H alpha decomposition in SNAP

Hi all!

Could someone tell me the steps to follow (in Graph Builder) when trying to produce H-alpha decomposition of a S1 SLC product? I do not need the H-alpha plane, just an image of alpha angles. I have the following, but not sure if the “Calibration” step is required and the output image looks peculiar… -

When I try the above procedure, the output images does look peculiar, with some wave-like artefacts - see snapshot -

Maybe not related to your question: Why do you reproject after Terrain Correction? You could directly use the desired coordinate reference system in the TC module.

About the result: Sentinel-1 has only 2 polarizations. It is currently discussed if this is enough for a decomposition which was originally based on quad-pol data.

Most studies indicate that the quality is lower than for quad-polarized data:

Sources:


2 Likes

Thanks for the comments ABraun. You are right that the reprojection step may be avoided. I also came across another post, where a slightly longer process chain was described. Will this have clear benefits over the chain I describe above?

I don’t think so.
To my understanding, complex calibration, debursting, decomposition and terrain correction are sufficient. Multi-looking and speckle filtering could theoretically increase your image quality but I would first have a look at the data and then decide to apply a filter.
And the decomposition module doesn’t need the matrix notation of your backscatter information. I’m not even sure if matrices can be derived from dual-pol data.

3 Likes

Great, thank you for the help!

Could I jump the step of “TOPSAR-Split”?Thank you.

Basically the TOPSAR split, is applied to capable selecting the IW sub Swath and the number of bursts, where your study area is located, in case this step is skipped, the entire image of IW-IW2-W3, will be processed.

I think I know how to do it.Thank you very much for your careful reply!