Closed BaptisteVandecrux closed 10 months ago
Right now, the percentiles are made on the l3 product (only on t_l, t_u, p_u, rh_u and wspd_u), which makes it easier for filtering. But it could be an advantage to do it on the l1a data, but will it not change a lot downstream, making it more difficult?
Or is it not possible to just do QC on the l3 product?
not urgent, found a local solution
A lot of QC work needs to be done on the merged raw and tx files with readable units. Having an aws-l1 repo, automatically updated by glacio-01 or azure, could save some local processing. @RasmusBahbah any opinion on that? What data do you use for the percentile work?