sarproz-group

Accuracy of my data

Viewing 3 reply threads
  • Author
    Posts
    • #7585
      paulinanino
      Participant

        I have 2 questions regarding the accuracy of my data.

        1. I understand that to obtain the deformation precision in a particular trajectory I must occupy the “stdev” which is the standard deviation.
        However, when I obtain the vertical deformation using two trajectories (ascending and descending) in the multi-sensor analysis it only gives me the coherence. How could I get the precision in this case?

        2. I would also appreciate if you could tell me how sarproz calculates the “stdev” parameter. I am especially interested in how to know what generates variation in my data, is there any document that explains it?

        Thank you very much

      • #7587
        periz
        Keymaster

          Hi,
          first of all, to be able to obtain accuracy, we would need the truth, which usually we don’t have. So, it would be better to talk about precision.
          In any case, the standard deviation is calculated from the phase residuals (observed interferometric phase after removal of the model you choose for the processing). As a consequence, you can understand that it’s a way measure the match between a model and the observations. Sometimes you might have a discrepancy caused by noise, sometimes the mismatch comes from un-modelled deformation. To try to understand more, you need to observe your data from multiple perspectives (like space and time).
          The coherence is an other way to measure the match, in the complex domain. This is the parameter we normally use to quantify the goodness of an insar time series estimation. We introduced the standard deviation just to simplify the interpretation to those not used to InSAR processing.
          To understand deeper this topics you might consider taking the course EO59 offers twice a year. There will be a session in about 1 month from now.
          At the moment we did not include the projection of standard deviation from ascending/descending orbits to horizontal/vertical directions in the multi’sensor module. If needed, you can obtain it from the single stdevs using the infos about incidence angles.
          best

        • #7674
          paulinanino
          Participant

            Thank you very much for your response.
            So I understand that in order to see the precision of my results I should check which are the components that constitute the coherence?
            – Is there any formula or document you could tell me about how it is calculated or exactly what terms affect it?
            I understand that my results can be affected by distortion, baseline, surface area, number of images, etc. However, I would not be able to discern which of these things the software fixes and which it does not, to understand how accurate my deformations are.
            – The things that I could say don’t affect my deformation results are topography and atmosphere because of the PSI technique, and the APS processing. Is this correct?

            Thank you,

          • #7675
            periz
            Keymaster

              Hi Paulina,
              to answer your questions we usually take 5 days of classes: those the core topics of multi-temporal insar…
              may I suggest you to take the online course? or alternatively we can organize a course in person in your country…
              best

          Viewing 3 reply threads
          • You must be logged in to reply to this topic.