The landsat time value seems to be the acquisition or ingestion date instead of a acquisition timestamp

[aaime]

A more compelling example can be shown, providing more time scattering, if the time value in the database is actually the full acquisition timestamp

[simboss]

I believe the problem here is that the time we have in the filename is like this LC81390452014295LGN00 and the interesting part is like this 2014295 which we are parsing this way: (?<=[a-zA-Z0-9]{9}){6}[0-9]+(?=[a-zA-Z0-9]{5})

We cannot change the filenames since we use landsat-utils to download and ingest, therefore for the moment we don't have a better granularity for the TIME dimension.