about time series

This topic contains 1 reply, has 2 voices, and was last updated by  periz 1 year, 7 months ago.

  • Author
  • #2461


    I got maybe a stupid question but I got suddenly unsure.
    So when in Geocoding (Small processing) I click on “Data series” I get the time series of the point but, please correct me:
    -if I do not tick anything I get the residues phase, which is the phase minus the topographic and the movement component
    -if I tick “mov” I add the movement component (a linear model), which is the phase minus only the topographic component
    is it correct?
    So if I want to see the actual deformation trend along time of a PS point then I choose to tick the box, but if I want to compare the time series with external factors like water level (is a dam) then I should look at the phase residues. Is it the right thinking? And also, is the concept of linear and non-linear deformation connected to this? Or is it something different?
    I know that there are papers and the manual talking about this but if you can very kindly explain the concept and tell me if my reasoning is wrong would be great.


  • #2464


    Hi Cecilia,
    yes, on the manual you find some explanation, here in particular
    what you have written is correct if no correlation with temperature was estimated. If on the contrary you did remove a temperature-correlated trend, then you have to tick on the corresponding box to see it, otherwise it is removed from the displayed data.
    Yes, it is correct, if you want to look at the correlation of residues with other factors, then better to look at the residues alone.
    And yes, this can work also for non-linear components.

    Tricks in Sarproz:
    – to estimate non-linear movement (assuming it has a temporal correlation), input smart=5 in the corresponding time series processing module. This corresponds to a 5 samples time filter. it’s non-parametric movement estimation.
    – To estimate correlation with external factors as water level, you can import them as “aux data” and then estimate a correlation value in the “azimuth pos” box in time series processing parameters. Here a few notes

You must be logged in to reply to this topic.