emontgomery-usgs
commented
6 years ago Maybe this is a non-issue, since by the time the data gets to CF, it's just dimensioned by time.
Our existing CF files have a z variable that has "long_name: z of the sensor relative to the water surface", and we could certainly populate that for waves data, if considered appropriate.
Two examples from the IOOS catalog have no "z" variable for wave height observations.
BODC seems to reference from the seafloor, but I didn't find a CF representation https://www.bodc.ac.uk/data/documents/series/425403/
related to #279, for when code is implemented to ingest our waves data...
Our waves files (mostly) don't have a depth dimension because they're related to the "surface", not the depth of the sensor making the measurement. Variables like wh_4061 are dimensioned by (time, lat,lon). see https://stellwagen.er.usgs.gov/thredds/dodsC/TSdata/DIAMONDSHOALS/8542p-cal.nc.html
Don't know if there's community agreement on what to put in "z". 0 makes a lot of sense, but there may be better options. In any case, this will need to be addressed before we do the run that treats the waves files.
@rsignell-usgs , can you find examples of what others do for the Z dimension in CF for wave params? my search got empty results and broken TDS...