def model_files(multi):
"""
Computes respective metric and returns single model files and multi-model cube at different aggregations
"""
#this is necessary to make multiprocessing an option
variable = multi[0]
scenario = multi[1]
bc_and_resolution = multi[2]
inpath = multi[3]
outpath = multi[4]
season = multi[5]
metric = multi[6]
region = multi[7]
overwrite = multi[8]
###region box
box = region[2]
xmin = box[0]
xmax = box[1]
ymin = box[2]
ymax = box[3]
for sc, bc, seas, var in itertools.product(scenario, bc_and_resolution, season, variable):
if metric == 'pet' and var == 'multivars':
var = 'rsds'
files_good, modelID = load_file_names(inpath, var, sc, bc)
if files_good == []:
print inpath, sc, bc, seas, var
print 'No files found. Check your input directory!!!'
continue
#sys.exit('No files found. Check your input directory!!!')
out = outpath + os.sep + bc
calc_to_call = getattr(calc, metric) # calls the metric calc function from calc.py
file_searcher = out + os.sep + str(metric) + '_' +str(var) + '_' + str(bc) + \
'_' + str(sc) + '_' + str(seas) +'_' + str(region[0])
for file, nme in zip(files_good, modelID):
cubeout=None
print('Doing:', nme, metric, var)
if metric == 'pet' and var == 'rsds':
print(nme, 'in')
# This is a different case because PET needs multiple variables, rather than one
# NB: We use rsds, because generally, less models have this variable as opposed to tasmin or tasmax
tasmin = atlas_utils.load_data(file.replace(var, 'tasmin'), xmin, xmax, ymin, ymax)
tasmax = atlas_utils.load_data(file.replace(var, 'tasmax'), xmin, xmax, ymin, ymax)
rsds = atlas_utils.load_data(file.replace(var, 'rsds'), xmin, xmax, ymin, ymax)
cubeout = iris.cube.CubeList([tasmin, tasmax, rsds])
file_searcher = file_searcher.replace('rsds', 'multivars')
print cubeout
nc_file = file_searcher + '_' + str(nme) + '_singleModel_tseries.nc'
if not os.path.isfile(nc_file) or (overwrite == 'Yes'):
print 'nc_file: ' + nc_file
print 'Load file: '+ file
if not cubeout:
print(nme, 'no cubeout yet, loading file')
print 'newcube'
cubeout = atlas_utils.load_data(file, xmin, xmax, ymin, ymax)
calc_to_call(cubeout, seas, nc_file) # saves single model netcdf
print '#######################################'
print 'Saving data for: '
print nme, metric, var, seas, region[0]
print '#######################################'
# runs big_cube for all available aggregations
for agg in cnst.AGGREGATION:
big_cube(file_searcher, agg)
def wfdei(multi):
# this is necessary to make multiprocessing an option
variable = multi[0]
outpath = multi[1]
season = multi[2]
metric = multi[3]
region = multi[4]
overwrite = multi[5]
for seas, var in itertools.product(season, variable):
bc = 'WFDEI'
sc = 'historical'
if metric == 'pet' and var == 'multivars':
var = 'rsds'
filepath = cnst.DATADIR + os.sep + bc + os.sep + str(var) + '_daily*.nc'
file = glob.glob(filepath)
if (len(file) != 1):
print('No or too many WFDEI files found, path problem')
return
file = file[0]
box = region[2]
xmin = box[0]
xmax = box[1]
ymin = box[2]
ymax = box[3]
if metric == 'pet' and var == 'multivars':
var = 'rsds'
out = outpath + os.sep + bc
calc_to_call = getattr(calc, metric) # calls the metric calc function from calc.py
file_searcher = out + os.sep + str(metric) + '_' + str(var) + '_' + str(bc) + \
'_' + str(sc) + '_' + str(seas) + '_' + str(region[0])
cubeout = None
# Check if we have any missing files for all aggregation types, if so, run the metric calculation again
# Note: the calc functions run for 2 or 3 aggregation methods
if metric == 'pet' and var == 'rsds':
# This is a different case because PET needs multiple variables, rather than one
# NB: We use rsds, because generally, less models have this variable as opposed to tasmin or tasmax
tasmin = atlas_utils.load_data(file.replace(var, 'tasmin'), xmin, xmax, ymin, ymax)
tasmax = atlas_utils.load_data(file.replace(var, 'tasmax'), xmin, xmax, ymin, ymax)
rsds = atlas_utils.load_data(file.replace(var, 'rsds'), xmin, xmax, ymin, ymax)
cubeout = iris.cube.CubeList([tasmin, tasmax, rsds])
file_searcher = file_searcher.replace('rsds', 'multivars')
print cubeout
nc_file = file_searcher + '_WFDEI_tseries.nc'
if not os.path.isfile(nc_file) or (overwrite == 'Yes'):
print 'nc_file: ' + nc_file
print 'Load file: ' + file
if not cubeout:
print 'newcube'
cubeout = atlas_utils.load_data(file, xmin, xmax, ymin, ymax)
calc_to_call(cubeout, seas, nc_file) # saves single model netcdf
print '#######################################'
print 'Saving data for: '
print metric, var, seas, region[0]
print '#######################################'
import atlas_utils, calc, constants as cnst
#### We only need these seven strings to create clean output files.
# Please give the correct variable and scenario information, the rest is less important
metric = 'metricname' # identifies the metric you're calculating
var = 'tas' # the CMIP5 variable name
bc = 'dummy' # we use this, you need to provide a random string (it's a bias correction flag)
sc = 'rcp45' # scenario, could form a loop from cnst.SCENARIO
seas = 'jas' # season
reg = 'WestAfrica' # regional flag
aggregation = cnst.AGGREGATION[0]
lonlatbox = [-18, 25, 4, 25]
cmip5_file = '/users/global/cornkle/CMIP/CMIP5_Africa/BC_mdlgrid/GFDL-CM3/rcp45/' + var + '_WFDEI_1979-2013_mdlgrid_day_GFDL-CM3_west-africa_rcp45_r1i1p1_full.nc'
outfilename = str(metric) + '_' +str(var) + '_' + str(bc) + \
'_' + str(sc) + '_' + str(seas) +'_' + str(reg) +'_singleModel_' + aggregation +'.nc'
outpath = '/users/global/cornkle/test/' + outfilename
cube = atlas_utils.load_data(cmip5_file, lonlatbox[0], lonlatbox[1],
lonlatbox[2], lonlatbox[3])
calc.annualHotDays(cube, 'jas', outpath)