I am trying to do H-A dual pol decomposition of sentinel-1 dual pol (VV/VH) SAR data. I am using server with 128 GB of RAM and multiple cores but my processing is taking a lot of time. Is anything wrong with my processing step or can something be done to speed up the process in SNAP.
I got stuck here and any suggestion would be of great help.
I think the cause of problem is terrain correction, try up to remove terrain correction, and run it again,
Similar issue I had here,
Source of the post
I guess you are right. But any alternative for terrain correction?
Have you checked if SNAP is able to use all of your 128 GB? It is described here: Gpt performance and here: S1 data calibration error
I would furthermore suggest to split the graph into two parts. Graphs which are too long have problems passing the intermediate products to the next operator. That releases used memory and is more effective in terms of computing time: Coregistrating more than two Sentinel-1 SLC IW products and Problems obtaining an interferogram from two product sets
Thank you for suggestions. SNAP is using nearly about 100 GB. I splitter the graph into two : First graph upto multi looking and second graph continuing from speckle filter to terrain correction. I saved first graph’s output as .tif file. But in the second graph terrain correction operator need incidence angle from tie-grid points which is not present in the output of first graph.
- Polarimetric speckle filter
- Polarimetric decomposition (H-A dual pol)
- Terrain Correction (problem occurring here)
You should use BEAM-Dimap during your processing and export to other formats in the end.
Besides the choice for tif after the first graph, I think that is a good solution.
You would save even more memory if you put the subset and multi-looking in the beginning. You can still perform complex calibration afterwards.
BEAM-Dimap format is good solution.
My points are:
- When to use complex calibration
- No able to perform terrain correction. output is zero data (blank image). When terrain correction is not performed then I am getting the Entropy, alpha bands.
Graph 1 works fine
if you want to perform polarimetric analyses (as in your case) complex calibration is required.
Try SRTM 1 Sec (AutoDownload) instead of 3Sec because SNAP has problems lately, resulting in black images.
I tried SRTM 1 sec and it works Thanks a ton for this. Regarding calibration, at which point in graph I should apply this??
SNAP supports calibration after multi-looking (unless you check “output intensity”) so you can make a subset first and multi-look to reduce the data size to be processed.
@ABraun The problem with the SRTM 3Sec has been reported somewhere.
Thank you very much
Hey thank you very much for this information, it helped a lot. I divided the graph into two parts and now it is working fine and faster also.
These are the two graphs:
- Complex calibration
- Write (as BEAM-DIMAP)
- Read (output of graph 1)
- Polarimetric speckle filtering (refined LEE 7x7)
- Polarimetric Decomposition (H-A dual pol)
- Terrain Correction (DEM : SRTM 1 Sec)
- Write (as tif)
Thank you all for suggestions and tips
What actually the problem is? How these two DEMs are different?
SRTM 3 Sec is 90 meters and has bigger tiles https://dwtkns.com/srtm/
SRTM 1 Sec is 30 meters and has smaller tiles https://dwtkns.com/srtm30m/
the first is somehow differently accessed by SNAP and lately the download is not successful in many cases.
Hey thanks for the info. It recalled helped a lot. I tried the graph and it is working fine But can you help me explaining the difference between:
Product -> Deburst -> Output
Product -> split(IW1/IW2/IW3) -> Deburst -> Merge -> Output
i.e. splitting and debursting individual sub swath and then merge to combine them vs Just debursting the product. The output of both these are different.
I tried H-A decomposition over a subset and the values of H-A are different.
is the same color range (min max as defined in the Color Manipulation tab) applied to both images?
Color range is different in both case. The above two steps: should they be producing the same output??