sarproz-group

GCP location error

This topic contains 14 replies, has 4 voices, and was last updated by  periz 2 months, 3 weeks ago.

  • Author
    Posts
  • #849

    cnu_lee
    Participant

    Hi professor,

    In the “GCP selection” module, no matter I choosed “keep current orbits” or “through external DEM”, the result of “Mean kml” has shown the location of automatic GCP was not matched with my study area and was too far away .And the amplitude map was far away from study area in Google earth ,but they couldn’t be displayed in the same screen. I don’t know why is this, my DEM data is SRTM, my dataset is 11 scenes Radarsat-2 images.

    Thanks, hope your reply.

    Lee

  • #850

    periz
    Keymaster

    In the GCP selection module there is a flag: ‘OVW’ which stays for “overwrite”.
    If you do not flag it, it will load the previous “mean kml” map.
    Since you are not the first one asking about it, I switched it to checked as default…
    You’ll get it in the next version, for the moment just check it manually.
    dp

  • #873

    cnu_lee
    Participant

    Thanks for your reply.

    I went back and extract the master area again, but I found the SAR image of extracted master area isn’t the area that I want. The centre of wrong area isn’t what I input in “latitude” and “longtitude”, but is the centre of one complete image. And I set the “latitude” and “longitude”, but the “samples” and “lines” doesn’t change.

  • #874

    periz
    Keymaster

    if you start the SLC processing over again,
    the Master and all previously generated files are overwritten.
    Also, the Master area is set as default: center in the middle of the scene.
    So, you have to set the Master area again. By the way, these operations were slightly improved recently: download the last version (or check in the updates history the version you are using is already containing the improvements).

    • #875

      cnu_lee
      Participant

      The version of software is May 6.

      I delete the old folders and import the data again.I set the master area again in “longitude” and “latitude” ,but the error was still here. The center of master area was still the center of the scene.By the way, the master area of kml format was right, but the SAR format image was not matched with the kml file.

  • #876

    periz
    Keymaster

    Can you please grab a few screenshots to show us your case?

    • #877

      cnu_lee
      Participant

      The first picture is SAR image,the other one is kml format master area.It’s obvious to find that area on the two pictures don’t match.

    • #878

      cnu_lee
      Participant

      I’m sorry for that the two pictures couldn’t be uploaded successfully.These two ones are new.

      Attachments:
      You must be logged in to view attached files.
  • #881

    periz
    Keymaster

    A possible explanation is a wrong timing of the image relative to the orbital data.
    Suggestion: plot the footprints of the full-frame, extract the full-frame Master and plot it.
    Based on the 2 pictures, you can make a rough estimation of the shift between the real geographic location of the image and the apparent one retrieved from the orbits.
    You can then proceed in 2 different ways:
    either you make use of the estimated shift to extract your area of interest (you have to shift the apparent Master area center)
    or you extract and coregister the full-frame, and later you extract a sub-dataset over the area of interest (from the site processing module).
    This misgeocoding can be later corrected through the GCP.

    check please if the software returned a message like “warning, the geographic center of the image retrieved from orbital data does not match …”.
    We can check whether we can in some ways improve/correct it or not.

    • #882

      cnu_lee
      Participant

      I don’t understand how to process specifically,but I found Matlab gave me a message”center latitude and longitude from MASTER SLC data differ from center of calculated footprint”. Is this message important for this problem?

  • #2466

    ABraun
    Participant

    Dear all, I use 16 Sentinel-1 SLC scenes (same track) for change detection analysis.
    I followed the tutorials I and II for data import. Master selection and coregistration worked well but I keep struggling with the GCPs as the scene is always shifted to a large degree (please see screenshot below).

    Do I miss anything that is not contained in the data import tutorials? How many GCPs are required?
    If you need further information, please just let me know.

    https://picload.org/view/dgapdiaa/gcps.png.html

  • #2470

    periz
    Keymaster

    1. check that you have a DEM correctly loaded. to check it, you can click “dem visualization” in the frame “preliminary geocoding” in “site processing”. if you get a green flat picture, probably it failed downloading SRTM data (you should have got messages about that). In such a case delete the corresponding SRTM files (SRTM directory in the folder containing the project directory) and plot the DEM again. The sw will try to re-download the SRTM files.
    2. assuming the DEM is ok, go to “GCP selection” and reset GCP and orbits. Then click on “keep current orbits”. You can then check again your geocoded reflectivity map. in normal conditions, you should solve your issue.
    3. if on the contrary, you still have problems, then there is something wrong. Go then to “extended data export” in “results exporting” and choose to orthorectify the reflectivity map. If this solves the problem please let us know, we’d be happy to check ur data, it could be a bug.

    If points 2 gives you a solution, and if you were getting problems by manually imputing GCP data, a possible explanation is that you confused lines and samples…

  • #2471

    ABraun
    Participant

    the dem was all fine, but the second point helped to load it at a correct position, thank you!
    I however noticed that the subset in the “SLC Data Import” wasn’t applied correctly so that confused me a lot 😀 Was probably my fault, I’ll run it again with the correct subset.

  • #2472

    qingli
    Participant

    Dear dan,
    i met nearly the same problem.
    I choose a GCP and set it.
    i have checked the DEM file via visualization and it seems fine.
    the location of GCP is correct in reflectivity map and in google earth.
    However, the geocoding reflectivity map is offset far from the right location.
    Please have a look with the attached files.
    Thanks,
    Regards,
    Qingli

  • #2473

    periz
    Keymaster

    You are on the mountains and you are geocoding your data, not orthorectifying it.
    You should orthorectify it instead.

You must be logged in to reply to this topic.