name_f = "/T%d_spr_err.png" % T
f_name_f = str(figsdir + name_f)
plt.savefig(f_name_f)
print ' '
print ' *** %s at time level %d saved to %s' % (name_f, T, figsdir)
##################################################################
### 3 panel subplot: CRPS of 3 vars for fc and an
##################################################################
CRPS_fc = np.empty((Neq, Nk_fc))
CRPS_an = np.empty((Neq, Nk_fc))
for ii in h_mask:
CRPS_fc[0, ii] = crps_calc(X[ii, :, T], X_tr[ii, 0, T])
CRPS_fc[1, ii] = crps_calc(X[ii + Nk_fc, :, T], X_tr[ii + Nk_fc, 0, T])
CRPS_fc[2, ii] = crps_calc(X[ii + 2 * Nk_fc, :, T], X_tr[ii + 2 * Nk_fc, 0,
T])
CRPS_an[0, ii] = crps_calc(Xan[ii, :, T], X_tr[ii, 0, T])
CRPS_an[1, ii] = crps_calc(Xan[ii + Nk_fc, :, T], X_tr[ii + Nk_fc, 0, T])
CRPS_an[2, ii] = crps_calc(Xan[ii + 2 * Nk_fc, :, T], X_tr[ii + 2 * Nk_fc,
0, T])
lw = 1. # linewidth
axlim0 = np.max(CRPS_fc[0, :])
axlim1 = np.max(CRPS_fc[1, :])
axlim2 = np.max(CRPS_fc[2, :])
ft = 16
xl = 0.65
def ave_stats(i, j, k, dirname):
'''
INPUT:
## e.g. if i,j,k... etc a coming from outer loop:
i=1
j=0
k=0
##
##
dirname = '/addinfv7_4dres'
##
OUTPUT:
spr, rmse, crps, OI for fc and an
'''
# LOAD DATA FROM GIVEN DIRECTORY
cwd = os.getcwd()
dirn = str(cwd+dirname+dirname+str(i+1)+str(j+1)+str(k+1))
if os.path.exists(dirn):
print ' '
print 'Path: '
print dirn
print ' exists... calculating stats...'
print ' '
# parameters for outer loop
o_d = [20,40]
loc = [1.5, 2.5, 3.5, 0.]
#inf = [1.1, 1.25, 1.5, 1.75]
inf = [1.01, 1.05, 1.1]
# LOAD DATA FROM GIVEN DIRECTORY
X = np.load(str(dirn+'/X_array.npy')) # fc ensembles
X_tr = np.load(str(dirn+'/X_tr_array.npy')) # truth
Xan = np.load(str(dirn+'/Xan_array.npy')) # an ensembles
Y_obs = np.load(str(dirn+'/Y_obs_array.npy')) # obs ensembles
OI = np.load(str(dirn+'/OI.npy')) # obs ensembles
#np.set_printoptions(formatter={'float': '{: 0.3f}'.format})
#print 'X_array shape (n_d,n_ens,T) : ', np.shape(X)
#print 'X_tr_array shape (n_d,1,T) : ', np.shape(X_tr)
#print 'Xan_array shape (n_d,n_ens,T) : ', np.shape(Xan)
#print 'Y_obs_array shape (p,n_ens,T) : ', np.shape(Y_obs)
#print 'OI shape (Neq + 1,T) : ', np.shape(OI)
##################################################################
Neq = np.shape(OI)[0] - 1
n_d = np.shape(X)[0]
Nk_fc = n_d/Neq
Kk_fc = 1./Nk_fc
n_ens = np.shape(X)[1]
n_obs = np.shape(Y_obs)[0]
obs_dens = n_d/n_obs
t_an = np.shape(X)[2]
time_vec = range(0,t_an)
# masks for locating model variables in state vector
h_mask = range(0,Nk_fc)
hu_mask = range(Nk_fc,2*Nk_fc)
hr_mask = range(2*Nk_fc,3*Nk_fc)
# masks for locating obs locations
row_vec = range(obs_dens,n_d+1,obs_dens)
obs_mask = np.array(row_vec[0:n_obs/Neq])-1
h_obs_mask = range(0,n_obs/Neq)
hu_obs_mask = range(n_obs/Neq,2*n_obs/Neq)
hr_obs_mask = range(2*n_obs/Neq,3*n_obs/Neq)
nz_index = np.where(OI[0,:])
if (len(nz_index[0]) < t_an-15):
print 'Runs crashed before Tmax...'
return float('nan'), float('nan'), float('nan'), float('nan'), float('nan'), float('nan'), float('nan'), float('nan'), float('nan')
else:
##################################################################
print 'Runs completed... '
# for means and deviations
Xbar = np.empty(np.shape(X))
Xdev = np.empty(np.shape(X))
Xanbar = np.empty(np.shape(X))
Xandev = np.empty(np.shape(X))
Xdev_tr = np.empty(np.shape(X))
Xandev_tr = np.empty(np.shape(X))
# for errs as at each assim time
rmse_fc = np.empty((Neq,len(time_vec)))
rmse_an = np.empty((Neq,len(time_vec)))
spr_fc = np.empty((Neq,len(time_vec)))
spr_an = np.empty((Neq,len(time_vec)))
ame_fc = np.empty((Neq,len(time_vec)))
ame_an = np.empty((Neq,len(time_vec)))
crps_fc = np.empty((Neq,len(time_vec)))
crps_an = np.empty((Neq,len(time_vec)))
ONE = np.ones([n_ens,n_ens])
ONE = ONE/n_ens # NxN array with elements equal to 1/N
print ' *** Calculating errors from ', dirn
for T in time_vec[1:]:
plt.clf() # clear figs from previous loop
Xbar[:,:,T] = np.dot(X[:,:,T],ONE) # fc mean
Xdev[:,:,T] = X[:,:,T] - Xbar[:,:,T] # fc deviations from mean
Xdev_tr[:,:,T] = X[:,:,T] - X_tr[:,:,T] # fc deviations from truth
Xanbar[:,:,T] = np.dot(Xan[:,:,T],ONE) # an mean
Xandev[:,:,T] = Xan[:,:,T] - Xanbar[:,:,T] # an deviations from mean
Xandev_tr[:,:,T] = Xan[:,:,T] - X_tr[:,:,T] # an deviations from truth
##################################################################
### ERRORS ####
##################################################################
# FORECAST: mean error
fc_err = Xbar[:,0,T] - X_tr[:,0,T] # fc_err = ens. mean - truth
fc_err2 = fc_err**2
# ANALYSIS: mean error
an_err = Xanbar[:,0,T] - X_tr[:,0,T] # an_err = analysis ens. mean - truth
an_err2 = an_err**2
# FORECAST: cov matrix for spread...
Pf = np.dot(Xdev[:,:,T],np.transpose(Xdev[:,:,T]))
Pf = Pf/(n_ens - 1) # fc covariance matrix
var_fc = np.diag(Pf)
# ... and rmse
Pf_tr = np.dot(Xdev_tr[:,:,T],np.transpose(Xdev_tr[:,:,T]))
Pf_tr = Pf_tr/(n_ens - 1) # fc covariance matrix w.r.t. truth
var_fct = np.diag(Pf_tr)
# ANALYSIS: cov matrix for spread...
Pa = np.dot(Xandev[:,:,T],np.transpose(Xandev[:,:,T]))
Pa = Pa/(n_ens - 1) # analysis covariance matrix
var_an = np.diag(Pa)
# ... and rmse
Pa_tr = np.dot(Xandev_tr[:,:,T],np.transpose(Xandev_tr[:,:,T]))
Pa_tr = Pa_tr/(n_ens - 1) # fc covariance matrix w.r.t truth
var_ant = np.diag(Pa_tr)
##################################################################
### CRPS ####
##################################################################
CRPS_fc = np.empty((Neq,Nk_fc))
CRPS_an = np.empty((Neq,Nk_fc))
for ii in h_mask:
CRPS_fc[0,ii] = crps_calc(X[ii,:,T],X_tr[ii,0,T])
CRPS_fc[1,ii] = crps_calc(X[ii+Nk_fc,:,T],X_tr[ii+Nk_fc,0,T])
CRPS_fc[2,ii] = crps_calc(X[ii+2*Nk_fc,:,T],X_tr[ii+2*Nk_fc,0,T])
CRPS_an[0,ii] = crps_calc(Xan[ii,:,T],X_tr[ii,0,T])
CRPS_an[1,ii] = crps_calc(Xan[ii+Nk_fc,:,T],X_tr[ii+Nk_fc,0,T])
CRPS_an[2,ii] = crps_calc(Xan[ii+2*Nk_fc,:,T],X_tr[ii+2*Nk_fc,0,T])
#################################################################
# domain-averaged errors
ame_an[0,T] = np.mean(np.absolute(an_err[h_mask]))
ame_fc[0,T] = np.mean(np.absolute(fc_err[h_mask]))
spr_an[0,T] = np.sqrt(np.mean(var_an[h_mask]))
spr_fc[0,T] = np.sqrt(np.mean(var_fc[h_mask]))
rmse_an[0,T] = np.sqrt(np.mean(an_err2[h_mask]))
rmse_fc[0,T] = np.sqrt(np.mean(fc_err2[h_mask]))
crps_an[0,T] = np.mean(CRPS_an[0,:])
crps_fc[0,T] = np.mean(CRPS_fc[0,:])
ame_an[1,T] = np.mean(np.absolute(an_err[hu_mask]))
ame_fc[1,T] = np.mean(np.absolute(fc_err[hu_mask]))
spr_an[1,T] = np.sqrt(np.mean(var_an[hu_mask]))
spr_fc[1,T] = np.sqrt(np.mean(var_fc[hu_mask]))
rmse_an[1,T] = np.sqrt(np.mean(an_err2[hu_mask]))
rmse_fc[1,T] = np.sqrt(np.mean(fc_err2[hu_mask]))
crps_an[1,T] = np.mean(CRPS_an[1,:])
crps_fc[1,T] = np.mean(CRPS_fc[1,:])
ame_an[2,T] = np.mean(np.absolute(an_err[hr_mask]))
ame_fc[2,T] = np.mean(np.absolute(fc_err[hr_mask]))
spr_an[2,T] = np.sqrt(np.mean(var_an[hr_mask]))
spr_fc[2,T] = np.sqrt(np.mean(var_fc[hr_mask]))
rmse_an[2,T] = np.sqrt(np.mean(an_err2[hr_mask]))
rmse_fc[2,T] = np.sqrt(np.mean(fc_err2[hr_mask]))
crps_an[2,T] = np.mean(CRPS_an[2,:])
crps_fc[2,T] = np.mean(CRPS_fc[2,:])
###########################################################################
spr_fc_ave = spr_fc[:,nz_index].mean(axis=-1)
err_fc_ave = ame_fc[:,nz_index].mean(axis=-1)
rmse_fc_ave = rmse_fc[:,nz_index].mean(axis=-1)
crps_fc_ave = crps_fc[:,nz_index].mean(axis=-1)
spr_an_ave = spr_an[:,nz_index].mean(axis=-1)
err_an_ave = ame_an[:,nz_index].mean(axis=-1)
rmse_an_ave = rmse_an[:,nz_index].mean(axis=-1)
crps_an_ave = crps_an[:,nz_index].mean(axis=-1)
OI_ave = 100*OI[0,nz_index].mean(axis=-1)
spr_fc_ave = spr_fc_ave.mean()
err_fc_ave = err_fc_ave.mean()
rmse_fc_ave = rmse_fc_ave.mean()
crps_fc_ave = crps_fc_ave.mean()
spr_an_ave = spr_an_ave.mean()
err_an_ave = err_an_ave.mean()
rmse_an_ave = rmse_an_ave.mean()
crps_an_ave = crps_an_ave.mean()
print 'spr_fc ave. =', spr_fc_ave
print 'err_fc ave. =', err_fc_ave
print 'rmse_fc ave. =', rmse_fc_ave
print 'crps_fc ave. =', crps_fc_ave
print 'spr_an ave. =', spr_an_ave
print 'err_an ave. =', err_an_ave
print 'rmse_an ave. =', rmse_an_ave
print 'crps_an_ave. =', crps_an_ave
print 'OI ave. =', OI_ave
return spr_fc_ave, err_fc_ave, rmse_fc_ave, crps_fc_ave, spr_an_ave, err_an_ave, rmse_an_ave, crps_an_ave, OI_ave
else:
print ' '
print ' Path:'
print dirn
print 'does not exist.. moving on to next one...'
print ' '
return float('nan'), float('nan'), float('nan'), float('nan'), float('nan'), float('nan'), float('nan'), float('nan'), float('nan')