orbit file (with a standard deviations of less than 2.5 meters) for Sentinel. In the case where reading precise orbit files fails (because precise orbit files aren’t available after one month of the acquisition, so if your image is within one month of your current date there’ll be no precise orbit available), the software will read the less precise Restituted Orbit (with a standard deviation of less than 10 meters) until precise orbit is published;

        (2) Sarproz is now able to read in TOPS dual polarization data (if available). Thus if you want to play with dual polarization you will have the option to import all polarizations files or either polarization files.

        (3) TOPS Fine-Coregistration algorithm has been updated. The updated algorithm will increase the accuracy of fine-coregistration and decrease the chance of getting phase jumps at the edge of stitched bursts for your interferograms.

        (4) Other minor fixes.

        For the next step of Sentinel S1A module development, we will concentrate on improve computational efficiency, and implement codes that will automatically stitch adjacent bursts and automatically stitch adjacent scene. We also have all the reason to believe sarproz will be able to handle the newly launched S1B once it is released.

        Thanks to you all for your precious comments and feedbacks to the code!

        Yours,
        Yuxiao

      • #1651
        maghsoudi
        Participant

          Hi Yuxiao,

          Thanks for the ongoing updates for the sentinel-1. I previously applied the coregistration for a series of sentinel-1 images using the older version. However, using this new version an error message pops up just after doing the deramping. I attached the error message.

          Regards,
          Yasser

          • #1654
            Yuxiao
            Spectator

              Hi Yasser,

              you will need to zip the lasterror.mat together with the log file and upload the zip file so that I can take a look.

              Thanks!

              Yuxiao

            • #1655
              Yuxiao
              Spectator

                Hi Yasser,

                you will need to zip the lasterror.mat together with the log file and upload the zip file so that I can take a look.

                Thanks!

                Yuxiao

            • #1656
              maghsoudi
              Participant

                Attached!

                Attachments:
                You must be logged in to view attached files.
              • #1677
                benl
                Participant

                  actually, i face the same problem ever since updating to the latest version. I tried setting the Max Area flag, but still, there is an error during the coregistration process.

                  • #1683
                    Yuxiao
                    Spectator

                      Hi Benl,

                      If you are selecting maximum area, you are not having the same error as Yasser. It would be great if you could post your lasterror.mat and log file so that I can know where the error occurs.

                      For this particular issue, I have located another bug in the code that sometimes will result in an error. I have fixed it and will update sarproz really soon. If the new one still doesn’t fix your error please submit us the log and error file.

                      Thanks,

                      Yuxiao

                  • #1717
                    maghsoudi
                    Participant

                      Hi Yuxiao,

                      I am still receiving an error message while I am trying the co-registration for the sentinel-1 images. I attached the error code. I would be grateful if you could help me on that.

                      Yasser

                      Attachments:
                      You must be logged in to view attached files.
                      • #1719
                        Yuxiao
                        Spectator

                          Hi Yasser,

                          The bug is fixed and the next version in sarproz update should see the solving of this bug.

                          Meanwhile, please take a look at your log. There are a number of warnings like this:

                          WARNING!!! WARNING!!! WARNING!!! WARNING!!! WARNING!!! WARNING!!!
                          Your area is TOO SMALL for an accurate subpixel coregistration!!!
                          A linear azimuth ramp may appear on your interferograms as the result!!!
                          You can try to discard this problem by selecting lines greater than 1500!!!
                          WARNING!!! WARNING!!! WARNING!!! WARNING!!! WARNING!!! WARNING!!!

                          It is generally a good idea to take a look at the warning and follow the recommendations from that warning message, or you might get unreliable results..

                          yuxiao

                      • #1720
                        maghsoudi
                        Participant

                          Thanks Yuxiao. I noticed the warnings. My area was about 4 by 4 km2 in the beginning. I increased the area to 5 by 5 km2 but I received the same warning message. I am not able to increase my area as it will contain a very large non-coherent area which again may cause problems to the co-registration process.

                        • #1721
                          maghsoudi
                          Participant

                            By the way, when do you think the new update will be released?

                            • #1722
                              Yuxiao
                              Spectator

                                Tomorrow by latest. Meanwhile, I would suggest to increase your area step by step until you reach maximum area. If you get these warnings, it would mean that it will be very difficult to obtain reliable results in your later on processing.

                                yuxiao

                            • #1723
                              maghsoudi
                              Participant

                                Hi Yuxiao,

                                I installed the latest version as of 06 June and repeated the procedure. Again, the same error message pops up as per attached.
                                Also, I would like to know what exactly is being performed in the “TOPS Post Processing for each image pair”. In my case, for 5 pairs I am receiving a warning message showing lack of coherency in my selected area and the pair is rolled back to SLC before Post Coregistration. I was curious to know what exactly is being missed in this rolling back.

                                Regards,
                                Yasser

                                Attachments:
                                You must be logged in to view attached files.
                              • #1725
                                maghsoudi
                                Participant

                                  when I increase the region to the ‘max area’, the co-registration works without any error. But the problem is that this area is too big for the performing the ‘site processing’ steps and is very time consuming. It would be good if there was an option right at the end of the co-registration in which we could subset the area of interest (this option is before doing the co-registration at the moment).

                                  • #1726
                                    Yuxiao
                                    Spectator

                                      Hi Yasser,

                                      Sorry for the late reply. I forgot to submit the code last weekend. I’ve just submitted the correct code to the developer yesterday and you should be bug-free in the next update of sarproz. Sorry again.

                                      A description of tops post processing can be find in chapter 4.3 of this conference paper: http://seom.esa.int/fringe2015/files/Contribution116.pdf. Generally speaking this process do the coregistration again to make sure the accuracy of TOPS coregistration meets the requirement (roughly to the accuracy of 1/1000 pixel). If this operation fails (for example, due to lack of coherency, or not large enough of AOI), a backup of SLC image generated with steps in chapter 4.2 of the paper will be restored, called “rolling back”.

                                      The rolled back version usually only has a coregistration accuracy to the magnitude of 0.1 pixel. This is very much enough for usual stripmap mode, but not for TOPS mode. Thus if you’re using the rolled back SLCs, there might be a linear phase ramp in your interferograms.

                                      At last, if your area is too big and too time consuming, after you do the coregistration, you can go to site processing -> selection and extraction to extract a smaller area. Here you can select an arbitrary small size (that make sense in your analysis) for your processing.

                                      Yours,
                                      Yuxiao

                                  • #1728
                                    ivca-cz
                                    Participant

                                      Hello,

                                      please, if the new S1 (coregistration) scripts were already updated, can you also please upload the linux-compiled version? The last version available is still 12-May.

                                      Thank you very much, Ivana

                                    • #1729
                                      periz
                                      Keymaster

                                        Hi Ivana,
                                        a new linux version is now available!
                                        d

                                      • #1730
                                        ivca-cz
                                        Participant

                                          Hello,

                                          thank you for the new version.

                                          I still encounter an error with multiple imaging modes detected. For this one set, on my computer (did not try yet this dataset on the supercomputer, and did not try other datasets on my computer). I already reported this problem in the beginning of April.

                                          The problem emerges in the “get contents” step.

                                          I finally managed to delete the one image (reported to be both asc and desc) from the dataset and the problem repeats with the following image (previously, it was 20150329, now it is 20150410). Last year, I successfully processed both images, but now I want to update the dataset with new images.

                                          Please, do you know what can be the problem in this case?

                                          Thank you, Ivana

                                        • #1743
                                          ivca-cz
                                          Participant

                                            Hi Yuxiao,

                                            I am sorry for this error, it is my mistake. I really mixed one ascending image into the descending set. However, the date reported is significantly different from the one it really was: while the ascending image is 20160224, the reported one was 20150329, and then 20150410.

                                            Very sorry for that,

                                            Ivana

                                        Viewing 12 reply threads
                                        • You must be logged in to reply to this topic.