no time_coverage_start and time_coverage_end attributes on the csv read?

[jerabaul29]

Using the same test case as in https://github.com/OpenDrift/trajan/issues/67 , seeing now that the time_coverage attributes are missing:

load SealsCruise_Martin data
<xarray.Dataset>
Dimensions:        (trajectory: 5, obs: 2792)
Coordinates:
  * trajectory     (trajectory) int64 0 1 2 3 4
  * obs            (obs) int64 0 1 2 3 4 5 6 ... 2786 2787 2788 2789 2790 2791
Data variables:
    drifter_names  (trajectory) <U2 'T1' 'T2' 'T3' 'T4' 'T5'
    time           (trajectory, obs) datetime64[ns] 2022-03-21T18:00:08 ... 2...
    lat            (trajectory, obs) float64 72.57 72.57 72.56 ... 68.26 68.26
    lon            (trajectory, obs) float64 -16.95 -16.97 ... -23.61 -23.64
Attributes:
    Conventions:         CF-1.10
    featureType:         trajectory
    geospatial_lat_min:  66.83907
    geospatial_lat_max:  72.87927
    geospatial_lon_min:  -31.53676
    geospatial_lon_max:  -15.77999

[gauteh]

No, actually the geospatial bounds are also not required by the standard. Should we keep these? Also drifter_names should maybe be trajectory coordinate.

[knutfrode]

time_coverage and geospatial bounds are not required by CF-convention, but are commonly used/parsed by many softwares as "discovery metadata": https://gcw.met.no/node/4 So I would recommend including these automatically, whenever possible.

[gauteh]

Sounds good, another item for our "adapted cf convention"

fre. 14. apr. 2023, 18:11 skrev Knut-Frode Dagestad < @.***>:

time_coverage and geospatial bounds are not required by CF-convention, but are commonly used/parsed by many softwares as "discovery metadata": https://gcw.met.no/node/4 So I would recommend including these automatically, whenever possible.

— Reply to this email directly, view it on GitHub https://github.com/OpenDrift/trajan/issues/70#issuecomment-1508904817, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAN366VWOQPD6TYPPEE6ELXBFZLFANCNFSM6AAAAAAW6PEYDU . You are receiving this because you commented.Message ID: @.***>

[jerabaul29]

A note: if I remember correctly, when I pushed my data to adc.met.no , they asked me to add these, as they use these to automatically generate som of their metadata on the dataset webpage. This is one example of what @knutfrode says, and if we want to use adc in the future (which is likely), this makes generating these automatically very useful :) .