I'm having 4 [GFS] temperature datasets: they are all the same in spatial resolution, the only difference is timestamp - they are for 00 UTC, 06 UTC, 12 UTC, 18 UTC within one day.
I need to compute the mean daily temperature dataset. Is there a way to do instrumentally, but not, like, manually pop values from corresponding nodes, compute mean and insert into a dataset?
import xarray as xr
dst00 = xr.open_dataset('gfs.t00z.pgrb2.1p00.f000', engine='cfgrib')
dst06 = xr.open_dataset('gfs.t06z.pgrb2.1p00.f000', engine='cfgrib')
dst12 = xr.open_dataset('gfs.t12z.pgrb2.1p00.f000', engine='cfgrib')
dst18 = xr.open_dataset('gfs.t18z.pgrb2.1p00.f000', engine='cfgrib')
Direct links to download sample datasets:
You can use xr.concat:
dst = xr.concat([dst00, dst06], dim='time').mean('time')
print(dst)
# Output
<xarray.Dataset>
Dimensions: (latitude: 31, longitude: 101)
Coordinates:
step timedelta64[ns] 00:00:00
heightAboveGround float64 2.0
* latitude (latitude) float64 60.0 61.0 62.0 63.0 ... 88.0 89.0 90.0
* longitude (longitude) float64 0.0 1.0 2.0 3.0 ... 98.0 99.0 100.0
Data variables:
t2m (latitude, longitude) float32 279.8 279.5 ... 243.4 243.4
Check:
import pandas as pd
out = pd.concat([d.to_dataframe()['t2m'] for d in [dst00, dst06, dst]], axis=1,
keys=['dst00', 'dst06', 'mean'])
print(out)
# Output
dst00 dst06 mean
latitude longitude
60.0 0.0 279.837494 279.800812 279.819153
1.0 279.557495 279.370789 279.464142
2.0 279.437469 279.100800 279.269135
3.0 279.227478 278.850800 279.039124
4.0 278.417480 278.190796 278.304138
... ... ... ...
90.0 96.0 243.507477 243.370804 243.439148
97.0 243.507477 243.370804 243.439148
98.0 243.507477 243.370804 243.439148
99.0 243.507477 243.370804 243.439148
100.0 243.507477 243.370804 243.439148
[3131 rows x 3 columns]