cumulative displacement isn't accuarte
 This topic has 6 replies, 4 voices, and was last updated 4 weeks, 1 day ago by chronomanz.

AuthorPosts


June 12, 2015 at 3:40 am #991cnu_leeParticipant
Hi, Professor Daniele,
The cumulative displacement value in DBF file of result are relative to the reference point. But I don’t know the cumulative displacement of reference point.So the value isn’t accurate?
How to understand the value of cumulative displacement? Whether I should plus or minus a value to obtain the true displcement value?
Thank you.

June 12, 2015 at 7:06 am #993perizKeymaster
the title of this post is pretty misleading.
insar is relative, but it can still be accurate.
so, deformation trend, time series, cumulative displacement and any other insar product are RELATIVE to a reference point.
unless you know whether the ref. point is stable, your measurements will always be RELATIVE to it.
the true displacement value does not exist (better: it exists, but you do not know it). in any case, you need to know what is the displacement relative to.
June 21, 2015 at 1:12 pm #1011seppisantiagoParticipant
Hello Daniele,
I’ve tried to generate Cumulative Displacement files covering different periods of time, entering different dates in the Cumulative Displacement tool. However, when exporting to KML/KMZ I think I’m always getting the same file,as the initially generated for the complete temporal length. covered by mi images. Looking at the logfile it does not show any error when saving the CumDisp.mat file.
Another question related to this, when exporting time series the displacement values (corresponding to each acquisition date )stored in the .csv file are calculated using the same polynomial coefficients in the Cumulative displacement tool?
Thanks,
S


June 21, 2015 at 11:55 pm #1013perizKeymaster
yes, that is something to be tided up.
Initially, the Cum Disp tool was generated to calculate it based on the estimated coefficients of a polynomial.
Later on, I invented the smart tool, which is a nonparametric time series analysis.
So, the Cum Disp tool is consistent with the rest of the results (and it calculates the proper things) only if you estimated a polynomial.
However, I strongly suggest you to become familiar with the smart option, which is much more powerful.
In such case, the Cum Disp is only calculated as a difference between the last and the first samples.
We will generalize it and we will tidy up this matter when we will renew the time series exporting tool. 
November 16, 2019 at 3:07 am #4813chronomanzParticipant
hello prof,
i have almost similar question. i used Smart5 with Small Temporal Baseline image graph on 6 images (which is less than u suggested for its robustness; in message #3231) to detect subsidence
obtained results from 2 weeks baselines are:
weeks: 0 2 4 6 8 12
a) individual displacement for 6 dates are: 0mm, 1.4mm, 0.5mm, 8.9mm, 11.7mm, 12.0mm (which total is 34.5mm)
b) reference point displacement for 6 dates are: 0mm, 0.2mm, 1.7mm, 1.7mm, 0.5mm, 0.23mm (which total is 0.07mm)
c) while cumulative displacement is: 11.2mm (but diff[lastfirst] = 12(1.4) = 10.6mm)1. based on small temp baseline graph – SB
* does each displacement measured in relative to consecutive image (date) since i used SB?
so, displacement of 0.5mm week 4 is in relative to week 2. so, total in week 4 is 0+[1.4]+[0.5]=1.9. am i correct?
* or in relative to the Ref Point for between 2 dates?
but, how to explain?2. how do i get explain the relationship between individual displacement (a) and cum. disp (c)?
* total of (a) differs from total of (c)Attachments:
You must be logged in to view attached files. 
November 16, 2019 at 5:52 am #4816perizKeymaster
In the attachment you see the difference between calculating the cum disp from the data or from the model (the drawing is made by hands, but the concept is that the smart filter results in a smoothing of the data).
We have chosen to calculate the cumulative displacement from the model. It’s a choice, that does not mean it’s the truth.
If you want to calculate the displacement from the data it’s easy. If the time series starts from zero, it’s the last value of the displacement. Otherwise, for more options, simply export the csv file that contains all numbers you want. From there, you can extract anything you like…
BestAttachments:
You must be logged in to view attached files. 
November 16, 2019 at 12:41 pm #4824chronomanzParticipant
hi prof,
understood. tqvm


AuthorPosts
 You must be logged in to reply to this topic.