def cs_contrib(a,b,bins,subsamp=10):
from metobs.generic import get_dims
'''
a,b: raw time series (not perturbations)
Compute the contribution of coherent structures to the total flux of a'b'.
bins: list of arrays that contain N.arange(start_index,stop_index) for CS events
the list is returned from metobs.vis.wavelet_plt() as the first member.
subsamp: number of samples in original time series for each sample in subsampled
timeseres (if timeseries was subsampled to calculate wavelet).
'''
sensors,obs,blks=get_dims(a,b)
a=a.reshape((sensors,obs,blks))
b=a.reshape((sensors,obs,blks))
cohf_n=[]
tcoh=[]
ap=get_pert(a)
bp=get_pert(b)
for i in range(0,len(bins[0])):
tcoh.append(len(bins[0][i])*subsamp)
cohf_n.append(tcoh[i]*N.average(
ap[0,bins[0][i][0]*subsamp:bins[0][i][-1]*subsamp+subsamp-1,0]*
bp[0,bins[0][i][0]*subsamp:bins[0][i][-1]*subsamp+subsamp-1,0]))
F_coh=N.array(cohf_n).squeeze().sum()/(a.shape[1]*turb_covar(a[0,:,0],b[0,:,0]))
TE=F_coh/((1./a.shape[1])*N.array(tcoh).sum())
return (F_coh,TE,tcoh)
def tke(u, v, w):
'''
Compute the turbulence kinetic energy from the time series of the velocity \
components u,v, and w.
'''
ax = 1
up = get_pert(u)
vp = get_pert(v)
wp = get_pert(w)
tke = N.power(N.average(N.power(up,2),axis=ax)+\
N.average(N.power(vp,2),axis=ax)+\
N.average(N.power(wp,2),axis=ax),0.5)
return tke
def tke(u,v,w):
'''
Compute the turbulence kinetic energy from the time series of the velocity \
components u,v, and w.
'''
ax=1
up = get_pert(u)
vp = get_pert(v)
wp = get_pert(w)
tke = N.power(N.average(N.power(up,2),axis=ax)+\
N.average(N.power(vp,2),axis=ax)+\
N.average(N.power(wp,2),axis=ax),0.5)
return tke
def kurtosis(a):
'''
Compute the kurtosis of the timeseries a
'''
ax = 1
ap = get_pert(a)
K = N.average(N.power(ap, 4), axis=ax) / N.power(N.std(ap), 4)
return K
def skewness(a):
'''
Compute the skewness of the timeseries a
'''
ax = 1
ap = get_pert(a)
S = N.average(N.power(ap, 3), axis=ax) / N.power(N.std(ap), 3)
return S
def kurtosis(a):
'''
Compute the kurtosis of the timeseries a
'''
ax=1
ap = get_pert(a)
K = N.average(N.power(ap,4),axis=ax)/N.power(N.std(ap),4)
return K
def skewness(a):
'''
Compute the skewness of the timeseries a
'''
ax = 1
ap = get_pert(a)
S = N.average(N.power(ap,3),axis=ax)/N.power(N.std(ap),3)
return S
