def run_minor_crossvalidation():
# Initialize OGGM and set up the run parameters
# initialization, select region if wanted and return GDIRs
gdirs = initialization_selection()
# some preprocessing
gdirs = preprocessing(gdirs)
# ----- cross validation loop
# dataframe to store results
xval = pd.DataFrame([], columns=['prcpsf', 'tliq', 'tmelt', 'tgrad',
'std_quot', 'bias', 'rmse', 'core',
'std_oggm', 'std_ref', 'nans'])
xval = calibration(gdirs, xval, major=0)
xval_perglacier, mb_perglacier = minor_xval_statistics(gdirs)
outdict = {'statistic': xval,
'per_glacier': xval_perglacier,
'massbalance': mb_perglacier,
'date_created': datetime.datetime.now().strftime('%Y-%m-%d'),
'oggmversion': mbcfg.PARAMS['oggmversion']}
pout = os.path.join(mbcfg.PATHS['storage_dir'],
'xval_%s_minor.p' % mbcfg.PARAMS['oggmversion'])
pickle.dump(outdict, open(pout, 'wb'))
def run_major_crossvalidation():
# Initialize OGGM and set up the run parameters
if mbcfg.PARAMS['histalp']:
histalp()
# initialization, select region if wanted and return GDIRs
gdirs = initialization_selection()
# some preprocessing
gdirs = preprocessing(gdirs)
# ----- cross validation loop
# dataframe to store results
xval = pd.DataFrame([],
columns=[
'prcpsf', 'tliq', 'tmelt', 'tgrad', 'std_quot',
'bias', 'rmse', 'core'
])
# define each parameter range
prcpsf = np.arange(mbcfg.PARAMS['prcp1'],
mbcfg.PARAMS['prcp2'] + mbcfg.PARAMS['prcp_step'],
mbcfg.PARAMS['prcp_step'])
tliq = np.arange(mbcfg.PARAMS['tliq1'],
mbcfg.PARAMS['tliq2'] + mbcfg.PARAMS['tliq_step'],
mbcfg.PARAMS['tliq_step'])
tmelt = np.arange(mbcfg.PARAMS['tmel1'],
mbcfg.PARAMS['tmel2'] + mbcfg.PARAMS['tmel_step'],
mbcfg.PARAMS['tmel_step'])
tgrad = np.arange(mbcfg.PARAMS['tgra1'],
mbcfg.PARAMS['tgra2'] + mbcfg.PARAMS['tgra_step'],
mbcfg.PARAMS['tgra_step'])
# loop over all
for PR in prcpsf:
cfg.PARAMS['prcp_scaling_factor'] = PR
for TL in tliq:
cfg.PARAMS['temp_all_liq'] = TL
for TM in tmelt:
cfg.PARAMS['temp_melt'] = TM
for TG in tgrad:
cfg.PARAMS['temp_default_gradient'] = TG
# execute calibraion and crossvalidation
log.info('prcpSF={}, Tliq={}, Tmelt={}, Tgrad={}'.format(
PR, TL, TM, TG))
xval = calibration(gdirs, xval, major=1)
outdict = {
'statistic': xval,
'date_created': datetime.datetime.now().strftime('%Y-%m-%d'),
'oggmversion': mbcfg.PARAMS['oggmversion']
}
pout = os.path.join(mbcfg.PATHS['storage_dir'],
'xval_%s_major.p' % mbcfg.PARAMS['oggmversion'])
pickle.dump(outdict, open(pout, 'wb'))