def merge_jsons(mip, exp, case_id, pmprdir):
json_file_dir_template = os.path.join(
pmprdir,
'%(output_type)', 'mjo',
'%(mip)', '%(exp)', '%(case_id)')
json_file_dir_template = StringConstructor(json_file_dir_template)
json_file_dir = json_file_dir_template(
output_type='metrics_results', mip=mip, exp=exp, case_id=case_id)
print('json_file_dir:', json_file_dir)
json_file_template = 'mjo_stat_%(mip)_%(exp)_da_atm_%(model)_%(realization)_1985-2004'
json_file_template = StringConstructor(json_file_template)
# Search for individual JSONs
json_files = sorted(glob.glob(
os.path.join(
json_file_dir,
json_file_template(
mip=mip, exp=exp, case_id=case_id, model='*', realization='*')+'.json')))
# Remove diveDown JSONs and previously generated merged JSONs if included
json_files_revised = copy.copy(json_files)
for j, json_file in enumerate(json_files):
filename_component = json_file.split('/')[-1].split('.')[0].split('_')
if 'diveDown' in filename_component:
json_files_revised.remove(json_file)
elif 'allModels' in filename_component:
json_files_revised.remove(json_file)
elif 'allRuns' in filename_component:
json_files_revised.remove(json_file)
# Load individual JSON and merge to one big dictionary
for j, json_file in enumerate(json_files_revised):
print(j, json_file)
f = open(json_file)
dict_tmp = json.loads(f.read())
if j == 0:
dict_final = dict_tmp.copy()
else:
dict_merge(dict_final, dict_tmp)
f.close()
# Dump final dictionary to JSON
final_json_filename = json_file_template(
mip=mip, exp=exp, case_id=case_id,
model='allModels', realization='allRuns')+'.json'
final_json_file = os.path.join(json_file_dir, final_json_filename)
print('final_json_filename:', final_json_filename)
with open(final_json_file, 'w') as fp:
json.dump(dict_final, fp, sort_keys=True, indent=4)
print("Done: check ", final_json_file)
def read_json_and_merge_axes(json_dir,
json_file,
statistics,
modes,
minimizeText=False):
model_run_list = []
model_run_list_label = []
mode_season_list = []
a = []
for mode in modes:
# open json
input_file = StringConstructor(json_file)(mode=mode)
with open(os.path.join(json_dir, input_file)) as f:
d = json.load(f)
# Get potential x-axis first
if mode == modes[0]:
models_list = sorted(list(d["RESULTS"].keys()))
for model in models_list:
#if model in ['GFDL-CM3', 'GFDL-CM4']:
if model in ['E3SM']:
runs_list = sort_human(list(d["RESULTS"][model].keys()))
for run in runs_list:
model_run_list.append(model + '_' + run)
if run == runs_list[0]:
model_run_list_label.append(model + '_' + run)
else:
model_run_list_label.append(run)
print(model_run_list)
# season depending on mode
if mode == 'PDO':
seasons = ['monthly']
else:
seasons = ['DJF', 'MAM', 'JJA', 'SON']
# season loop
for season in seasons:
mode_season_list.append(mode + '_' + season)
for model_run in model_run_list:
model = model_run.split('_')[0]
run = model_run.split('_')[-1]
try:
tmp = d["RESULTS"][model][run]["defaultReference"][mode][
season][statistics]
if statistics == 'std_pseudo_pcs':
tmp = tmp / d["REF"]["obs"]["defaultReference"][mode][
season]["pc1_stdv"]
except:
tmp = np.nan
a.append(tmp)
# convert to array and decorate axes
a = np.array(a).reshape(len(mode_season_list), len(model_run_list))
if minimizeText:
X = cdms2.createAxis(model_run_list_label)
else:
X = cdms2.createAxis(model_run_list)
Y = cdms2.createAxis(mode_season_list)
a = MV2.array(a, axes=(Y, X), id=statistics)
return a
def merge_json(mode, eof, mip, exp, case_id, obs, syear, eyear, pmprdir):
json_file_dir_template = 'metrics_results/variability_modes/%(mip)/%(exp)/%(case_id)/%(mode)/%(obs)'
json_file_dir_template = StringConstructor(json_file_dir_template)
json_file_dir = os.path.join(
pmprdir,
json_file_dir_template(mip=mip, exp=exp, case_id=case_id, mode=mode, obs=obs))
json_file_template = 'var_mode_%(mode)_%(eof)_stat_%(mip)_%(exp)_mo_atm_%(model)_%(run)_%(syear)-%(eyear).json'
json_file_template = StringConstructor(json_file_template)
# Search for individual JSONs
json_files = sorted(glob.glob(
os.path.join(
json_file_dir,
json_file_template(mode=mode, eof=eof, mip=mip, exp=exp, model='*', run='*', syear='*', eyear='*'))))
# Remove diveDown JSONs and previously generated merged JSONs if included
json_files_revised = copy.copy(json_files)
for j, json_file in enumerate(json_files):
filename_component = json_file.split('/')[-1].split('.')[0].split('_')
if 'allModels' in filename_component:
json_files_revised.remove(json_file)
elif 'allRuns' in filename_component:
json_files_revised.remove(json_file)
# Load individual JSON and merge to one big dictionary
for j, json_file in enumerate(json_files_revised):
print(j, json_file)
f = open(json_file)
dict_tmp = json.loads(f.read())
if j == 0:
dict_final = dict_tmp.copy()
else:
dict_merge(dict_final, dict_tmp)
f.close()
# Dump final dictionary to JSON
final_json_filename = json_file_template(
mode=mode, eof=eof, mip=mip, exp=exp, model='allModels', run='allRuns', syear=str(syear), eyear=str(eyear))
final_json_file = os.path.join(json_file_dir, final_json_filename)
with open(final_json_file, 'w') as fp:
json.dump(dict_final, fp, sort_keys=True, indent=4)
print('mode:', mode)
# Variables
var = param.varModel
# Check dependency for given season option
seasons = param.seasons
print('seasons:', seasons)
# Observation information
obs_name = param.reference_data_name
obs_path = param.reference_data_path
obs_var = param.varOBS
# Path to model data as string template
modpath = StringConstructor(param.modpath)
if LandMask:
modpath_lf = StringConstructor(param.modpath_lf)
# Check given model option
models = param.modnames
# Include all models if conditioned
if ('all' in [m.lower() for m in models]) or (models == 'all'):
models = ([
p.split('/')[-1].split('.')[1] for p in glob.glob(
modpath(mip=mip, exp=exp, model='*', realization='*',
variable=var))
])
# remove duplicates
models = sorted(list(dict.fromkeys(models)), key=lambda s: s.lower())
required=False)
args = P.get_parameter()
infile_template = args.infile
outfile_template = args.outfile
outpath_template = args.outpath
outfilename_template = args.outfilename
varlist = args.vars
start = args.start
end = args.end
print('start and end are ', start, ' ', end)
print('variable list: ', varlist)
InFile = StringConstructor(infile_template)
OutFile = StringConstructor(outfile_template)
OutFileName = StringConstructor(outfilename_template)
OutPath = StringConstructor(outpath_template)
for var in varlist:
# Build filenames
InFile.variable = var
OutFile.variable = var
OutFileName.variable = var
OutPath.variable = var
infile = InFile()
outfile = OutFile()
outfilename = OutFileName()
outpath = OutPath()
def __get(self):
nfree = 0
names = []
for p in self.parameters_list:
if not p in self.dummies and not p in self.auto_dummies:
v = getattr(self, p)
if v is None \
or \
(isinstance(v,(list,tuple)) and len(v)>1):
already = 0
for pn in names:
if p == pn:
already = 1
elif isinstance(pn, list):
if p in pn: already = 1
if already == 0:
nfree += 1
added = 0
for g in self.grouped:
if p in g:
names.append(g)
added = 1
if added == 0:
names.append(p)
if nfree != 2:
raise 'Error MUST end up with 2 multiple values ! (we have ' + str(
nfree) + ':' + str(names) + ')'
# Now determines length of each axis
axes_length = [1, 1]
# First make sure with have 2 list of parameters
for i in range(2):
if not isinstance(names[i], list):
names[i] = [names[i]]
for n in names[i]:
v = getattr(self, n)
if v is None:
if n == 'component':
axes_length[i] *= 28
elif n == 'time_domain':
axes_length[i] *= 19
else:
raise 'Error, ' + n + ' is not defined correctly, please specify which values you wish to extract'
else:
axes_length[i] *= len(v)
# Creates the dummy array
output = MV2.ones((axes_length[0], axes_length[1]))
# Now mask everywhere
output = MV2.masked_equal(output, 1)
# Indices for filling
i = 0
j = 0
# First creates the filler object and sets all the fixed values !
F = StringConstructor(self.files_structure)
# Ok let's fill it
for p in self.parameters_list:
if not p in self.dummies and not p in self.auto_dummies:
v = getattr(self, p)
if isinstance(v, (list, tuple)):
if len(v) == 1:
v = v[0]
if p in self.slaves.keys():
vslvs = v[1:]
v = v[0]
setattr(F, p, v)
if p in self.slaves.keys():
slvs = self.slaves[p]
for js in range(len(slvs)):
s = slsvs[js]
setattr(F, s, vslvs[js])
else:
setattr(F, p, '*')
else:
if p in self.slaves.keys():
vslvs = v[1:]
v = v[0]
setattr(F, p, v)
if p in self.slaves.keys():
slvs = self.slaves[p]
for js in range(len(slvs)):
s = slsvs[js]
setattr(F, s, vslvs[js])
else:
setattr(F, p, '*')
#fnms=F()
nms = names[0] + names[1]
t1, t2, t3 = self.string_construct(nms)
output = output.ravel()
sp1 = t1.split()
n = len(sp1)
for i in range(len(t2)):
sp2 = t2[i].split()
for j in range(n):
v = sp2[j]
if sp1[j] == 'time_domain':
try:
v = int(v)
except:
pass
if v == 'NONE':
v = ''
setattr(F, sp1[j], v)
#print 'Search string is:',fnms
#f=os.popen('ls '+F()).readlines()
#ip,op,ep=os.popen3('ls '+F())
if self.verbose: print 'command line:', F()
#f=op.readlines()
f = glob.glob(F())
#print 'F is:',f
files = []
for file in f:
files.append(file)
for e in self.exclude:
if file.find(e) > -1:
files.pop(-1)
break
if self.verbose: print 'files:', files
try:
# now we get the one value needed in this file
f = cdms2.open(files[0])
V = f[F.statistic]
component = F.component
time_domain = F.time_domain
if isinstance(component, str):
dic = eval(f.components)
for k in dic.keys():
if dic[k] == F.component:
component = k
if isinstance(F.time_domain, str):
dic = eval(f.time_domain)
for k in dic.keys():
if dic[k] == F.time_domain:
time_domain = k
value = V(time_domain=time_domain,
component=component,
squeeze=1)
output[i] = value
# In case sometihng goes wrong (like modle not processed or inexsitant for this var, etc...)
f.close()
except Exception, err:
#print 'Error:',err
pass
realization = '*'
print('realization: ', realization)
# Metrics Collection
mc_name = param.metricsCollection
# case id
case_id = param.case_id
print('case_id:', case_id)
# Output
outdir_template = param.process_templated_argument("results_dir")
outdir = StringConstructor(
str(
outdir_template(output_type='%(output_type)',
mip=mip,
exp=exp,
metricsCollection=mc_name,
case_id=case_id)))
# Debug
debug = param.debug
print('debug:', debug)
# =================================================
# Create output directories
# -------------------------------------------------
for output_type in ['graphics', 'diagnostic_results', 'metrics_results']:
if not os.path.exists(outdir(output_type=output_type)):
os.makedirs(outdir(output_type=output_type))
print(outdir(output_type=output_type))
def merge_jsons(mip, exp, case_id, pmprdir):
json_file_dir_template = os.path.join(pmprdir, "%(output_type)", "mjo",
"%(mip)", "%(exp)", "%(case_id)")
json_file_dir_template = StringConstructor(json_file_dir_template)
json_file_dir = json_file_dir_template(output_type="metrics_results",
mip=mip,
exp=exp,
case_id=case_id)
print("json_file_dir:", json_file_dir)
json_file_template = (
"mjo_stat_%(mip)_%(exp)_da_atm_%(model)_%(realization)_1985-2004")
json_file_template = StringConstructor(json_file_template)
# Search for individual JSONs
json_files = sorted(
glob.glob(
os.path.join(
json_file_dir,
json_file_template(mip=mip,
exp=exp,
case_id=case_id,
model="*",
realization="*") + ".json",
)))
# Remove diveDown JSONs and previously generated merged JSONs if included
json_files_revised = copy.copy(json_files)
for j, json_file in enumerate(json_files):
filename_component = json_file.split("/")[-1].split(".")[0].split("_")
if "diveDown" in filename_component:
json_files_revised.remove(json_file)
elif "allModels" in filename_component:
json_files_revised.remove(json_file)
elif "allRuns" in filename_component:
json_files_revised.remove(json_file)
# Load individual JSON and merge to one big dictionary
for j, json_file in enumerate(json_files_revised):
print(j, json_file)
f = open(json_file)
dict_tmp = json.loads(f.read())
if j == 0:
dict_final = dict_tmp.copy()
else:
dict_merge(dict_final, dict_tmp)
f.close()
# Dump final dictionary to JSON
final_json_filename = (json_file_template(mip=mip,
exp=exp,
case_id=case_id,
model="allModels",
realization="allRuns") + ".json")
final_json_file = os.path.join(json_file_dir, final_json_filename)
print("final_json_filename:", final_json_filename)
with open(final_json_file, "w") as fp:
json.dump(dict_final, fp, sort_keys=True, indent=4)
print("Done: check ", final_json_file)
description='Runs PCMDI Modes of MJO Computations',
formatter_class=RawTextHelpFormatter)
P = AddParserArgument(P)
param = P.get_parameter()
# Pre-defined options
mip = param.mip
exp = param.exp
print('mip:', mip)
print('exp:', exp)
# Variables
var = param.varModel
# Path to model data as string template
modpath = StringConstructor(param.modpath)
# Check given model option
models = param.modnames
# Include all models if conditioned
if ('all' in [m.lower() for m in models]) or (models == 'all'):
model_index_path = param.modpath.split('/')[-1].split('.').index(
"%(model)")
models = ([
p.split('/')[-1].split('.')[model_index_path] for p in glob.glob(
modpath(mip=mip,
exp=exp,
realm='atmos',
model='*',
realization='*',
# remove duplicates
models = sorted(list(dict.fromkeys(models)), key=lambda s: s.lower())
print('models:', models)
# Realizations
realization = param.realization
print('realization: ', realization)
# case id
case_id = param.case_id
# Output
outdir_template = param.process_templated_argument("results_dir")
outdir = StringConstructor(str(outdir_template(
output_type='%(output_type)',
mip=mip, exp=exp, case_id=case_id)))
# Create output directory
for output_type in ['graphics', 'diagnostic_results', 'metrics_results']:
if not os.path.exists(outdir(output_type=output_type)):
os.makedirs(outdir(output_type=output_type))
print(outdir(output_type=output_type))
# Debug
debug = param.debug
print('debug: ', debug)
# Year
# model
msyear = param.msyear
fac = param.fac
nperseg = param.nperseg
noverlap = param.noverlap
print(modpath)
print(mod)
print(prd)
print(nperseg, noverlap)
# Get flag for CMEC output
cmec = param.cmec
# Create output directory
case_id = param.case_id
outdir_template = param.process_templated_argument("results_dir")
outdir = StringConstructor(
str(outdir_template(output_type="%(output_type)", mip=mip,
case_id=case_id)))
for output_type in ["graphics", "diagnostic_results", "metrics_results"]:
if not os.path.exists(outdir(output_type=output_type)):
try:
os.makedirs(outdir(output_type=output_type))
except FileExistsError:
pass
print(outdir(output_type=output_type))
# Check data in advance
file_list = sorted(glob.glob(os.path.join(modpath, "*" + mod + "*")))
data = []
for file in file_list:
if mip == "obs":
model = file.split("/")[-1].split(".")[2]
def read_json_and_merge_axes(json_dir, json_file, statistics, modes, AverageRuns=False):
runs_list_dic = {}
model_run_list = []
mode_season_list = []
a = []
for mode in modes:
# open json
input_file = StringConstructor(json_file)(mode=mode)
with open(os.path.join(json_dir, input_file)) as f:
d = json.load(f)
# Get potential x-axis first
if mode == modes[0]:
models_list = sorted(list(d["RESULTS"].keys()))
for model in models_list:
runs_list = sort_human(list(d["RESULTS"][model].keys()))
runs_list_dic[model] = runs_list
for run in runs_list:
model_run_list.append(model+'_'+run)
print(model_run_list)
# season depending on mode
if mode == 'PDO':
seasons = ['monthly']
else:
seasons = ['DJF', 'MAM', 'JJA', 'SON']
# season loop
for season in seasons:
mode_season_list.append(mode+'_'+season)
if AverageRuns:
for model in models_list:
b = []
for run in runs_list_dic[model]:
try:
tmp = d["RESULTS"][model][run]["defaultReference"][mode][season][statistics]
if statistics == 'std_pseudo_pcs':
tmp = tmp / d["REF"]["obs"]["defaultReference"][mode][season]["pc1_stdv"]
b.append(tmp)
except:
tmp = np.nan
tmp2 = float(np.mean(np.array(b)))
a.append(tmp2)
else:
for model_run in model_run_list:
model = model_run.split('_')[0]
run = model_run.split('_')[-1]
try:
tmp = d["RESULTS"][model][run]["defaultReference"][mode][season][statistics]
if statistics == 'std_pseudo_pcs':
tmp = tmp / d["REF"]["obs"]["defaultReference"][mode][season]["pc1_stdv"]
except:
tmp = np.nan
a.append(tmp)
# convert to array and decorate axes
if AverageRuns:
#xaxis_label = models_list
xaxis_label = [model+' ('+str(len(runs_list_dic[model]))+')' for model in models_list]
else:
xaxis_label = model_run_list
print('xaixs_label:', xaxis_label, len(xaxis_label))
a = np.array(a).reshape(len(mode_season_list), len(xaxis_label))
X = cdms2.createAxis(xaxis_label)
Y = cdms2.createAxis(mode_season_list)
a = MV2.array(a, axes=(Y,X), id=statistics)
return a
