[bridgetwang]

Now it is totally clear to me. I appreciated all of your answers, thank you!

[bridgetwang]

Thank you so much for the prompt reply. Now I understand where the problem is.

So does it also mean that, everytime if I change the graph e.g. from STAR to Full graph or MST, I should re-process the ASI for the current graph?

[bridgetwang]

Dear Prof. Daniele，

I had asked in the forum that I coregistered and processed a long time series dataset including 280 images, and then I changed the “Dataset.txt” to process different subsets individually in one project. Here is the question link:Big dataset processing

The 3 subset is 2010-2012,2012-2014,2014-2016, and the whole dataset is from 2010-2016. I use the same threshold of APS(0.75) and MISP(0.7) for the whole dataset as well as for different subsets. After I process in this way, I find the number of PS points is not what I expected. The PS points I get in APS and in MISP for subsets are excatly the same number as I get from the whole dataset: for the whole 280 images from 2010-2016, I get 30694 PSC in APS and 334472 PS in MISP. The number I get is exactly the same for subset 2010-2012, subset 2012-2014. subset 2014-2016.

This is not what I expected. As the urban area experience fast building construction, the extracted PS points of subset should be more than the that of the whole dataset which is affected by temporal decorrelation/changes.

It is true because I then build a new project, choose the same area and coregister the 2010-2012 dataset and process PSI, I do get more PS with the same threshold.

I could get more PS points in an individual project as I expected, while not in the whole dataset project. What should I do except from changing the Dataset.txt to extract more PS points for different subset just like in an individual project?

[bridgetwang]

Thank you very much for your answer as quick as ever! I’ll follow the tutorial to do a test.
Sure, I’d like to take the course when our financial support could afford it. 🙂

[bridgetwang]

Thank you so much for the rapid reply!

As for “estimate as well as possible the non linear model”, I wonder what’s the difference between using a non-linear model built by myself(e.g. include deformation induced by temperature or water level) and using the “smart” non-linear model?

Additionally, I know that I can use higher threshold for point selection in APS module, and then complete the APS processing. But I don’t quite understand how to implement the 2 steps estimation in APS module. Could you please explain further on how to implement step1 and step2 in the software?

Best regards

[bridgetwang]

finally the linear model with thermal expansion is added.

By the way, the temporal coherence with index “_1” is the coherence before removing APS.

Looking forward to your reply. Thanks very much.

Best regards

Attachments:

You must be logged in to view attached files.

[bridgetwang]

the result of linear model is added here

Attachments:

You must be logged in to view attached files.

[bridgetwang]

Thank you very much for your rapid and detailed answers!

1.

Cons: if your area is affected by temporal decorrelation/changes, keeping long temporal baselines will decrease the overall coherence.

For the cons you mentioned, I wonder if I individually corigester let’s say 50 images ; On the other hand, I select the same 50 images from the 280 coregisterd imgs. So the 1st and 2nd 50 images are the same, just they are coregistered to different master. And after same PSI processing, is there going to be much different result between the 1st 50 imgs and 2nd 50 imgs?

I’d like to divide the 280 images by 2 years, meaning the temporal baseline of every subdataset is within 2 years. I suppose in this case the long temporal baselines problem won’t exist, right?

2.
As for the error I come across, I got this error after I delete some imgs in Dataset.txt. When I load the prohect again, I got this error. I’m sure I didn’t change the SLC folder, and already complete the coregistration. How can I fix it ?

3.
According to your answer, in order to keep some image connections and discard those exceeding 600m baseline threshold, I should just set the theshold via the dataset selcetion module to get the connections I want, and ignore the matlab warning? In that case, when I come to APS, it will be e.g. “Images Nr. 58 Connection Nr. 39”, is this OK for next processing ?

Best regards

[bridgetwang]

Please find the attached error files, thanks a lot !

Attachments:

You must be logged in to view attached files.

[bridgetwang]

Thank you very much! I’ll try as you recommended 🙂

[bridgetwang]

Thank you for your rapid answer!

[bridgetwang]

Hi,

it works after I delete the imgs in dataset.txt.

Thank you so much !

Best regards,
Zoe

[bridgetwang]

Thank you very much !

[bridgetwang]

Thank you for your kind reply.

With respect to question 3, the “previous points” you pointed to are points selected by e.g. Amplitude dispersion index in APS?

Also, I have the confusion that, since QPS make use of DS points, shall I comprehend like this: The points selected with star graph are PS points, and points selected by other kind of graphs(MST,Full graph…) are PS+DS points?

[bridgetwang]

Dear Prof. Daniel,
After I read your answers, I tried some experiments with GAR to comprehend better in the past days. I got some more questions here during the experiments:

1. Does GAR always behave better than MISP? And do you recommend to use GAR in most cases?

2.You mentioned GAR could detect strong movements which exceeds wavelength, I want to know what the limit amount of moving GAR could detect? why could it detect movements larger than wavelength?

3. Are the “local reference points” in GAR still PS points, or are they other points like DS points?

4. With respect to your suggestions, I’d like to ask has date of the course in Italy been set yet ?

Best regards,
Zoe

[Author]

Posts