Sentinel-1 ortho-image graph fails

I am trying to create a custom Sentinel-1 ortho-image generation graph, which is supposed to provide a Geotiff in the end to use it externally. Here’s what I do, including the options:

  1. Read (Data Format: Sentinel-1)
  2. Apply-Orbit-File (Orbit State Vectors: Sentinel Precise, Polynomial Degree: 3)
  3. ThermalNoiseRemoval (Polarisations: VV, Remove Thermal Noise)
  4. Speckle-Filter (Source Bands: Intensity_VV, Filter: Refined Lee)
  5. Terrain-Correction (Source Bands: Intensity_VV, DEM: SRTM 1Sec HGT, DEM Resampling: BILINEAR, Image Resampling: BILINEAR, Pixel Spacing: 10 m, Map Projection: UTM Zone 32, WGS84, Output bands for: Selected source band, nothing else checked)
  6. Write (Save as: GeoTIFF)

I tested the graph on a recent Sentinel-1 IW GRD image, and at first it seems to work like a charm. But when I open the final GeoTIFF (about 2.5 GB in size), I don’t get anything: QGIS tells me only about NaNs, GIMP opens an empty, completely transparent image, and SNAP itself shows some JAVA exception error.

I am using SNAP 5.0.8. on a 64bit Ubuntu machine.

Does anybody have any idea what went wrong?

Could be related to the use of GeoTIFF. Try processing in BEAM-DIMAP and converting to GeoTIFF in a separate step.

This way it worked. Thank you!

There are a few other questions remaining. Maybe they can be solved in this context.

  1. Although I clearly oversampled the image to a 10m pixel spacing, the eventual quality seems somehow degraded. Is the order of actions I applied reasonable? Maybe I change the speckle filter or the interpolation scheme…

  2. Is there any other way to automate the output to GeoTIFF? I am actually seeking to batch process a whole bunch of images this way…

  3. Can I include a subset module already at the beginning of the processing chain? Or will this lead to erroneous results in the geometric modules (e.g. terrain correction)?

  4. Is there any way to output amplitude rather than intensity? After thermal noise removal, the only option I have left is Intensity_VV…

Thanks a lot!