MODIS LAI STAC ('modis-15A2H-061') may have broken time metadata?

[erthward]

Hi, MSPC.

I'm prototyping some work on the Hub, and I noticed that when I used the same basic code to load and peruse DataArrays from different collections I got null time data returned for the MODIS LAI dataset ('modis-15A2H-061'). I don't believe this would be expected given my code, especially given that identical code returns working data for other collections, so I took this as a sign that something is wrong with the STAC, but of course please let me know if I'm doing something dumb on my end!

Sample code to reproduce:

import pystac_client, planetary_computer

catalog = pystac_client.Client.open("https://planetarycomputer.microsoft.com/api/stac/v1")
bbox = [-79.21787, 37.25838, -79.20933, 37.26692] # ~8km ENE of Evington, VA
dt = "2018-01-01/2022-05-09"

search = catalog.search(
        collections=['modis-15A2H-061'],
        bbox=bbox,
        datetime=dt,
    )

items = search.get_all_items()

da = stackstac.stack(planetary_computer.sign(items),
                                 bounds_latlon=bbox,
                                 epsg=4326,
                                )

print(da.time)

Screenshot of resulting problem:

Thanks!

[TomAugspurger]

Thanks for the easily reproducible example.

The MODIS items (I think all of the collections) use a null datetime since they are composites over some time period. If you check items[0].datetime you'll see it's None, while items[0].properties["start_datetime"] and end_datetime are valid.

The time dimension returned by stackstac is comes from the datetime. The properties like start_datetime should be available as coordinates, and you should be able to make them a dimension with something like da.swap_dims({"time": "start_datetime"})

[erthward]

Oh, perfect! That's what I've been doing in the meantime, but I thought it was a hack.
Thanks, Tom!

D