yes my co-registration was successful and yes i can visually identify the movement. But i tried also on 2 different windows size but same type result i am getting red image
When i tried using SLC product offset tracking was not working giving error that cannot find amplitude& Intensity band.
Guys, please advice.
I am trying to get a subsidence information within a basin area. Given my Basin study area with Wrapped-, Unwrapped-phase and Displacement (uncalibrated) of S1 Multi-looked IFgram from Dec2014/Dec2018 pair.
Mask of coherence gt 0.2 during filtering. How can i tell that i am getting the right output? What i meant is, should we get a high displacement where there are lots of fringes on an area (right above the basin boundary)?
the patterns in your interferogram and the ramp in your displacement indicates that your result is not reliable at all. It is just randomly aggregated phase noise. The fringes should at least be connected over the area you want to observe.
This is most probably because you use images from 2014 and 2018, which is a way too long time for differential InSAR, because your signal is completely decorrelated.
These things are well discussed in here, you might have a look: Some explanations about concepts of fringes of interferogram and coherence
High coherence does not indicate high displacement, it simply is a measure on where your signal can be used for displacement mapping. But this only works if the area of interest has connected areas of higher coherence because otherwise unwrapping errors occur and produce a trend as seen in your result.
Is your study area covered by vegetation? This is one of the major reasons for decorrelation. I would suggest using image pairs with smaller temporal baselines but if there is much vegetation, you cannot do much about it at all.
SW: large oil palm area + paddy fields + swampy area
Centre: urban area
NE: forested area of the main range mountains
I will try smaller baselines: 12/6/3 months or shorter
Thinking of PSI method but can my 24gb ram, 24.ghz pc handle it? i am also very new to this. Trying to finish my masters prog due Feb 2020. Still running thru the posts in the forum. Lots of info. Thanks
too much forest for traditional DInSAR, I’m afraid. You supervisor should have mentioned this in my opinion.
24 GB RAM is not bad, PS InSAR should work if you prepare a subset using SNAP2StaMPS
1 Like
Dear all,
what do you think about the idea to use DInSar to detect vertical displacement of gravel-bed rivers due to flood events?
The question is deliberately generic so you can express your own view 
!
Thanks
Daniele
Dear all,
I’ve followed the entire process to get the displacement recorded between two images (17th August - 10th September) in the area of Nord-East of Italy, and I’ve obtained the same type of results gained from @chronomanz.
I completely agree with @ABraun when he says that the presence of forest compromises the results. So, how can I evaluate reliable displacement only of the gravel bed of the river surrounded by the black line?
Thank you in advance for the suggestions
You can try to mask out all forest areas and see what remains. Is the gravel bed in the forest aa well?
I’ll try. At wich step you suggest to mask out the forest (12, 13 … 19)?
In your opinion is better to mask the forest using the blue polygon or analyze only the riverbed inside the red polygon masking out all the rest?
As you can see it isn’t covered by forest
yes, if the coherence outside the forests is acceptable, masking the forest would be the better choice, because it leaves you more area to be unrwapped. It is unlikely that unwrapping only the riverbed will work.
Can you please share a screenshot of your interferogram?
this is the inferogram obtained from the Inferogram Formation tool.
As you can see, there are not present fringes maybe because the land surface has not changed during the time step between the two acquisitions 17 Aug - 10 Sep.
Only the riverbed faced probably some variation due to a flood event. That is my attempt to assess the riverbed height variations trough SAR data.
if the surface did not change between this period, the interferogram should rather look like this:
Personally, I see no potential for traditional DInSAR here, because the phase is almost entirely decorrelated. Even with masked forest areas, there is no real pattern to unrwap and your unwrapped phase looks accordingly random.
Have you tested if multi-looking increases the interferogram quality?
I have processed 72 images of Sentinel-1 in pairs (master + slave). Then I created a displacement map in SNAP software, can I analyze this process as a subsidence time series? Although SNAP software does not have the ability to do time series?
a suggestion on how to combine the deformations of several image pairs is made in this tutorial:
https://eo-college.org/resources/insar_deformation/
But note that this was created using NEST and some of the described steps (unwrapping, phase to deformation) are outdated and now easier to achieve.
A more recent example is given at the end of this tutorial: Sentinel-1 TOPS interferometry
Besides that, we hope to include pyrate support in SNAP in the near future. It especially targets the multi temporal analysis of DInSAR pairs.
this is announced but not yet made available to the users.
The G-TEP in general provides services to registered users. There are different user levels, most can just view the results generated by others, higher level users can start the services themselves.
Please have a look at the documentation: https://geohazards-tep.eu/
So how can I use this chat service and process with it?
I’m not going to read the documentation to you.
We do our best to help each other in this forum, but if things are well documented like this portal, you are more than qualified to find out yourself.
You have thankfully adviced me before to subset the area before unwrapping especially when there are huge lakes in the middle of the bursts. According to zahra0729, the displacement values change depending on the area of the image. which begs a question. Are unwrapping errors cumulative?
So, in addition to masking out low coherence values, is it preferable to subset the study area before unwrapping because the unwrapping errors may potentially multiply?
I mean, let’s assume the study area is contained in nearly 4 bursts, should it be divided into smaller sections and each of which will undergo interferogram formation and phase unwrapping, then the results are merged?
If such strategy makes sense, would there be discontinuities along the regions where the unwrapped/displacement images merge?
I don’t think so, because the merging of different sub-swaths potentially leaves unwanted marks along the edges. If a sub-swath is noisy its better to fully leave it out from the beginning.