There are a few dozen assets with unexpected polarizations - eg, S1A_IW_GRDH_1SVH_20141017T163045_20141017T163110_002872_0033F7_6B4A or S1A_IW_GRDH_1SHH_20150201T062114_20150201T062133_004426_00569D_C56B
None such assets are produced after July 2015.
Are these errors? Is it normal to have a VH or HV polarization by itself? And should SVV and SHH assets be treated like SSV and SSH?
It depends on the observation plan. There may have been one-time special acquisitions for calibration purposes or there could be a particular part of the world where a specific mode is always acquired.
The general hlop is:
- HH+HV or HH polarization for the monitoring of polar environments, sea-ice zones
- VV+VH or VV polarization for all other observation zones (with an exception for the Baltic Sea observed partially in HH-HV during northern winter, on descending orbits)
but there are no guarantees that this will always be the case.
Figure 3-7 of the Sentinel-1 Product Specification (https://earth.esa.int/documents/247904/349449/Sentinel-1_Product_Specification) indicates that these products are partial dual polarisation acquisitions. So the first product should be treated as a 1SVH polarisation acquisitions (i.e. not 1SSV which would be VV), while for the second this is 1SHH (which is the same as 1SSH). Note the first S in 1SVH indicates that it is a standard product (rather than an annotated product).
Thanks, Peter. For my own education, can you say why acquisitions with partial dual polarization happen? Does it mean there was a problem with the second polarization, or is this mode intentional?
I wasn’t aware that these were being acquired and so I’m not sure why they occurred. Certainly they are not part of the expected acquisition plan as Iveci indicated in his reply.
Is this worth investigating further? There are 224 such assets, which is a large enough number that I’d like to either ingest them into Earth Engine, or understand why they are not worth ingesting.
I’ve investigated further with the S1 QC team. The conclusion is that these partial polarisation products should not be used as they are non-nominal acquisitions and have not been validated. Further steps will be taken to ensure they are not made available to users.
The generation of the partial product should not happen. Those products are not qualified.
When is this happening:
The instrument downlinks the polarimetrics channels (H and V) in two different dowlink channels.
The PDGS contructs first a partial L0 HH andv VH and the reconstructs the full dual pol product DH.
However when a dump is incomplete the full reconstruction is sometimes not possible. After a timeout, the partial L0 product is sent to the S-1 IPF (generation of SLC/GRD) while it shouldn’t.
Along the months, we improved the IPF robustness to support degraded cases and the IPF was more and capable of generating a L1 product out this not expected scenario. After that we tuned the processing flow to avoid the distribution of such products.
Is there quality issues?
Those products were never qualified as they shouldn’t exists.
Furthermore, we believe that they are not radiometrically calibrated.
What to do with those:
You should discard them and send the list to the copernicus helpdesk email@example.com that will remove them from the hub and regenerate (if possible the standard products)
S-1 Data Quality Manager
Thank you, Nuno. List sent.