I need to look at this. But if you need the original data you can find it in the tape archive.
The data range of the 2D output variable total column integrated cloud water (tqc_dia
) has been discretised to only a few values, as can be seen via this MWE:
import eurec4a
import pandas as pd
import matplotlib.pyplot as plt
cat = eurec4a.get_intake_catalog()
timeslice = slice(pd.to_datetime('2020-01-11-10:00'),pd.to_datetime('2020-01-11-12:00'))
data = cat['simulations']['ICON']['LES_CampaignDomain_control']['surface_DOM01'].to_dask().sel(time=timeslice)
plt.hist(data['tqc_dia'].values.ravel(), range=(.01, .1), bins=100)
# or :
# plt.hist(np.log(data['tqc_dia'].values.ravel()), range=(-12, 2), bins=100)
plt.show()
Hauke Schulz (c9c19321) at 05 Jul 20:12
add DOM03 gridspec
From this run it becomes also clear that lprogccn
is False. (/work/mh0010/m300408/DVC-test/EUREC4A-ICON/EUREC4A/dev/run/logs/LOG.exp.DOM01+DOM02large.run.2187156.o)
Hauke Schulz (99291fe0) at 01 Jan 22:08
change postprocessing to EUREC4A_130CCN
Hauke Schulz (2998be77) at 01 Dec 07:51
reduce nodes and increase restart frequency (330 nodes currently fr...
The above notes were not enough to successfully start again another nest (DOM03) based on a different multifile-restart-file. After applying the above script snippet ICON tried to read patch3_*
files from the mfr-file, which do not exist. The solution seems to be to set the simulation end date to a time before the new nest (DOM03) should start, so that DOM03
is not active
:
If DOM03 does not become active within the simulation, no patch3_*
will be loaded, BUT, they will be written to the next mfr file. Once they are written to a mfr-file the simulation end date and DOM03 start time can be set to the actual one.