Rethink ancillary table
Closed this issue · 0 comments
bpbond commented
Breaking out from #179
Right now the ancillary
table is exceedingly free form: just a single value for any number of fields, with no checks on those field names or values at all.
In addition @avnimalhotra points out that time series would be highly useful.
We could restructure it, and the input file in each dataset folder, to allow time- and port-specific ancillary data:
CSR_TIMESTAMP_BEGIN | CSR_TIMESTAMP_END | CSR_PORT | CSR_x | CSR_y | CSR_z |
---|---|---|---|---|---|
So all fluxes are per unit time_interval, where the interval is defined by the timestamps; e.g. for annual NPP you'd enter 2020-01-01 00:00
and 2021-01-01 00:00
. Right?