It is quite normal. more pictures means that your points should remained stable in more pictures
so if I want more points, i should reduce the number of images? even that I can’t go less than 20 , otherwise all this process wont work
Thats a problem that I also faced. When you have vegetation in your area, it is much harder to find permanent scatters. If i were you, I would try different combination of data. I have read somewhere that even time of the acusition or humadity level affect the coherency between pictures. Chose pictures with similar situation.
I can’t choose more pictures anymore. I will have to stick to these images i already have. Indeed, the best I could do is to take pictures only during spring/summer/autumn to do this process.
Hello there. I have another question. as i have too many images for the same area, i’m splitting the images in three different stacks. My question is if it is right to put the first master image in the other two sacks in order to have the same PSI points overlapped on the map of the results of every stack? otherwise. since they are processed differently, I’ll have different points for the three stacks. I don’t know if I explained myself.
how many image sdo you have in total and how much are left for each of the three periods?
there are in total 56 images, over ten years. Si I decided to split them in 19 - 18 - 18 images. otherwise, i could take only 3 images for every year ( only of June July and August) for the total of 30 images over 10 years, this way I could use them without having to do different stacks.
in this case I would use three different masters at the middle of each period.
I did something similar here to avoid too much decorrelation caused by landcover changes: https://www.tandfonline.com/doi/full/10.1080/22797254.2020.1789507
Hello, I am doing a study over a bridge that has fallen in april of this year. But from the process of the PSI of the last 10 years, there is no prove of great movement of that area. How can be that possible? the process did not gave me the result I was aspecting
I don’t know the exact setting, but I could think of many reasons:
- the bridge did not show a subtle deformation prior to the incident, because it was a sudden collapse
- the spatial resolution does not target the PS which are responsible for the movement
- there is temporal decorrelation or strong filtering which either suppressed the PS with a distinct movement or smoothed it in a spatial way - what is the size of the atmospheric filter from the last step?
1.that is right, the bridge collapsed suddenly, but i had the interferograms after that months, so I think there would be some more movement after that day.
2. there are multiple target PS in the entire bridge, enough to cover the collapse area ( it was a long and important bridge)
3. how can I see what is the size of the atmospheric filter?
The standard is set by
That means each PS is filtered by all others with a temporal radius of one year.
You can reduce it to 150 days, for example.
But if you expect movement after the incident you should create two stacks, one before and one after this date (but not containing it).
If the incident is within one time series, you will drastically lose many PS because of temporal decorrelation.
I’m sorry, maybe I am confusing something. How is it possible I can’t see this kind of deformation? At the end, this is the scope of the work I am doing, being able to detect strong deformations to prevent disasters. But this way I can’t tell anymore …
When was the incident and what is the first and last date of the time series.
The bridge collapsed on 8 April 2020. The First date of images is on May 2011 untill July of 2020
For master selection, it should be based on normal baseline and snap already taking care of it. I am not sure playing with that would really result in better PS density or more accurate results.
based on your explanations, If I were in your place I would try simple Dinsar method. Cuz, Multiple time DinSAR is for gradual deformations.
I could try to use DinSAR and maybe confront the results. Do you know any tutorial for the DinSAR method?
just one point to add if your deformation is in the range 20-40 cm, the atmospheric effect would not dominate the results. but for smaller deformation, less than 10mm. noises, especially atmospheric in high humid area, would affect the reliability of your result which in that case, MT DinSAR can be helpful as it removes the atmospheric and other noises more effectively.
in my case, there is a total collapse in the bridge. we are not talking about deformation anymore. I want to prove that if we use this method, we are able to detect unusual deformation, and in this case we can prevent the collpase. @ABraun was right, i shouldn’t have included the images after the collapse. So i’m going to give a try at this other method ( I can’t even try to re-do the PSI-StaMPS without the collapse images, because the images are way to big, and it takes a whole week to finish just one process) and will confront the results.