SNAP guidelines

This repository contains a list of guidelines to use Sentinel-1 SLC assets inside SNAP to create different products:

1. interferograms;
2. Digital Elevation Models (DEMs).

Interferograms

Based on the tutorials:

• on creating the interferogram of NASA, Skywatch, ESA;
• on phase-unwrapping of NASA

Follow this steps:

1. download two Sentinel-1 SLC images as zip files, one before and one after the date of earthquake or landslide. There are many sources for Sentinel-1 data. I suggest you to try with Earthdata NASA and Copernicus Dataspace. Both websites require free registration;
2. SAR instruments have different configurations and expert can give you advice on which data is better to choose to create proper interferograms. I am here to suggest to choose that images before and after should be from the same configuration (IW,SM,EW,WV,ascending,descending) but can be from different instrument (S1A and S1B);
3. open the SLC layers on SNAP. As a check, try to double click on an amplitude image and look at the result on the right panel;
4. _Orb_Stack coregistration of images. The tools is called S-1 TOPS coregistration and can be done with or without ESD (bursts-splitted images). Select in the TOPSAR-Split tab a sub-swath of the full image to coregister the images. If your S1 images have more than one polarization here you can choose which band to use for coregistration. In the Apply-Orbit-File tab you can choose orbital corrections: the default Sentinel Precise will choose the effemerides. Co-registration can also be assisted by DEM in Back-Geocoding tab. You can choose the default SRTM 3Sec or SRTM 1sec HGT (the one I choose for this tutorial). The Output Deramp and Demod Phase flag will try to improve the coregistration. After co-registration you will obtain a new object called (name_of_your_first_image)_Orb_Stack; After the coregistration has ended, we can check the result using the Open RGB Image Window tool. We can set the master (mst) image to be in the R and G channels while slave (slv) to be in the B channel. The coregistration has been corrected if the blue color does not emerge over yellow colours and if the borders are sharp.
5. _Orb_Stack_ifg creation of the interferogram. The idea is to multiply the coregistered master image to the complex conjugate of the coregistered slave image. Use the tool interferogram formation. We are interested in looking only at the phase difference due to the landslide or earthquake so we want to get rid of other contributions to that phase difference. First, we will subtract the flat-earth phase that is the Earth's curvature contribution to the phase that can be removed using orbital data and ellipsoid parameters. During this steps we also create a coherence layer that is used to determine the quality of the final interferogram using the fact that neighbour terrain pixels should have similar values of phase value. Low coherence (< 0.3) will be found in forests where trees will mask the terrain, while high coherence (> 0.6) will be found in cities where flat surfaces quite parallel to the terrain will be stable between images. The coherence layer will be used in the unwrapping step to create a significative displacement map. Run the interferogram formation tool with default parameters. The topographic contribution to the phase can be removed here but we prefer to remove in a different step. THe result of the interferogram formation has a Phase band that is the interferogram;
6. _Orb_Stack_ifg_deb eliminate the grid between bursts. Sentinel-1 images are taken into a bursts mode so they are a collage of spot. In between, there are blind spots that can be filled using the TOPS deburst tool
7. _Orb_Stack_ifg_deb_dinsar remove the topographic contribution to the phase. The Topographic phase removal tool will delete topographic contribution to the phase which comes from a topographic respect to the flat terrain (relative to the ellipsoid surface). A DEM formation wants to preserve and use this phase change but here we want to see terrain changes in time. We then use a DTM to eliminate this topographic contribution. The DEM of this tutorial is SRTM 1sec HGT
8. _Orb_Stack_ifg_deb_dinsar_ML_flt other phase contribution removal. Atmospheric noise, volume scattering, time variation of scatterers and different view angles or orbit configuration (ascending or descending) give other contribution to the phase change between images. Removing these will first require multi-looking tool that is a bidimensional spatial filter, acting as a convolutional kernel. Choose the bands i (real), q (imaginary), coh (coherence). The phase band (which is a derived band respect to i and q) will disapper and come back later. Use look number as 6
9. _Orb_Stack_ifg_deb_dinsar_ML_flt Goldstein phase filtering. We use a Fast Fourier Transform (FFT) to improve SNR (signal to noise ratio)
10. subset_Orb_Stack_ifg_deb_dinsar_ML_flt subset. Select a smaller region right clicking on the area in the right panel that you want to keep and choose Spatial subset from View;
11. _Orb_Stack_ifg_deb_dinsar_ML_flt_subset_unw unwrap the phase. The 2π-periodicity of the phase hides the linear displacement of the land. Unwrapping is the way you go from periodical phase to displacement. First we export the wrapped phase using Snaphu export. In the parameter window choose 200 as row overlap and column overlap. Insert the folder you want to save the file in. Speaking of SNAP version 9 there is a bug where you cannot browse the folder but you can insert the path manually. Secondly, use Snaphu unwrapping selecting the product subset_Orb_Stack_ifg_deb_dinsar_ML_flt you exported and the folder has to be the same as the one you put in Snaphu export. Use the Display execution output to see if the procedure is working. If not, go to the setting Manage External Tools, select Snaphu unwrapping and choose Edit the selected operator. In the system variable USERPROFILE put the folder where you exported the file with Snaphu export. If it still fails to work, check to have enough space on your system. I did not find any adjustment to make it work properly for every data source. Then use Snaphu Import:

Far partire nuovamente lo strumento. Se non dovesse funzionare controllare di avere abbastanza spazio su disco; usiamo “Snaphu Import”: in ReadPhase seleziono il prodotto che è stato scelto da esportare. In Read-Unwrapped-Phase seleziono il file .hdr della unwrapped phase. In Snaphu-Import lascio non selezionato il non-salvataggio. In Write aggiungo _unw al nome. (_Orb_Stack_ifg_deb_ML_flt_subset_unw_dsp) vogliamo ora passare dalla fase unwrapped allo spostamento. Per questo usiamo “Phase to Displacement”. Quando si effettua questo passaggio viene lasciato indietro il layer di coherence. Se si vuole riportarlo si può usare Band Maths sul layer di displacement, tolgo la selezione Virtuale e come espressione preso il file _unw e seleziono la banda coherence; (_Orb_Stack_ifg_deb_ML_flt_subset_unw_dsp_TC) applico la funzione “Range Doppler Terrain correction” per riportare dalle slant range coordinates (che sono le coordinate radar) alle ground range coordinates (che sono le coordinate sul terreno). Per far questo si usa un DEM di riferimento. Se si vuole esportare su Google Earth conviene usare WGS84; esportare con “View as Google Earth KMZ” come file KMZ in modo da poterlo caricare su Google Earth.

DEM generation

Based on the tutorials:

• ESA