uataq.Network.get_availability#

Network.get_availability(time_range=None, freq='D', lvls=None, num_processes=1)[source]#

Find when each site has data for the pollutant, and at what level.

Every data file is parsed, so this reflects real measurements rather than which files exist: a time bin counts for a level when any row in it has a non-null concentration, calibrated or _raw (see uataq.pollutants.concentration_columns()). Each bin is labeled with the highest level that covers it, so a bin measured but dropped from final data shows as qaqc.

Parameters:

time_range (TimeRange | str | tuple[str | datetime | datetime64 | None, str | datetime | datetime64 | None] | list[str | datetime | datetime64 | None] | slice | None) – The time range to check, in UTC (see Time range). Default is None (all data).
freq (str) – The size of a time bin, as a pandas period alias: ‘h’, ‘D’, ‘W’, ‘M’, … Default is ‘D’.
lvls (str | list[str] | None) – The data levels to check; one level may be given as a string. Default is None, every level in uataq.filesystem.lvls. Checking fewer is faster.
num_processes (int | Literal['max']) – Number of processes used to parse files. Default is 1. A whole network is thousands of files; use a compute node.

Returns:

One row per segment, with columns SID, lvl, start and stop. Segments are half-open [start, stop). Sites with no data have no rows.

Return type:

DataFrame

Notes

If the network was built without a group, every group that operates an instrument is checked (TRX01 ozone, for instance, is in both the lin and horel archives), not only the one get_obs would read.