hello,
Summary: S1C imagery collected over the Amazon shows large differences between beams. According to links below, there is a correction available, but either I have been unable to apply it correctly, or it does not work.
This article refers to a new calibration to correct beam-to-beam differences:
It mentions an updated antenna pattern. This work was performed by SAR-MPC and the link above basically copies information that can be found in the “Recalibration of Sentinel-1C” item that can be found Sentinel-1 MPC News
Both articles refer to “a new configuration of processing for Sentinel-1C based on updated AUX_CAL and AUX_PP1 files”. On the SAR-MPC site, one can download the updated “(list of AUX_CAL and list of AUX_PP1)” files referred to.
How does one apply this correction? Below are 5 Options I have considered, having tested the first four with no success.
My test image is S1C_IW_GRDH_1SDV_20260428T230359_20260428T230424_007422_00F0FA_8D26.SAFE and I have tested with separate recent downloads from ASF and Copernicus. I first ran SNAP 12 and then updated to SNAP 13 (and have also used GAMMA). I use a series of gpt commands bundled in a python script to run SNAP.
- Option 1: do nothing but run an updated version of SNAP. Generally speaking, updated calibration files of all sorts are automatically implemented in the processing chain from RAW data to Level1 to user-run SNAP (as long as SNAP is updated). This produced no improvement in beam-to-beam offset.
- Option 2: AI suggested that I run gpt with a flag: gpt Calibration -PauxFile=“Latest Auxiliary File” . This would force SNAP to download and apply the latest calibration files. This made no difference, and then I learned “Latest Auxiliary File” is already the default, so the AI suggestion would not have made a difference.
- Option 3: AI suggested that I run gpt with a flag: gpt Calibration -PauxFile=“External Auxiliary File” -PexternalAuxFile="/archive1/Sentinel1/testSnap/xxx.xml” where the last path in quotes is the path to an aux_file I downloaded from SAR-MPS using links above. This produced a SNAP error saying that the ability to call external calibration files in SNAP does not apply to Sentinel-1 processing. So again AI provided a bad lead.
- Option 4: An AI suggestion that one could unzip the SAFE file of the image and go to the subdirectory /annotation/calibration to find and then either replace the file (e.g., “calibration-s1c-iw-grd-vh-20260428t230359-20260428t230424-007422-00f0fa-002.xml”) with the corrected aux_file downloaded from SAR-MPC (“s1c-aux-cal.xml”), or edit the calibrationXXX.xml file and replace parts of it with contents of the updated aux_file. But one then learns that the calibrationXXX.xml file has totally different fields (e.g.,calibrationVector) than the updated aux_file s1c-aux-cal.xml (e.g., azimuthAntennaPattern). Seeing that these updated files are antennae patterns, which is intuitively something that should be utilized at a lower level in the processing chain than the L1 calibration vectors in the SAFE product, suggests that these updated files to implement the beam-to-beam TOPS correction are not meant to be applied through SNAP by the user. In other words, it should be automatic. In that case, see Options 1-3 above.
- Option 5: This is the “follow the science in the paper and apply the correction manually yourself” option. The link above says “Users generating long time series of radiometric measurements are advised to use the retro compensation methodology described in Schmidt” et al. This paper is from 2023 and while our group is capable and are in fact exploring methods to manually correct calibration issues ourselves, the point of this post is to ensure that we are not somehow missing an automatic correction applied lower in the processing chain (which is nearly always the preferable path). Furthermore, the article says to follow the method in the linked paper “to align the radiometry of data processed before the 03.02.2026”. This inclusion of a date suggests that one should not need to manually correct the imagery if it was collected after February, 2026, presumably because the corrections are applied automatically since that date. But my test image was collected after then, and none of my attempts to correct the beam-to-beam difference show an improvement.
How are these corrections supposed to be applied?