[Bug]: COG overviews dominated by NaNs about rtc

[hjkristenson]

It looks like this issue has been addressed on the SDS side. There are OPERA RTC products with good overviews starting April 1, 2024. There are duplicate products in the archive, however, so users have to take care to use products that were processed April 18 or later. Are there any details available about the implementation of this change? It looks like there was no change in the PGE version in the metadata for the new products.

[gshiroma]

Thank you for the feedback @hjkristenson ! The software version (SOFTWARE_VERSION for GeoTIFF files or /metadata/sourceData/softwareVersion for the HDF5 file) has been updated from 1.0.1 to 1.0.2. It looks like the SDS team kept the same product version as 1.0. Does this answer your question?

[gshiroma]

Hi @scottyhq , thank you for reporting this issue! You are right, the OPERA RTC-S1 products produced as of today do present this problem. We were able to track this issue down, and we found out the problem is in the re-generation of overviews after this RTC-S1 software (Science Application Software) is executed.

I was informed me that this needs to be done because the RTC-S1 products metadata needs to be updated later in the process and that invalidates the COG format. The problem is that this second processing is not using the overview sampling algorithm set to "AVERAGE", as we have here.

So, the issue is not really in this RTC-S1 software. But I'll keep this issue open until the OPERA Science Data System fix it on their side. Thank again!

[gshiroma]

The regeneration of RTC-S1 imagery overviews was removed from the SDS workflow, resolving this issue on April 17, 2024. All OPERA RTC-S1 products generated thereafter will be free from this problem.