Spatial ramp for the distribution of PS points

This topic contains 1 reply, has 2 voices, and was last updated by  periz 1 year, 3 months ago.

  • Author
  • #3088



    When i do PS time series analysis, i got the result as shown in the first figure ‘PS overview’. This figure shows all the PS points with their color rendered by the deformation rate. There is a very obvious spatial ramp pattern that northeast corner shows stronger subsiding while the southwest corner shows much less subsidence trend. It looks like orbit error or tropospheric error, which resulting in the ramp.

    The next three figures i posted shows the statistics of velocity, residual height and height of the study area. These figures were resulted from choice of reference point. All of the histogram peaked at/around 0, so i believe the reference point is relatively stable.

    I processed the in such a procedure:

    First – SLC data import module, choose study area, master and slave co-registration. (after coregistration, i manually check the reflectivity map, and it is well aligned in google earth. The whole study area is 20km radius.)

    Second, i applied the APS module. In the APS module, I input the estimated linear trend and height as close to what is shown in the figure (as you emphasized in the manual).

    Third i applied the sparse point processing module, also input the estimated linear trend and height according to the figures.

    At last, i output the results in Post Analysis- time series module.

    I will appreciate it if you can suggest on how to solve this spatial ramp pattern problem.


    You must be logged in to view attached files.
  • #3093


    Spatial ramps can arise from the spatial graph integration. Such operation is weighted with the coherence of the connections. Check the coherence and check the Non-linear parameters used to set the range of reliable coherence values. Choosing carefully those parameter should help the stabilization of the integration. If that does not help, it means that the topology of the graph is not good enough (e.g. too few points and/or too long spatial connections). Build a better graph then.
    an other option (should be the last chance if you have no ways to improve the graph inversion) is the manual removal of ramps. ramps can be removed from interferograms in case of orbital errors, from the graph results (but this implies using the inverted parameters APS which should be avoided) or at the very end of the processing, when exporting sparse points.

You must be logged in to reply to this topic.