def loadEnsembleFit(lfit=True, dist=None, dist_args=None, reference=None, target=None,
lglobalScale=False, lrescale=False, lflatten=True, sample_axis=None,
lcrossval=False, ncv=0.2, lbootstrap=False, nbs=100,
WRF_exps=None, CESM_exps=None, WRF_ens=None, CESM_ens=None, variable_list=None,
load_list=None, lproduct='outer', lshort=True, datatype=None, **kwargs):
''' convenience function to load ensemble time-series data and compute associated distribution ensembles '''
if lrescale and not lfit: raise ArgumentError
load_list = [] if load_list is None else load_list[:] # use a copy, since the list may be modified
# load shape ensemble
if datatype.lower() == 'shape':
ensemble = loadShapeEnsemble(variable_list=variable_list, WRF_exps=WRF_exps, CESM_exps=CESM_exps,
WRF_ens=WRF_ens, CESM_ens=CESM_ens, load_list=load_list, lproduct=lproduct, **kwargs)
if datatype.lower() == 'station':
ensemble = loadStationEnsemble(variable_list=variable_list, WRF_exps=WRF_exps, CESM_exps=CESM_exps,
WRF_ens=WRF_ens, CESM_ens=CESM_ens, load_list=load_list, lproduct=lproduct, **kwargs)
# N.B.: kwargs are first passed on to loadShapeEnsemble/loadStationEnsemble and then to loadEnsembleTS
# generate matching datasets with fitted distributions
fitens, sclens = addDistFit(ensemble=ensemble, lfit=lfit, dist=dist, dist_args=dist_args, lflatten=lflatten,
lrescale=lrescale, reference=reference, target=target, lglobalScale=lglobalScale,
lbootstrap=lbootstrap, nbs=nbs, lcrossval=lcrossval, ncv=ncv,
sample_axis=sample_axis, load_list=load_list, lproduct=lproduct, **kwargs)
# N.B.: kwargs are mainly needed to infer the expanded shape of the ensemble list
# return ensembles (will be wrapped in a list, if BatchLoad is used)
if lshort:
if lfit and lrescale: return ensemble, fitens, sclens
elif lfit: return ensemble, fitens
else: return ensemble
else:
return ensemble, fitens, sclens
def loadStationEnsemble(variable_list=None, **kwargs):
''' wrapper for clim.load.loadStationEnsemble that sets experiment and variable lists '''
if variable_list is None: variable_list = variables_rc
return clim_load.loadStationEnsemble(variable_list=variable_list,
WRF_exps=WRF_exps,
CESM_exps=CESM_exps,
WRF_ens=WRF_ens,
CESM_ens=CESM_ens,
**kwargs)
def loadStationEnsemble(variable_list=None, **kwargs):
""" wrapper for clim.load.loadStationEnsemble that sets experiment and variable lists """
if variable_list is None:
variable_list = variables_rc
return clim_load.loadStationEnsemble(
variable_list=variable_list,
WRF_exps=WRF_exps,
CESM_exps=CESM_exps,
WRF_ens=WRF_ens,
CESM_ens=CESM_ens,
**kwargs
)
print fitens[0][0].MaxPrecip_1d.atts.sample_axis; print ''
print fitens[0][1] if fitens is not None else fitens ; print ''
print sclens[0] if sclens is not None else sclens ; print ''
assert len(stnens) == len(seasons)
print stnens[0][0]
print stnens[0][1].MaxPrecip_1d.mean()
elif test == 'rescaling':
# some settings for tests
exps = ['max-val','max-all']; exps = [exps_rc[exp].exps for exp in exps]
provs = ('BC',); seasons = 'summer'
lflatten = True; lfit = True
# rescale
stnens,fitens,scalens = loadStationEnsemble(names=exps, provs=provs, seasons=seasons, varlist='precip',
lrescale=True, reference='Observations', target='max-ens',
filetypes='hydro', domain=None, lfit=lfit, lflatten=lflatten,
lbootstrap=False, nbs=10,
variable_list=variables_rc, default_constraints=constraints_rc,
WRF_exps=WRF_exps, CESM_exps=None, WRF_ens=ensembles, CESM_ens=None,
load_list=['names'], lproduct='outer')
# special rescaling for projection plots
scalens = [rescaleDistributions(ens, reference=fitens[0]['Observations'], target='max-ens') for ens in fitens]
# print diagnostics
print fitens[0]; print ''
print scalens[0]; print ''
print scalens[1][1]
def loadStationEnsemble(variable_list=variables_rc, stationtype=default_stationtype, **kwargs):
''' wrapper for clim.load.loadStationEnsemble that sets experiment and variable lists '''
return clim_load.loadStationEnsemble(variable_list=variable_list, stationtype=stationtype, WRF_exps=WRF_exps,
CESM_exps=CESM_exps, WRF_ens=WRF_ens, CESM_ens=CESM_ens, obs_ts='EC', **kwargs)
def loadEnsembleFit(lfit=True,
dist=None,
dist_args=None,
reference=None,
target=None,
lglobalScale=False,
lrescale=False,
lflatten=True,
sample_axis=None,
lcrossval=False,
ncv=0.2,
lbootstrap=False,
nbs=100,
WRF_exps=None,
CESM_exps=None,
WRF_ens=None,
CESM_ens=None,
variable_list=None,
load_list=None,
lproduct='outer',
lshort=True,
datatype=None,
**kwargs):
''' convenience function to load ensemble time-series data and compute associated distribution ensembles '''
if lrescale and not lfit: raise ArgumentError
load_list = [] if load_list is None else load_list[:] # use a copy, since the list may be modified
# load shape ensemble
if datatype.lower() == 'shape':
ensemble = loadShapeEnsemble(variable_list=variable_list,
WRF_exps=WRF_exps,
CESM_exps=CESM_exps,
WRF_ens=WRF_ens,
CESM_ens=CESM_ens,
load_list=load_list,
lproduct=lproduct,
**kwargs)
if datatype.lower() == 'station':
ensemble = loadStationEnsemble(variable_list=variable_list,
WRF_exps=WRF_exps,
CESM_exps=CESM_exps,
WRF_ens=WRF_ens,
CESM_ens=CESM_ens,
load_list=load_list,
lproduct=lproduct,
**kwargs)
# N.B.: kwargs are first passed on to loadShapeEnsemble/loadStationEnsemble and then to loadEnsembleTS
# generate matching datasets with fitted distributions
fitens, sclens = addDistFit(ensemble=ensemble,
lfit=lfit,
dist=dist,
dist_args=dist_args,
lflatten=lflatten,
lrescale=lrescale,
reference=reference,
target=target,
lglobalScale=lglobalScale,
lbootstrap=lbootstrap,
nbs=nbs,
lcrossval=lcrossval,
ncv=ncv,
sample_axis=sample_axis,
load_list=load_list,
lproduct=lproduct,
**kwargs)
# N.B.: kwargs are mainly needed to infer the expanded shape of the ensemble list
# return ensembles (will be wrapped in a list, if BatchLoad is used)
if lshort:
if lfit and lrescale: return ensemble, fitens, sclens
elif lfit: return ensemble, fitens
else: return ensemble
else:
return ensemble, fitens, sclens
seasons = 'summer'
lflatten = True
lfit = True
# rescale
stnens, fitens, scalens = loadStationEnsemble(
names=exps,
provs=provs,
seasons=seasons,
varlist='precip',
lrescale=True,
reference='Observations',
target='max-ens',
filetypes='hydro',
domain=None,
lfit=lfit,
lflatten=lflatten,
lbootstrap=False,
nbs=10,
variable_list=variables_rc,
default_constraints=constraints_rc,
WRF_exps=WRF_exps,
CESM_exps=None,
WRF_ens=ensembles,
CESM_ens=None,
load_list=['names'],
lproduct='outer')
# special rescaling for projection plots
scalens = [
rescaleDistributions(ens,
reference=fitens[0]['Observations'],
