Does anyone know how to modify the tolerance during the unwrapping processing in StaMPS? I would like to change the default values of pi in order to eliminate the steps in the ts of my previous post. I think that the matlab function is unwrap(x, tol)
Thank you for the link! I think that my issue is described here (step 6):
“The 2±π discontinuity of the extracted phase appears when an extreme value (−π/+π) is reached. The phase then jumps to the other end of the interval, even though physically the optical phase is continuous and relatively softly increasing or decreasing.”
I try to modify the parameters of step 6 and the “scn_time_win” as you suggest. Thanks again!
very good - can you please share the parameters which were required to change (and what you selected) so that the result has no more steps? Would be interested in how it looks now.
In particular, I used the command in matlab >>setparm(‘parm_name’, newvalue) for each parameter except “scla_deramp” before step 1. then I run steps 1 to 5 together (>>stamps(1,5)) and steps 6 to 8 together setting the “scal_deramp” to ‘y’ using >>stamps(6,8,‘y’).
I would like also to aplly the parameters suggested by Lazecky, 2011 to see the differences.
It will be also interesesting to modfiy each parameter individually to see how they influence the final product.
@Evandro: I have made the experience that setting parameters in StaMPS correctly is extremely challenging, at least for me. Simply using default values yields somewhat weird results in many cases, and applying parameter settings reported in Papers for similar process types is not guaranteed to yield the best results at first try, either.
I do agree that it would be interesting to modify each parameter individually to see how they influence the final product. @thho has made some efforts to assess the importance of various parameters on the results, which I basically reused in my documentation, but I guess this is just a very rough guideline, please feel free to report updates if you have new findings
Off-topic: I have made some changes to Thorsten’s visualizer, including the option for applying LOESS-filtering the time series, which is useful for seasonal time series. I should make a pull request at some point.
@m477h145 Yes I agree, although finding the correct setting is quite challenging. For example, the setting proposed by Hoser (2018) fits well with my situation while Lazecky’s does not resolve phase jumps. But Hoser’s itself works better only with 1 patch from mt_prep_snap and scla_deramp set to the default value of “n”.
I think that the correct setting is affected by both the geological situation (landslides, subsidence…) and the geomorphoological charachteristics (orientation and inclination of the slope…) of the AOI but also by the period considered (2, 3 or 5 years…). Do you agree?
Is there a maximum limit of amplitude dispersion index DA?
In the literature, I have verified that it is suggested that there are typical values (from 0.25 to 0.42) but I do not know if there is a maximum allowed limit since assuming that a value of DA=5.0 is applied, I think it is logical that a higher density of PS will be achieved.
here the two plots with almost the same vertical scale. I’m sorry for that “almost” but I don’t know how to change manually the x and y axis in StaMPS_Visualizer.
Seasonal effects is the first think I thought actually, but 20/40 mm of LOS displeacement in 12-24 days is hard to explain. I also cheked the precipitation rate (rain, snow…) during the jumps wondering about some kind of bulge but no correlations have emerged.
Hello everyone,
I have successfully performed steps till mt_prep_snap and now I am trying to run stamps(1,1) in MATLAB but i am encountering following error
Although fopen is being used to generate file id in readparams. Can somebody help me with that?
Note: I have performed mt_prep_snap in WSL and copied the folders made by stamps and matlab folder from stamps installation path in WSL to psi_insar folder in windows. And continue further processing in windows (i.e. I have matlab installed in windows. I don’t have license for matlab for linux)
Hello everyone! I am also facing a similar issue.
I used mt_load_initial_gamma provided by Mr. Delgado and compiled StaMPS as explained by Mr. ABraun with gcc7 to avoid segmentation core dump error. Also am using StaMPS version suggested by ABraun wich is called “StaMPS-master”.
I am using just 4 images just to test the method so maybe that is a problem.
This is a screenshot of the error in matlab.
Update: By deleting the virtual machine and installing Ubuntu 20.04 on a separate hard drive I was able to perform stamps(1,1), stamps(2,2) and stamps(3,3) but matlab exits everytime i run stamps(4,4). Just after this line:
Checking the ram usage with “top” shows that after free space get to 100Mib it exits.
I will need to try again with more RAM but is there another option? This time am processing 9 interferograms.
Dear @ABraun,
I have processed my time series into 5 separate parts(from 2016 to 2020). How many groundwater level maps do I need to draw now to validate their results? Do you need a piezometric map for each part?
Thanks a lot
there is no general answer on this, sorry. If you have some, use as many as you can. The more validation data you have, the more you can say about the accuracy of your results which makes them more valuable.