The rules for the XYZT variables longitude, latitude, pressure, and time, are a little more relaxed than #22. There only needs to be a value which labels each point or profile (depending on dimensionality of the coordinate). However, there is an expected order for data to appear in the files:
along the N_PROF dimension, the data should be ordered by date, and by easterly and northerly if date isn't sufficient (T, X, Y)
in the N_LEVELS dimension, the data should be ordered by increasing pressure (Z)
coordinate checks ensure that all data points are labeled by (X, Y, Z, T) and no data are unlabeled.
These checks should operate directly on the xr.Dataset or underlying ndarray objects.
This was not closed in #27 because, while that sorts things while converting, we still need a checker function that tells us if the Dataset is correct.
The rules for the XYZT variables
longitude
,latitude
,pressure
, andtime
, are a little more relaxed than #22. There only needs to be a value which labels each point or profile (depending on dimensionality of the coordinate). However, there is an expected order for data to appear in the files:These checks should operate directly on the xr.Dataset or underlying ndarray objects.