Open georgeboldeanu opened 1 month ago
It seems you have filtered out some low coherence areas like water surfaces, but not completely, as there are still invalid borders around them.
@AlexeyPechnikov I forgot to mention that I am referring to the rectangular gaps. Sorry for the omission. Also, I mention that all the data that enters in the computation of the displacement doesn't have those gaps, so I guess is from the processing of the displacement.
Trend correction is per-chunk in the spatial domain, so if you do not have enough pixels in the chunks, the correction is not possible for some chunks (prezented as rectangular gaps on the plots). It’s better to use a single chunk for this correction, and you can filter out low-correlation values for faster processing on large grids.
@AlexeyPechnikov But those areas that have gaps are in an urban area with high coherence, so no filtering is done. How is possible that there are not sufficient pixels in an area like this? If you look at the sample image, in the upper part is understandable, but in the middle part I don't understand the behavior.
You need to increase the chunk size and compare the results.
Hello! I do a sbas analysis over some area of Philippines. After I computed the displacement with the correction for trend and 2 turbulance phases I noticed some gaps in the resulted data (rectangular gaps, circled in red). It's this a known issue? I didn't get any error message or something like that.![Screenshot 2024-07-12 104803](https://github.com/user-attachments/assets/59028314-58a5-40d8-9b69-156e7a8576d3)
For a smaller area using SBAS and PS combination the results were clear without any gaps. It can be something resource related? The resulted displacement stack is not that big, with around 17GB.![image](https://github.com/AlexeyPechnikov/pygmtsar/assets/34105846/6488596e-4203-4178-bce2-a536e2576203)
Thank you!