Ascending Vs Descending

Oded · February 15, 2021, 4:48pm

Hello,

I managed to activate the snap2stamps chain from start to end, and finish StaMPS inside Matlab.
I did it successfully for ascending data.
I’m trying to do the same exact chain for descending data of the same area and time, but coreg_ifg_topsar.py gives me error -

Error: [NodeId: Subset(2)] Operator ‘SubsetOp’: Could not parse geometry.
– org.jblas INFO Deleting /tmp/jblas7283355022306156205/libgfortran-4.so
– org.jblas INFO Deleting /tmp/jblas7283355022306156205/libjblas_arch_flavor.so
– org.jblas INFO Deleting /tmp/jblas7283355022306156205/libjblas.so
– org.jblas INFO Deleting /tmp/jblas7283355022306156205/libquadmath-0.so
– org.jblas INFO Deleting /tmp/jblas7283355022306156205

Is there any reason for a problem with descending data?

Thank you,
Oded

marpet · February 15, 2021, 5:07pm

Not sure why the geometry can’t be parsed.
Can you provide the geometry string?

ABraun · February 15, 2021, 6:13pm

please also share the config file which defines the AOI

Oded · February 15, 2021, 7:33pm

So at the beggining I defined AOI and got the error.
Then I removed the AOI definition from the conf file and tried the coreg_ifg part again, the error remained.
I thought maybe the former step (splitting) kept that AOI information somewhere, so I started the whole process from the beginning. Still, the error remains.

When I run it with the AOI of the ascending data, it works.

Oded

Oded · February 15, 2021, 7:35pm

Hi,
Did you mean the AOI parameters?

################

AOI BBOX DEFINITION

LONMIN=34.6044
LATMIN=31.747633
LONMAX=34.700742
LATMAX=31.841665
################

ABraun · February 16, 2021, 6:15am

Which python version are you using?

marpet · February 16, 2021, 7:18am

Ah! You are defining the AOI in STaMPS and STaMPS converts it into a geometry for the SubsetOp.
So, this is either a problem in STaMPS or in your configuration.
I’ll move this from the SNAP category to the STaMPS category.

Oded · February 16, 2021, 8:04am

Actually,
I’m in the snap2stamps process, so it might considered SNAP now because i’m using snap’s kernel.

Oded · February 16, 2021, 8:05am

It’s Python-2.7

marpet · February 16, 2021, 8:29am

Please show us your code and how you feed the AOI into the subset operator.
As the error message indicates there must be something wrong. Which version of SNAP are you using?

Oded · February 16, 2021, 8:44am

Hi @marpet,
I’m using SNAP 8.0.3, i.e. the updated one.
As I wrote, I completely removed any AOI definitions, started the whole process from the beginning, and still got that error message.

Oded

ABraun · February 16, 2021, 8:50am

you can check which xml was created by snap2stamps during the processing in the graphs folder of the processing directory (not the snap2stamps directory). There is a graph file belonging to the coregistration process with the current datasets as inputs. Maybe checking the definition of the subset in this file gives an indication about the error source.

Oded · February 16, 2021, 8:58am

@ABraun, you were right.
coreg_ifg2run.xml:

<operator>Subset</operator>
<sources>
  <sourceProduct refid="TopoPhaseRemoval"/>
</sources>
<parameters class="com.bc.ceres.binding.dom.XppDomElement">
  <sourceBands/>
  <region>0,0,24608,2844</region>
  <geoRegion>POLYGON (( , , , , ))</geoRegion>
  <subSamplingX>1</subSamplingX>
  <subSamplingY>1</subSamplingY>
  <fullSwath>false</fullSwath>
  <tiePointGridNames/>
  <copyMetadata>true</copyMetadata>
</parameters>

While the templet coreg_ifg_computation_subset.xml is:

<operator>Subset</operator>
<sources>
  <sourceProduct refid="TopoPhaseRemoval"/>
</sources>
<parameters class="com.bc.ceres.binding.dom.XppDomElement">
  <sourceBands/>
  <region>0,0,24608,2844</region>
  <geoRegion>POLYGON</geoRegion>
  <subSamplingX>1</subSamplingX>
  <subSamplingY>1</subSamplingY>
  <fullSwath>false</fullSwath>
  <tiePointGridNames/>
  <copyMetadata>true</copyMetadata>
</parameters>

For some reason, the Polygon got empty enteries…
I believe that shoud stay ‘Polygon’ and not ‘Polygon (( , , , , ))’.

ABraun · February 16, 2021, 9:00am

I assume this is how it looks when you have not entered the AOI coordinates in the config file, right?

What happens when you add them back in and run it again?

Oded · February 18, 2021, 9:52am

Hello @ABraun,

It took me some time, but my conclustion is that this piece of data suffers from bad luck…
The SA1 descending path borders fall exactly at the city I’m interested in (marked with black X), and cut it to two frames from N to S, and two frames from W to E:
Screenshot from 2021-02-15 15-32-07

So, that’s right - there is an area of overlap, and the upper left rectangle should include that city, but look what happen after spliting and applying orbit correction:

The orbit correction tilts the rectangle and misses the city!!! (where the red arrow points to).
So I call that bad luck, and apparently that’s why I had problems with the subset coordinates. Ha ha.

The Ascending frame doesn’t suffer from this problem and therefore there were no problem processing it:

What would you say about that?

Oded

p.s.
If I understant correctly, mt_prep_snap + stamps(1,1) make the trasnformation from Range-Azimuth grid of sentinel to Lan/Lot grid.
I need the elavation data as well, and I see there is a hgt.mat file at the PATCH folder. Is that’s what I need?
I want to present the result on 3D grid.

p.s. 2
The abbreviation ‘p.s.’ is funny in this context.

ABraun · February 18, 2021, 12:26pm

have you checked the radar image visually if this is really the case? Sometimes the footprints in the overview are not very precise.

Oded · February 18, 2021, 2:04pm

By ‘radar image’ do you mean to the quicklook?

ABraun · February 18, 2021, 2:15pm

no the actual intensity image

Oded · February 18, 2021, 3:05pm

OK, I’ll check it.
What about the 3D grid data?
Is there a way to export it from SNAP in order to present the PS on a 3D grid?

ABraun · February 19, 2021, 9:53am

can you please specify what you meen with 3D grid data?