class BiasCorrecter(object):
def __init__(self, dt, mp, cr):
self.transformer = TransformerWrapper(cr)
self.retrender = RetrenderWrapper(dt, mp)
self.detrender = DetrenderWrapper(dt, mp)
self.unrealizable = mp == 'false' and cr == 'mean-scale'
def correct(self, sim, obs, tsim, tobs):
""" Corrects simulated sequence sim using observations obs,
with tsim and tobs arrays of years """
tmin = max([tsim[0], tobs[0]])
tmax = min([tsim[-1], tobs[-1]])
odT, T = self.detrender.detrend(obs) # detrend observations
sdT, _ = self.detrender.detrend(sim) # detrend simulations
dT = masked_array(zeros(len(tsim)), mask=ones(len(tsim)))
rT = dT.copy()
if not self.unrealizable and not sdT.mask.all() and not odT.mask.all():
odT_c = self.__toverlap(odT, tobs, tmin,
tmax) # transform distribution
sdT_c = self.__toverlap(sdT, tsim, tmin, tmax)
dT = self.transformer.transform(sdT, sdT_c, odT_c)
obs_c = self.__toverlap(obs, tobs, tmin, tmax) # combine
T_c = self.__toverlap(T, tobs, tmin, tmax)
dT_c = self.__toverlap(dT, tsim, tmin, tmax)
rT_c = self.retrender.retrend(T_c, dT_c, obs_c)
rT[logical_and(tsim >= tmin, tsim <= tmax)] = rT_c
return dT, rT
def __toverlap(self, x, tx, tmin, tmax):
return x[logical_and(tx >= tmin, tx <= tmax)]
class BiasCorrecter(object):
def __init__(self, dt, mp, cr):
self.transformer = TransformerWrapper(cr)
self.retrender = RetrenderWrapper(dt, mp)
self.detrender = DetrenderWrapper(dt, mp)
self.unrealizable = mp == 'false' and cr == 'mean-scale'
def correct(self, sim, obs, tsim, tobs):
""" Corrects simulated sequence sim using observations obs,
with tsim and tobs arrays of years """
tmin = max([tsim[0], tobs[0]])
tmax = min([tsim[-1], tobs[-1]])
odT, T = self.detrender.detrend(obs) # detrend observations
sdT, _ = self.detrender.detrend(sim) # detrend simulations
dT = masked_array(zeros(len(tsim)), mask = ones(len(tsim)))
rT = dT.copy()
if not self.unrealizable and not sdT.mask.all() and not odT.mask.all():
odT_c = self.__toverlap(odT, tobs, tmin, tmax) # transform distribution
sdT_c = self.__toverlap(sdT, tsim, tmin, tmax)
dT = self.transformer.transform(sdT, sdT_c, odT_c)
obs_c = self.__toverlap(obs, tobs, tmin, tmax) # combine
T_c = self.__toverlap(T, tobs, tmin, tmax)
dT_c = self.__toverlap(dT, tsim, tmin, tmax)
rT_c = self.retrender.retrend(T_c, dT_c, obs_c)
rT[logical_and(tsim >= tmin, tsim <= tmax)] = rT_c
return dT, rT
def __toverlap(self, x, tx, tmin, tmax):
return x[logical_and(tx >= tmin, tx <= tmax)]
def __init__(self, dt, mp, cr):
self.transformer = TransformerWrapper(cr)
self.retrender = RetrenderWrapper(dt, mp)
self.detrender = DetrenderWrapper(dt, mp)
self.unrealizable = mp == 'false' and cr == 'mean-scale'
def __init__(self, dt, mp, cr):
self.transformer = TransformerWrapper(cr)
self.retrender = RetrenderWrapper(dt, mp)
self.detrender = DetrenderWrapper(dt, mp)
self.unrealizable = mp == 'false' and cr == 'mean-scale'
