Newer
Older
# coding=utf-8
import glob
import shutil
import uuid
import os
from datetime import datetime
import pygrib
from autosubmit.config.log import Log
from autosubmit.date.chunk_date_lib import parse_date, chunk_end_date, previous_day, date2str, add_months
from earthdiagnostics.frequency import Frequency, Frequencies
from earthdiagnostics.modelingrealm import ModelingRealms
from earthdiagnostics.utils import TempFile, Utils
class Cmorizer(object):
"""
Class to manage CMORization
:param data_manager: experiment's data manager
:type data_manager: DataManager
:param startdate: startdate to cmorize
:type startdate: str
:param member: member to cmorize
:type member: int
"""
NON_DATA_VARIABLES = ('lon', 'lat', 'time', 'time_bnds', 'leadtime', 'lev', 'icethi',
'deptht', 'depthu', 'depthw', 'depthv', 'time_centered', 'time_centered_bounds',
'deptht_bounds', 'depthu_bounds', 'depthv_bounds', 'depthw_bounds',
Javier Vegas-Regidor
committed
'deptht_bnds', 'depthu_bnds', 'depthv_bnds', 'depthw_bnds',
'time_counter_bounds', 'ncatice', 'nav_lat_grid_V', 'nav_lat_grid_U',
'nav_lat_grid_T', 'nav_lon_grid_V', 'nav_lon_grid_U', 'nav_lon_grid_T',
'depth', 'depth_2', 'depth_3', 'depth_4',
'mlev', 'hyai', 'hybi', 'hyam', 'hybm')
ALT_COORD_NAMES = {'time_counter': 'time', 'time_counter_bnds': 'time_bnds', 'time_counter_bounds': 'time_bnds',
'tbnds': 'bnds', 'nav_lat': 'lat', 'nav_lon': 'lon', 'x': 'i', 'y': 'j'}
def __init__(self, data_manager, startdate, member):
self.data_manager = data_manager
self.startdate = startdate
self.member = member
self.config = data_manager.config
self.experiment = self.config.experiment
self.cmor = self.config.cmor
self.member_str = self.experiment.get_member_str(member)
self.original_files_path = os.path.join(self.config.data_dir, self.experiment.expid, 'original_files',
self.startdate, self.member_str, 'outputs')
self.atmos_timestep = None
self.cmor_scratch = os.path.join(self.config.scratch_dir, 'CMOR')
def cmorize_ocean(self):
"""
CMORizes ocean files from MMO files
:return:
"""
if not self.cmor.ocean:
return
self._cmorize_ocean_files('MMO')
self._cmorize_ocean_files('PPO')
self._cmorize_ocean_files('diags')
tar_folder = os.path.join(self.original_files_path, '{0}*'.format(prefix))
tar_files = glob.glob(tar_folder)
tar_files.sort()
count = 1
for tarfile in tar_files:
Log.info('Unpacking oceanic file {0}/{1}'.format(count, len(tar_files)))
self._unpack_tar_file(tarfile)
self._cmorize_nc_files()
Log.result('Oceanic file {0}/{1} finished'.format(count, len(tar_files)))
count += 1
def _cmorize_nc_files(self):
for filename in glob.glob(os.path.join(self.cmor_scratch, '*.nc')):
self._cmorize_nc_file(filename)
def _correct_fluxes(self):
fluxes_vars = ("prsn", "rss", "rls", "rsscs", "rsds", "rlds")
for filename in glob.glob(os.path.join(self.cmor_scratch, '*.nc')):
handler = Utils.openCdf(filename)
for varname in handler.variables.keys():
cmor_var = self.data_manager.variable_list.get_variable(varname, True)
if cmor_var.short_name not in fluxes_vars:
continue
handler.variables[varname][:] = handler.variables[varname][:] / self.experiment.atmos_timestep * 3600
handler.close()
def _unpack_tar_file(self, tarfile):
if os.path.exists(self.cmor_scratch):
shutil.rmtree(self.cmor_scratch)
os.makedirs(self.cmor_scratch)
Utils.untar((tarfile,), self.cmor_scratch)
Utils.unzip(glob.glob(os.path.join(self.cmor_scratch, '*.gz')))
def _merge_mma_files(self, tarfile):
temp = TempFile.get()
for filename in glob.glob(os.path.join(self.cmor_scratch, 'MMA_*_SH_*.nc')):
Utils.cdo.sp2gpl(options='-O', input=filename, output=temp)
shutil.move(temp, filename)
sh_files = glob.glob(os.path.join(self.cmor_scratch, 'MMA_*_SH_*.nc'))
Utils.cdo.mergetime(input=sh_files, output=os.path.join(self.cmor_scratch, 'sh.nc'))
gg_files = glob.glob(os.path.join(self.cmor_scratch, 'MMA_*_GG_*.nc'))
Utils.cdo.mergetime(input=gg_files, output=os.path.join(self.cmor_scratch, 'gg.nc'))
for filename in sh_files + gg_files:
os.remove(filename)
Utils.nco.ncks(input=os.path.join(self.cmor_scratch, 'sh.nc'),
output=os.path.join(self.cmor_scratch, 'gg.nc'), options='-A')
os.remove(os.path.join(self.cmor_scratch, 'sh.nc'))
tar_startdate = tarfile[0:-4].split('_')[5].split('-')
new_name = 'MMA_1m_{0[0]}_{0[1]}.nc'.format(tar_startdate)
shutil.move(os.path.join(self.cmor_scratch, 'gg.nc'), os.path.join(self.cmor_scratch, new_name))
def cmorize_atmos(self):
"""
CMORizes atmospheric data, from grib or MMA files
:return:
"""
if not self.cmor.atmosphere:
return
if self.cmor.use_grib and self.gribfiles_available():
self._cmorize_grib_files()
Javier Vegas-Regidor
committed
else:
self._cmorize_mma_files()
def _cmorize_mma_files(self):
tar_files = glob.glob(os.path.join(self.original_files_path, 'MMA*'))
tar_files.sort()
count = 1
for tarfile in tar_files:
Log.info('Unpacking atmospheric file {0}/{1}'.format(count, len(tar_files)))
self._unpack_tar_file(tarfile)
self._merge_mma_files(tarfile)
self._cmorize_nc_files()
Log.result('Atmospheric file {0}/{1} finished'.format(count, len(tar_files)))
count += 1
def _cmorize_grib_files(self):
count = 1
chunk_start = parse_date(self.startdate)
while os.path.exists(self.get_original_grib_path(chunk_start, 'GG')) or \
os.path.exists(self.get_original_grib_path(chunk_start, 'SH')):
chunk_end = chunk_end_date(chunk_start, self.experiment.chunk_size, 'month', 'standard')
chunk_end = previous_day(chunk_end, 'standard')
Log.info('CMORizing chunk {0}-{1}', date2str(chunk_start), date2str(chunk_end))
for grid in ('SH', 'GG'):
Log.info('Processing {0} variables', grid)
if not os.path.exists(self.get_original_grib_path(chunk_start, grid)):
self.cmorize_grib_file(chunk_end, chunk_start, count, grid)
chunk_start = chunk_end_date(chunk_start, self.experiment.chunk_size, 'month', 'standard')
def cmorize_grib_file(self, chunk_end, chunk_start, count, grid):
for month in range(0, self.experiment.chunk_size):
current_date = add_months(chunk_start, month, 'standard')
original_gribfile = self.get_original_grib_path(current_date, grid)
Log.info('Processing month {1}', grid, date2str(current_date))
gribfile = self.get_scratch_grib_path(current_date, grid)
if not os.path.isfile(gribfile):
Log.info('Copying file...', grid, date2str(current_date))
Utils.copy_file(original_gribfile, gribfile)
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
self._obtain_atmos_timestep(gribfile)
prev_gribfile = self.get_scratch_grib_path(add_months(current_date, -1, 'standard'), grid)
if os.path.exists(prev_gribfile):
self._merge_grib_files(current_date, prev_gribfile, gribfile)
full_file = 'ICM'
else:
full_file = gribfile
Log.info('Unpacking... ')
# remap on regular Gauss grid
if grid == 'SH':
Utils.cdo.splitparam(input='-sp2gpl {0}'.format(full_file), output=gribfile + '_',
options='-f nc4')
else:
Utils.cdo.splitparam(input=full_file, output=gribfile + '_', options='-R -f nc4')
# total precipitation (remove negative values)
Utils.cdo.setcode(228, input='-setmisstoc,0 -setvrange,0,Inf -add '
'{0}_{{142,143}}.128.nc'.format(gribfile),
output='{0}_228.128.nc'.format(gribfile))
Utils.remove_file('ICM')
next_gribfile = self.get_original_grib_path(add_months(current_date, 1, 'standard'), grid)
if not os.path.exists(next_gribfile):
os.remove(gribfile)
cdo_reftime = parse_date(self.startdate).strftime('%Y-%m-%d,00:00')
self._ungrib_vars(cdo_reftime, gribfile, current_date.month, '{0}hr'.format(self.atmos_timestep))
self._ungrib_vars(cdo_reftime, gribfile, current_date.month, '1d')
self._ungrib_vars(cdo_reftime, gribfile, current_date.month, '1m')
for splited_file in glob.glob('{0}_*.128.nc'.format(gribfile)):
os.remove(splited_file)
Log.result('Month {0}, {1} variables finished', date2str(current_date), grid)
count += 1
self._merge_and_cmorize_atmos(chunk_start, chunk_end, grid, '1m')
self._merge_and_cmorize_atmos(chunk_start, chunk_end, grid, '1d')
self._merge_and_cmorize_atmos(chunk_start, chunk_end, grid,
'{0}hr'.format(self.atmos_timestep))
def get_scratch_grib_path(self, current_date, grid):
return os.path.join(self.config.scratch_dir, self._get_grib_filename(grid, current_date))
def _obtain_atmos_timestep(self, gribfile):
if self.atmos_timestep is None:
self.atmos_timestep = self._get_atmos_timestep(gribfile)
def get_original_grib_path(self, current_date, grid):
return os.path.join(self.original_files_path,
self._get_grib_filename(grid, current_date))
def _get_grib_filename(self, grid, month):
return 'ICM{0}{1}+{2}.grb'.format(grid, self.experiment.expid, date2str(month)[:-2])
def _get_atmos_timestep(self, gribfile):
Log.info('Getting timestep...')
grib_handler = pygrib.open(gribfile)
mes1 = grib_handler.message(1)
mes2 = grib_handler.readline()
while mes2.analDate == mes1.analDate:
mes2 = grib_handler.readline()
atmos_timestep = mes2.analDate - mes1.analDate
atmos_timestep = int(atmos_timestep.total_seconds() / 3600)
self.experiment.atmos_timestep = atmos_timestep
grib_handler.close()
return atmos_timestep
def _cmorize_nc_file(self, filename):
Log.info('Processing file {0}', filename)
if not self._contains_requested_variables(filename):
os.remove(filename)
return
Utils.convert2netcdf4(filename)
frequency = self._get_nc_file_frequency(filename)
Utils.rename_variables(filename, Cmorizer.ALT_COORD_NAMES, False, True)
Javier Vegas-Regidor
committed
self._remove_valid_limits(filename)
Javier Vegas-Regidor
committed
self._add_common_attributes(filename, frequency)
self._update_time_variables(filename)
handler = Utils.openCdf(filename)
Log.info('Splitting file {0}', filename)
for variable in handler.variables.keys():
if variable in Cmorizer.NON_DATA_VARIABLES:
continue
self.extract_variable(filename, handler, frequency, variable)
Log.result('File {0} cmorized!', filename)
handler.close()
os.remove(filename)
# noinspection PyMethodMayBeStatic
Javier Vegas-Regidor
committed
def _remove_valid_limits(self, filename):
handler = Utils.openCdf(filename)
for variable in handler.variables.keys():
var = handler.variables[variable]
if 'valid_min' in var.ncattrs():
del var.valid_min
if 'valid_max' in var.ncattrs():
del var.valid_max
handler.close()
file_parts = os.path.basename(filename).split('_')
if self.experiment.expid in [file_parts[1], file_parts[2]]:
frequency = Frequency('m')
Javier Vegas-Regidor
committed
elif self.experiment.expid == file_parts[0]:
try:
parse_date(file_parts[1])
frequency = Frequency('m')
Javier Vegas-Regidor
committed
except ValueError:
frequency = Frequency(file_parts[1][1])
frequency = Frequency(file_parts[1][1])
return frequency
def _contains_requested_variables(self, filename):
variables = Utils.get_file_variables(filename)
return self.cmor.any_required(variables)
def extract_variable(self, file_path, handler, frequency, variable):
"""
Extracts a variable from a file and creates the CMOR file
:param file_path: path to the file
:type file_path: str
:param handler: netCDF4 handler for the file
:param frequency: variable's frequency
:type frequency: Frequency
:param variable: variable's name
:type variable: str
"""
alias, var_cmor = self.data_manager.variable_list.get_variable_and_alias(variable)
if var_cmor is None:
return
if not self.cmor.cmorize(var_cmor):
return
Utils.nco.ncks(input=file_path, output=temp, options='-v {0}'.format(variable))
self._rename_level_variables(temp, var_cmor)
self._add_coordinate_variables(handler, temp)
date_str = self.get_date_str(file_path)
if date_str is None:
Log.error('Variable {0} can not be cmorized. Original filename does not match a recognized pattern',
var_cmor.short_name)
raise CMORException('Variable {0}:{1} can not be cmorized. Original filename does not match a recognized '
'pattern'.format(var_cmor.domain, var_cmor.short_name))
self.data_manager.send_file(temp, var_cmor.domain, var_cmor.short_name, self.startdate, self.member,
frequency=frequency, rename_var=variable, date_str=date_str, region=region,
move_old=True, grid=alias.grid, cmorized=True)
def get_date_str(self, file_path):
file_parts = os.path.basename(file_path).split('_')
if file_parts[0] in (self.experiment.expid, 'MMA', 'MMO') or file_parts[0].startswith('ORCA'):
# Model output
return '{0}-{1}'.format(file_parts[2][0:6], file_parts[3][0:6])
elif file_parts[1] == self.experiment.expid:
# Files generated by the old version of the diagnostics
return '{0}-{1}'.format(file_parts[4][0:6], file_parts[5][0:6])
else:
return None
Javier Vegas-Regidor
committed
@staticmethod
def _add_coordinate_variables(handler, temp):
handler_cmor = Utils.openCdf(temp)
Utils.copy_variable(handler, handler_cmor, 'lon', False)
Utils.copy_variable(handler, handler_cmor, 'lat', False)
if 'time' in handler_cmor.dimensions.keys():
Utils.copy_variable(handler, handler_cmor, 'leadtime', False)
handler_cmor.close()
Javier Vegas-Regidor
committed
@staticmethod
def _rename_level_variables(temp, var_cmor):
if var_cmor.domain == ModelingRealms.ocean:
Utils.rename_variables(temp, {'deptht': 'lev', 'depthu': 'lev', 'depthw': 'lev', 'depthv': 'lev',
'depth': 'lev'}, False, True)
if var_cmor.domain in [ModelingRealms.landIce, ModelingRealms.land]:
Utils.rename_variables(temp, {'depth': 'sdepth', 'depth_2': 'sdepth', 'depth_3': 'sdepth',
'depth_4': 'sdepth'}, False, True)
if var_cmor.domain == ModelingRealms.atmos:
Utils.rename_variables(temp, {'depth': 'plev'}, False, True)
Javier Vegas-Regidor
committed
@staticmethod
def translate_frequency(frequency):
return Frequency(frequency)
@staticmethod
def _merge_grib_files(current_month, prev_gribfile, gribfile):
Log.info('Merging data from different files...')
fd = open('rules_files', 'w')
fd.write('if (dataDate >= {0.year}{0.month:02}01) {{ write ; }}\n'.format(current_month))
fd.close()
# get first timestep for each month from previous file (if possible)
if os.path.exists('ICM'):
os.remove('ICM')
Utils.execute_shell_command('grib_filter -o ICM rules_files '
'{0} {1}'.format(os.path.basename(prev_gribfile),
os.path.basename(gribfile)))
os.remove('rules_files')
Utils.remove_file(prev_gribfile)
def _ungrib_vars(self, cdo_reftime, gribfile, month, frequency):
Log.info('Preparing {0} variables'.format(frequency))
var_codes = self.config.cmor.get_variables(frequency)
for var_code in var_codes:
if not os.path.exists('{0}_{1}.128.nc'.format(gribfile, var_code)):
continue
new_units = None
cdo_operator = '-selmon,{0}'.format(month)
if frequency == Frequencies.monthly:
if var_code == 201:
cdo_operator = "-monmean -daymax {0}".format(cdo_operator)
elif var_code == 202:
cdo_operator = "-monmean -daymax {0}".format(cdo_operator)
else:
cdo_operator = "-monmean {0} ".format(cdo_operator)
if frequency == Frequencies.daily:
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
if var_code == 201:
cdo_operator = "-daymax {0} ".format(cdo_operator)
elif var_code == 202:
cdo_operator = "-daymin {0} ".format(cdo_operator)
else:
cdo_operator = "-daymean {0} ".format(cdo_operator)
if var_code in (144, 146, 147, 169, 175, 176, 177, 179, 180, 181, 182, 201, 202, 205, 212, 228):
cdo_operator = '{0} -shifttime,-{1}hours'.format(cdo_operator, self.experiment.atmos_timestep)
if var_code == 129:
# geopotential
new_units = "m"
cdo_operator = "-divc,9.81 {0}".format(cdo_operator)
elif var_code in (146, 147, 169, 175, 176, 177, 179, 212):
# radiation
new_units = "W m-2"
cdo_operator = "-divc,{0} {1}".format(self.experiment.atmos_timestep * 3600, cdo_operator)
elif var_code in (180, 181):
# momentum flux
new_units = "N m-2"
cdo_operator = "-divc,{0} {1}".format(self.experiment.atmos_timestep * 3600, cdo_operator)
elif var_code in (144, 182, 205, 228):
# precipitation/evaporation/runoff
new_units = "kg m-2 s-1"
cdo_operator = "-mulc,1000 -divc,{0}".format(self.experiment.atmos_timestep * 3600)
levels = self.config.cmor.get_levels(frequency, var_code)
if levels:
cdo_operator = "{0} -sellevel,{1}".format(cdo_operator, levels)
Utils.execute_shell_command('cdo -t ecmwf setreftime,{0} '
'{1} {2}_{3}.128.nc '
'{2}_{3}_{4}.nc'.format(cdo_reftime, cdo_operator,
gribfile, var_code, frequency))
h_var_file = '{0}_{1}_{2}.nc'.format(gribfile, var_code, frequency)
handler = Utils.openCdf(h_var_file)
if new_units:
for var in handler.variables.values():
if 'code' in var.ncattrs() and var.code == var_code:
var.units = new_units
break
var_name = None
for key in handler.variables.keys():
if key + '_2' in handler.variables and key not in handler.dimensions:
var_name = key
handler.close()
if var_name is not None:
Utils.nco.ncks(input='{0}_{1}_1m.nc'.format(gribfile, var_code),
output='{0}_{1}_1m.nc'.format(gribfile, var_code),
options='-O -v {0}'.format(var_name))
def _merge_and_cmorize_atmos(self, chunk_start, chunk_end, grid, frequency):
merged_file = 'MMA_{0}_{1}_{2}_{3}.nc'.format(frequency, date2str(chunk_start), date2str(chunk_end), grid)
files = glob.glob(os.path.join(self.config.scratch_dir,
'{0}_*_{1}.nc'.format(self._get_grib_filename(grid, chunk_start), frequency)))
for first_file in files:
shutil.move(first_file, merged_file)
current_month = add_months(chunk_start, 1, 'standard')
while current_month < chunk_end:
month_file = first_file.replace('+{0}.grb'.format(date2str(chunk_start)[:-2]),
'+{0}.grb'.format(date2str(current_month)[:-2]))
Utils.concat_variables(month_file, merged_file, True)
current_month = add_months(current_month, 1, 'standard')
self._cmorize_nc_file(merged_file)
Javier Vegas-Regidor
committed
def _update_time_variables(self, filename):
handler = Utils.openCdf(filename)
time_var = handler.variables['time']
Javier Vegas-Regidor
committed
if "time_bnds" in handler.variables:
time_var.bounds = "time_bnds"
handler.variables['time_bnds'].units = time_var.units
handler.close()
temp = TempFile.get()
Utils.cdo.setreftime('1850-01-01,00:00:00,days', input=filename, output=temp)
Utils.move_file(temp, filename)
self._set_leadtime_var(filename)
Javier Vegas-Regidor
committed
def _set_leadtime_var(self, filename):
handler = Utils.openCdf(filename)
if 'leadtime' in handler.variables:
var = handler.variables['leadtime']
else:
var = handler.createVariable('leadtime', float, 'time')
var.units = "days"
var.long_name = "Time elapsed since the start of the forecast"
var.standard_name = "forecast_period"
leadtime = Utils.get_datetime_from_netcdf(handler)
startdate = parse_date(self.startdate)
leadtime = [datetime(time.year, time.month, time.day, time.hour, time.minute, time.second) - startdate
for time in leadtime]
for lt in range(0, len(leadtime)):
var[lt] = leadtime[lt].days
Javier Vegas-Regidor
committed
handler.close()
Javier Vegas-Regidor
committed
def _add_common_attributes(self, filename, frequency):
cmor = self.config.cmor
experiment = self.config.experiment
Javier Vegas-Regidor
committed
handler = Utils.openCdf(filename)
handler.associated_experiment = cmor.associated_experiment
handler.batch = '{0}{1}'.format(experiment.institute, datetime.now().strftime('%Y-%m-%d(T%H:%M:%SZ)'))
Javier Vegas-Regidor
committed
handler.contact = 'Pierre-Antoine Bretonniere, pierre-antoine.bretonniere@bsc.es , ' \
'Javier Vegas-Regidor, javier.vegas@bsc.es '
handler.Conventions = 'CF-1.6'
handler.creation_date = datetime.now().strftime('%Y-%m-%d(T%H:%M:%SZ)')
handler.experiment_id = experiment.experiment_name
handler.forecast_reference_time = parse_date(self.startdate).strftime('%Y-%m-%d(T%H:%M:%SZ)')
handler.frequency = frequency.frequency
handler.institute_id = experiment.institute
handler.institution = experiment.institute
handler.initialization_method = cmor.initialization_method
handler.initialization_description = cmor.initialization_description
handler.physics_version = cmor.physics_version
handler.physics_description = cmor.physics_description
handler.model_id = experiment.model
handler.associated_model = cmor.associated_model
handler.project_id = self.config.data_convention.upper()
handler.realization = str(self.member + 1)
handler.source = cmor.source
handler.startdate = 'S{0}'.format(self.startdate)
handler.tracking_id = str(uuid.uuid1())
handler.title = "{0} model output prepared for {2} {1}".format(experiment.model, experiment.experiment_name,
self.config.data_convention.upper())
Javier Vegas-Regidor
committed
handler.close()