Sentinel 1 Back geocoding produces occasional No Data values

Could someone shine light upon why i am getting maybe 2% of pixels that come out of S-1 Back geocoding with values set to the No Data value?

I will say in advance that my DEM (Copernicus 30m) is complete and shows no pixel effects such as missing data values, and that I have experimented with different dem resampling methods and resampling types, and also that I have tried toggling, and untoggling Mask out areas with no elevation.

Further I have tried changing the no data value using Raster → Data-Conversion → Set No Data Value , and that didn’t seem to have an effect either.

none of these seem to have an effect, on the number of No-Data pixels after the Back propagation step.

I’m using snap 9.0 and no updates are found, so I assume I am up to date.

Many thanks for your help!

1 Like

@djagula can you have a look? Thank you

1 Like

Could you right click on the band name in SNAP, select “Properties” and uncheck “No-Data Value Used” in the Properties window? Give it a try to see if it helps

I am still testing this, but before I make any statements on it helping, could you tell me how I would introduce this step into a graph, so that I don’t need to manually edit files to make this change?

OK so i have delved into this a good bit further… i was wrong when I said that back geocoding was introducing the NaN / 0 values. @jun_lu @diana_harosa

Instead I would like to share the following observations.

My output interferogram contains pixels where both the I and Q complex numbers are set to 0, when In my opinion this should not be the case.
I use the following graph

the Interferogram step is defined as follows…

As previously stated the DEM is complete with no holes.

when I look at the raw data output at the Enhanced Spectral Density stage (write(2) in the graph) it looks like this (I have highlighted the master image i channel data that has 0 values in yellow, [ I find it odd that this data is 0, but i assume, its valid because i can trace it all the way back to the original data read step.]

After the Interferogram step, these pixels result in a 0 in both the complex channels of the output interferogram, shown here…

I believe this is not the correct result for their values, I can confirm all values are correct when the Master I component is not 0, but not when it is 0 i get a different value.

I’m not convinced that i am not making a mistake here, but i think its likely my not understanding something correctly, In all cases the data is set that “No Data Value Used” is set to unchecked… though i have tested and this does not make a difference.

could you enlighten me as to why this is happening please, My Interferogram is getting so many 0i0j values as to question my findings, so i need a explanation as to why they are occuring.

A Jira ticket ([SNAP-3687] - JIRA) has been created to track the problem. We will look into it and thank you for pointing out the issue.

1 Like