0
#read data
dset = xr.open_mfdataset('/home/users/ruoou/projects/data/cmip6/scenaromip/ssp126/tas/*'+ mod+ '*.nc', parallel=True)

#define timescale
time = dset.time
clim_set_t = dset.sel(time=slice("2015-01-16", "2100-12-16"))
clim_ta = clim_set_t['tas'].groupby('time.month').mean('time', keep_attrs=True)
clim_ta_c = (clim_ta - 273.15)

# define areas
lat = dset.lat
lon = dset.lon
lon_range = lon[(lon>60) & (lon<150)]
lat_range = lat[(lat>10) & (lat<60)]


clim_ta_c.sel(lon=lon_range, lat=lat_range, month = 1).plot.contourf(ax=ax,
                                          levels=np.arange(-30, 30 + 1, 1),
                                          transform=ccrs.PlateCarree(),
                                          cmap='Spectral_r',
                                          cbar_kwargs={'orientation': 'vertical',     #colorbar
                                                   'label': 'temperature (℃)',
                                                   'ticks': np.arange(-30, 30 + 5, 5),
                                                   'pad': 0.05,
                                                   'shrink': 0.8})

model = dset.attrs['source_id']
title = f'{model} near-surface temperature climatology 2015-2100 (1)'
plt.title(title, fontsize = 13)

This is the main part of my script to plot BCC in ssp126 near-surface temperature climatology for January. Bcc in ssp126 What should I do to add up all models and plot the mean (like average the BCC in ssp126 and IPSL in ssp126)? Thanks for help in advance.

ClimateUnboxed
  • 7,106
  • 3
  • 41
  • 86
supperoppa
  • 3
  • 1
  • I would use Adrian Tompkins answer below. Though, if you are asking a question about averaging CMIP6 files you really should specify whether the time steps and grids are the same in each file. Otherwise the answer could be missing a number of steps – Robert Wilson Apr 04 '22 at 08:07

1 Answers1

2

You tagged the question cdo-climate, so I presume you are also happy to have a cdo based solution, I added a remapping step, but check out remapping questions on stackoverflow, or my remapping video guide for more details.

dir=/home/users/ruoou/projects/data/cmip6/scenaromip/ssp126/tas/
# remap all files to a 1x1 deg grid 
for file in $dir/*modelname*.nc ; do 
   cdo remapbil,r360x180 $file ${file%???}_remap.nc 
done 
# calculate ens mean
cdo ensmean *_remap.nc ensmean.nc

then you can open ensmean.nc and plot it as you wish.

ClimateUnboxed
  • 7,106
  • 3
  • 41
  • 86
  • Upvoted. Though based on the question the files likely have different grids, so another step might be missing – Robert Wilson Apr 04 '22 at 08:05
  • `from cdo import Cdo from cdo import * cdo = Cdo() nc_ensmean = cdo.ensmean(input = '/home/users/ruoou/projects/data/cmip6/scenaromip/ssp126/tas/*.nc', output = 'nc_ensmean.nc') CDOException: (returncode:1) Warning (cdfInqContents): Coordinates variable time can't be assigned! Warning (cdfInqContents): Coordinates variable time can't be assigned! cdo ensmean (Abort): Grid size of the input parameter tas do not match!` Just as Robert said, the grids are different, so it shown the error notification. But thank you all the same. – supperoppa Apr 04 '22 at 17:15
  • Then you need to map all fields to a common grid first, e.g. cdo remapbil,r360x180 in.nc out.nc – ClimateUnboxed Apr 04 '22 at 22:09