Skip to content
Snippets Groups Projects
Commit f87668b1 authored by lkugler's avatar lkugler
Browse files

update

parent 853c180a
No related branches found
No related tags found
No related merge requests found
......@@ -3,7 +3,14 @@ Specifically, one observation per column which is the maximum of the column
Use this script before running the OSSE workflow, to prepare obs_seq.out files.
Note:
path_3d_obsseq = '/path/exp_obs10_loc20/obs_seq_out/2008-07-30_%H:%M_obs_seq.out'
Note:
Only works in case there is 1 observation type!
Example:
python obsseq_2dim.py exp_v1.22_P2_rr_REFL_obs10_loc20_oe2.5 2008-07-30_13:00
"""
from copy import copy
......@@ -17,42 +24,41 @@ from dartwrf import utils
from dartwrf import assim_synth_obs as aso
from dartwrf import obsseq
def _get_n_obs_per_layer(oso):
"""Get number of observations per layer"""
height_all = np.array([a[2] for a in oso.df.loc3d])
height_first = height_all[0]
# count how often this height appears
n_obs_per_layer = int(np.sum(height_all == height_first))
return n_obs_per_layer
if __name__ == "__main__":
if __name__ == "__main__":
exp = sys.argv[1]
assim_time = dt.datetime.strptime(sys.argv[2], "%Y-%m-%d_%H:%M")
path_3d_obsseq = cluster.archive_base+exp+'/obs_seq_out/%Y-%m-%d_%H:%M_obs_seq.out'
oso_input = assim_time.strftime(path_3d_obsseq)
# load experiment config
sys.path.insert(0, cluster.archivedir+'/'+exp+'/DART-WRF')
from config import cfg
# only assured to work with single obstype
if len(cfg.exp.observations) > 1:
raise NotImplementedError()
# existing obsseq with multi levels
oso = obsseq.ObsSeq(oso_input)
n_obs = cfg.exp.observations[0]['n_obs']
nlev = len(oso.df)/n_obs
if nlev - int(nlev) != 0:
raise RuntimeError()
nlev = int(nlev) # levels per obs
print('nlev', nlev)
print('n_obs', n_obs)
n_obs_3d = len(oso.df)
n_obs_per_layer = _get_n_obs_per_layer(oso)
nlev = int(n_obs_3d/n_obs_per_layer)
assert np.allclose(nlev, n_obs_3d/n_obs_per_layer), 'n_obs not evenly divisible!'
print('n_obs_per_layer', n_obs_per_layer)
print('n_obs_3d', n_obs_3d)
output = copy(oso) # copy will be modified
# output.df = output.df.copy() # without this, we get a SettingWithCopyWarning
output.df = output.df.iloc[0::nlev] # every nth level = first level
#print(output.df, oso.df)
# iterate through, set value to max
for i_obs in range(0, n_obs): # go through n_obs (all columns)
for i_obs in range(0, ): # go through n_obs (all columns)
i_obs_subset = i_obs*nlev # jumps by nlev (from one to next column)
column = oso.df.loc[0 + i_obs_subset:nlev + i_obs_subset, :] # select column
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Please to comment