sarproz-group

Loading reference point in the APS module using command line mode

Viewing 1 reply thread
  • Author
    Posts
    • #5362
      dominika-malinowska
      Participant

      Dear Sarproz Team,

      I came across an issue with the command line mode.

      I first analysed my AOI manually. In the APS module, I loaded RefPoint_1134_65.mat file, which I created earlier, as the reference point. The message in MATLAB confirmed that the reference was loaded correctly. Then, I went to the MISP module, where again I loaded the same mat file as a reference. I finished by creating kml in the Time-series module.

      Then, I wanted to repeat the same analysis using the command-line mode. Thus, I copied the InputPar file, renamed it, and changed name of the export and import TS file to make sure it imports the correct TS (<TSExportFileName> AutoTS_06 </TSExportFileName> and <TSFileName> AutoTS_06 </TSFileName>). Then I run the automatic process in Matlab using command: fig = main('D:\Sarproz_Data\Sub-dataset_10by10km_small_displ\InputParFile_auto06.txt') . When kml was created, I noticed that all new points are shifted in comparison to the first ones. The easiest way to notice that is by comparing reference points – even though the reference points in both kml have the same line and sample numbers, they have different location on the map (one of them is shifted alongside the satellite line of sight). You can see that on the screenshot below.
      I analysed messages from the MATLAB command window, and I think the problem is created in the APS module. It seems that, when the command line mode is used, the first point from the automatically generated list of points is selected as a reference, instead of loading the point I wanted. You can see it in the messages from the MATLAB command window below. Point 1589,96 is used instead of 1134,65. Later on in the MISP module, the correct reference point is imported from the .mat file.

      MATLAB command window messages:

      calculating nodes coherence 
       <strong>Ref Point Nr 1 out of 25 </strong>
       reading file D:\Sarproz_Data\Sub-dataset_10by10km_small_displ\RESULTS\MATLAB\InSarParam 
       sel_par: neither file D:\Sarproz_Data\Sub-dataset_10by10km_small_displ\RESULTS\CoerSpace nor D:\Sarproz_Data\Sub-dataset_10by10km_small_displ\RESULTS\CoerSpace.mat found 
       files D:\Sarproz_Data\Sub-dataset_10by10km_small_displ\RESULTS\CoerSpace and D:\Sarproz_Data\Sub-dataset_10by10km_small_displ\RESULTS\CoerSpace.mat not found 
       <strong>selected reference S: 1589 L: 96, Temp Coher: 0.99, Sp Coher:</strong>  
       Regulating the middle value 
       applying non-linear weights on the connections coherence: 0.57 0.76 0.96 
       processing APS through RESIDUALS INVERSION 
       INVERTED RESIDUALS APS processing started

      To sum up, I believe the problem with the shift of points is caused by wrong selection of the reference point in the APS module. This issue does not appear when I manually select all the options using Sarproz the usual way. Therefore, my question is: how can I force Sarproz when using the command line mode to load the reference point I want in the APS module?

      I attached the InputPar txt file I used. Please let me know if you need any more details.

      Thanks,
      Dominika

      Attachments:
      You must be logged in to view attached files.
    • #5367
      periz
      Keymaster

      Hi Dominika,
      First of all, at the moment we did not think about letting the user pass a manual reference point in the APS module in the command line mode. The sw will load the RefPoint.mat file (the default way it is usually saved), if it is working in update mode. Otherwise, it’s the sw to choose the default ref point. The reason is because in 95% of cases the sw does a better job than the user. Also, if it’s the first time you analyze a given dataset, you cannot know which point can act as a good reference. In fact, here it’s more important to equalize statistics rather than interpreting your data. For that, you can change the reference point later on, e.g. when you export a csv file.
      Note that the shift in geocoding, if it is really due to the reference point, it means that something is wrong with your ref point selection. Otherwise, the sw would adjust the correction.
      About the 5% of cases when the user choice is better than the default one, you would need to run the sw with graphic interface for that…
      best

Viewing 1 reply thread
  • You must be logged in to reply to this topic.