Hi
We have a production system that downloading the ERA-5 hourly data at daily basis.
For some reason, the following request did not return 24 hours of data and only contains 2-3 hour data.
Any thoughts? Thanks
Request ID: 271d566a-df1a-41b6-9207-1593c58281cd
import cdsapi
dataset = “reanalysis-era5-land”
request = {
“variable”: [“snow_depth_water_equivalent”],
“year”: “2025”,
“month”: “02”,
“day”: [“14”],
“time”: [
“00:00”, “01:00”, “02:00”,
“03:00”, “04:00”, “05:00”,
“06:00”, “07:00”, “08:00”,
“09:00”, “10:00”, “11:00”,
“12:00”, “13:00”, “14:00”,
“15:00”, “16:00”, “17:00”,
“18:00”, “19:00”, “20:00”,
“21:00”, “22:00”, “23:00”
],
“data_format”: “netcdf”,
“download_format”: “unarchived”,
“area”: [55, -100, 41, -68]
}
client = cdsapi.Client()
client.retrieve(dataset, request).download()
I suppose it is linked to the general issue with timestamp data. Try to load 12-15 Feb instead of 14 only. Perhaps this would help. But I do not know what is the real time. My experiments show that they have 23 hour shift in timestamp, but maybe it is a wrong suggestion.
1 Like
Yes, I have the same issues. A 23 hour shift in timestamp.
It would be very useful if we could define the timezone when requesting. Getting a complete year of data is very painful, because when you do so, the last day of the year has only 1 hour of data, then 23 rows of nan’s. Meanwhile, the first day of the dataset is Dec-31 of the previous year and it also has just one hour of data (23:00). with 23 rows of nan’s.
I have found I can’t request two years of data to get around this, since that gets pushed back for ‘request too large’ (ERA5-land). Ideally, we would just be able to specify a timezone for our dates. Alternatively, it would be great if we could specify start and end dates, so that we could just go from Jan 1 2020 to Jan 2 2021 (ensuring we got all of Dec 31 2020), but the way selections work this would require requesting ALL of 2021.
The only workaround I’ve found is to patch up the missing Dec 31 with a second request for years [“2020”, “2021”], months [“01”, “12”] and days [“01”, “30”, “31”]. Which then gives me the full Dec 31 without running into the ‘request too large’ issue. But this is quite a cumbersome workaround…
1 Like