toe2read = ftoe.variables[var + 'ToE2'][:]
nMembers[i] = toe2read.shape[0]
print('- Computing ToE of',model['name'], 'with', nMembers[i], 'members')
nruns1 = nruns + nMembers[i]
toe_a[nruns:nruns1,:,:] = toe2read[:,1,:,:]
toe_p[nruns:nruns1,:,:] = toe2read[:,2,:,:]
toe_i[nruns:nruns1,:,:] = toe2read[:,3,:,:]
# -- Select domain to average for each model
domain = ToEdomainhistvshistNat(model['name'], domain_name)[0]
domain_char = ToEdomainhistvshistNat(model['name'], domain_name)[1]
# -- Average toe and determine median of the model
if domain['Atlantic'] != None:
varToEA[nruns:nruns1] = np.ma.around(averageDom(toe_a[nruns:nruns1,:,:], 3, domain['Atlantic'], lat, density)) + iniyear
medmodelsToEA[i] = np.ma.around(np.ma.median(varToEA[nruns:nruns1]))
print('Median ToE Atlantic:', medmodelsToEA[i])
if domain['Pacific'] != None:
varToEP[nruns:nruns1] = np.ma.around(averageDom(toe_p[nruns:nruns1,:,:], 3, domain['Pacific'], lat, density)) + iniyear
medmodelsToEP[i] = np.ma.around(np.ma.median(varToEP[nruns:nruns1]))
print('Median ToE Pacific:', medmodelsToEP[i])
if domain['Indian'] != None:
varToEI[nruns:nruns1] = np.ma.around(averageDom(toe_i[nruns:nruns1,:,:], 3, domain['Indian'], lat, density)) + iniyear
medmodelsToEI[i] = np.ma.around(np.ma.median(varToEI[nruns:nruns1]))
print('Median ToE Indian:', medmodelsToEI[i])
nruns = nruns1
# Method 2
toe1_i = np.ma.masked_all(len(domains))
# -- Initialize ouput variable
varToE1 = np.ma.masked_all((basinN, len(domains)))
varToE2 = np.ma.masked_all((basinN, len(domains)))
# -- Loop over domains
for j, domain_name in enumerate(domains):
print('- ' + domain_name)
# Select domain to average
domain = ToEdomain1pctCO2vsPiC(model['name'], domain_name)[0]
# Average signal and noise
if domain['Atlantic'] != None:
varsignal_a[:, j] = averageDom(
varCO2[:, 1, :, :] - meanvarpiC[1, :, :], 3,
domain['Atlantic'], lat, density)
if method_noise == 'average_std':
varnoise_a[j] = averageDom(varstd[1, :, :], 2,
domain['Atlantic'], lat, density)
else:
varnoise_a[j] = np.ma.std(averageDom(varpiC[:, 1, :, :], 3,
domain['Atlantic'], lat,
density),
axis=0)
if domain['Pacific'] != None:
varsignal_p[:, j] = averageDom(
varCO2[:, 2, :, :] - meanvarpiC[2, :, :], 3, domain['Pacific'],
lat, density)
if method_noise == 'average_std':
varnoise_p[j] = averageDom(varstd[2, :, :], 2,
varsignal = np.ma.masked_all((timN,nruns))
if method_noise == 'average_histNat':
# Here we read the std of the averaged histNat for all runs, then take the max as our noise
filenoise = 'cmip5.' + model['name'] + '.noise_domains_hist_histNat.std_of_average.nc'
fnoise = open_ncfile(indir_noise + filenoise,'r')
varstd = fnoise.variables[var+'stdhn'][:,ibasin,idomain].squeeze()
varnoise = np.ma.max(varstd)
# Select domain to average
domain = ToEdomainhistvshistNat(model['name'], domain_name)[0]
if method_noise == 'average_std':
# Average noise
if domain[basin_name] != None:
varnoise = averageDom(varstd, 2, domain[basin_name], lat, density)
# Average noise over projection period if we want to use PiControl
if use_piC == True:
if method_noise == 'average_std':
varnoise2 = averageDom(stdvarpiC, 2, domain[basin_name], lat, density)
else:
varnoise2 = np.ma.std(averageDom(varpiC, 3, domain[basin_name], lat, density), axis=0)
# Loop over number of runs
for k in range(nruns):
# Read file
fhrcp = open_ncfile(listruns[k],'r')
# Read var hist + rcp85
varh = fhrcp.variables[var][tstart:tend,ibasin,:,:].squeeze()
varrcp = fhrcp.variables[var][tend:tend+95,ibasin,:,:].squeeze()
# Loop over number of histNat runs
for khn in range(nruns_hn):
print(' . histNat run number', khn)
# Read file
fhn = open_ncfile(listruns_hn[khn], 'r')
# Read var histNat
varhn = fhn.variables[var][tstart:tend, :, :, :]
# Compute std of histNat
varstdhn = np.ma.std(varhn, axis=0)
# Average std
if domain['Atlantic'] != None:
if method_noise == 'average_std':
varnoise_hna[khn,
j] = averageDom(varstdhn[1, :, :], 2,
domain['Atlantic'], lat,
density)
else:
varnoise_hna[khn, j] = np.ma.std(averageDom(
varhn[:, 1, :, :], 3, domain['Atlantic'], lat,
density),
axis=0)
if domain['Pacific'] != None:
if method_noise == 'average_std':
varnoise_hnp[khn,
j] = averageDom(varstdhn[2, :, :], 2,
domain['Pacific'], lat,
density)
else:
varnoise_hnp[khn, j] = np.ma.std(averageDom(
varhn[:, 2, :, :], 3, domain['Pacific'], lat,
toe_a = np.ma.masked_all(len(domains))
toe_p = np.ma.masked_all(len(domains))
toe_i = np.ma.masked_all(len(domains))
# -- Initialize ouput variable
varToE = np.ma.masked_all((basinN, len(domains)))
# -- Loop over domains
for j, domain_name in enumerate(domains):
print('- ' + domain_name)
# Select domain to average
domain = ToEdomain1pctCO2vsPiC(model['name'], domain_name)[0]
# Average signal and noise
if domain['Atlantic'] != None:
varsignal_a[:,j] = averageDom(varCO2[:,1,:,:]-varpiC[:,1,:,:], 3, domain['Atlantic'], lat, density)
if method_noise == 'average_std':
varnoise_a[j] = averageDom(varstd[1,:,:], 2, domain['Atlantic'], lat, density)
else:
varnoise_a[j] = np.ma.std(averageDom(varpiC[:,1,:,:], 3, domain['Atlantic'], lat, density), axis=0)
if domain['Pacific'] != None:
varsignal_p[:,j] = averageDom(varCO2[:,2,:,:]-varpiC[:,2,:,:], 3, domain['Pacific'], lat, density)
if method_noise == 'average_std':
varnoise_p[j] = averageDom(varstd[2,:,:], 2, domain['Pacific'], lat, density)
else:
varnoise_p[j] = np.ma.std(averageDom(varpiC[:,2,:,:], 3, domain['Pacific'], lat, density), axis=0)
if domain['Indian'] != None:
varsignal_i[:,j] = averageDom(varCO2[:,3,:,:]-varpiC[:,3,:,:], 3, domain['Indian'], lat, density)
if method_noise == 'average_std':
varnoise_i[j] = averageDom(varstd[3,:,:], 2, domain['Indian'], lat, density)
else:
toei_a = np.reshape(findToE(varCO2_a-varpiC_a, varams, multStd),(levN,latN))
toei_p = np.reshape(findToE(varCO2_p-varpiC_p, varpms, multStd),(levN,latN))
toei_i = np.reshape(findToE(varCO2_i-varpiC_i, varims, multStd),(levN,latN))
# -- Save
toe_a[i,:,:] = toei_a
toe_p[i,:,:] = toei_p
toe_i[i,:,:] = toei_i
# -- Select domain to average for each model
domain = ToEdomain1pctCO2vsPiC(model['name'], domain_name)[0]
domain_char = ToEdomain1pctCO2vsPiC(model['name'], domain_name)[1]
# -- Average toe
if domain['Atlantic'] != None:
varToEA[i] = np.ma.around(averageDom(toe_a[i,:,:], 2, domain['Atlantic'], lat, density))
print('ToE Atlantic:',varToEA[i])
if domain['Pacific'] != None:
varToEP[i] = np.ma.around(averageDom(toe_p[i,:,:], 2, domain['Pacific'], lat, density))
print('ToE Pacific:',varToEP[i])
if domain['Indian'] != None:
varToEI[i] = np.ma.around(averageDom(toe_i[i,:,:], 2, domain['Indian'], lat, density))
print('ToE Indian:',varToEI[i])
if method == 'average_signal':
for i, model in enumerate(models):
print('- Reading', model['name'])
varnoise_i = varstd[3,:]
# Loop over 5 domains
for j, domain_name in enumerate(domains):
print('- ', j, domains[j])
# Select domain to average
domain = ToEdomainrcp85vshistNat(model['name'], domain_name)[0]
domain_char = ToEdomainrcp85vshistNat(model['name'], domain_name)[1]
if use_piC == False :
if method_noise == 'average_std':
# Average noise when using histNat
if domain['Atlantic'] != None:
varnoise_a[j] = averageDom(varstd_a, 2, domain['Atlantic'], lat, density)
if domain['Pacific'] != None:
varnoise_p[j] = averageDom(varstd_p, 2, domain['Pacific'], lat, density)
if domain['Indian'] != None:
varnoise_i[j] = averageDom(varstd_i, 2, domain['Indian'], lat, density)
else :
# Average noise when using PiControl
if domain['Atlantic'] != None:
if method_noise == 'average_std':
varnoise_a[j] = averageDom(stdvarpiC_a, 2, domain['Atlantic'], lat, density)
# else :
# varnoise_a[j] = np.ma.std(averageDom(varpiC_a, 3, domain['Atlantic'], lat, density), axis=0)
if domain['Pacific'] != None:
if method_noise == 'average_std':
varnoise_p[j] = averageDom(stdvarpiC_p, 2, domain['Pacific'], lat, density)
# else:
varnoise_i = np.ma.max(varstd[:, 3, :], axis=0)
# Loop over 5 domains
for j, domain_name in enumerate(domains):
print('- ', j, domains[j])
# Select domain to average
domain = ToEdomainhistvshistNat(model['name'], domain_name)[0]
domain_char = ToEdomainhistvshistNat(model['name'],
domain_name)[1]
if method_noise == 'average_std':
# Average noise
if domain['Atlantic'] != None:
varnoise_a[j] = averageDom(varstd_a, 2,
domain['Atlantic'], lat,
density)
if domain['Pacific'] != None:
varnoise_p[j] = averageDom(varstd_p, 2,
domain['Pacific'], lat,
density)
if domain['Indian'] != None:
varnoise_i[j] = averageDom(varstd_i, 2,
domain['Indian'], lat,
density)
# Average noise over projection period if we want to use PiControl
if use_piC == True:
if domain['Atlantic'] != None:
if method_noise == 'average_std':
varnoise2_a[j] = averageDom(
# Read main variables
lat = fCO2mme.variables['latitude'][:]
time = fCO2mme.variables['time'][:]
varname = defVarmme('salinity'); v = 'S'
density = fCO2mme.variables['lev'][:]
var = varname['var_zonal_w/bowl']
models = defModelsCO2piC()
var_change = np.ma.masked_all((len(models)+1, len(time)))
domain_name = 'Southern ST'
box = [-38,-18,25,26.4]
varCO2mme = fCO2mme.variables[var][:,2,:,:].squeeze()
varpiCmme = fpiCmme.variables[var][-20:,2,:,:].squeeze()
varpiCmme = np.ma.average(varpiCmme, axis=0)
var_change[-1,:] = averageDom(varCO2mme, 3, box, lat, density) - averageDom(varpiCmme, 2, box, lat, density)
# ----- Individual runs -----
indir = '/data/ericglod/Density_binning/Prod_density_april15/'
for imodel in range(len(models)) :
model = models[imodel]
file_CO2 = 'mme_1pctCO2/cmip5.' + model['name'] + '.1pctCO2.ensm.an.ocn.Omon.density.ver-' + model['file_end_CO2'] + '_zon2D.nc'
file_piC = 'mme_piControl/cmip5.' + model['name'] + '.piControl.ensm.an.ocn.Omon.density.ver-' + model['file_end_piC'] + '_zon2D.nc'
fCO2 = open_ncfile(indir + file_CO2, 'r')
fpiC = open_ncfile(indir + file_piC, 'r')
# ----- Variables -----
domain = ToEdomain1pctCO2vsPiC(model['name'], domain_name)[0]
toei_i = np.reshape(findToE(varCO2_i - varpiC_i, varims, multStd),
(levN, latN))
# -- Save
toe_a[i, :, :] = toei_a
toe_p[i, :, :] = toei_p
toe_i[i, :, :] = toei_i
# -- Select domain to average for each model
domain = ToEdomain1pctCO2vsPiC(model['name'], domain_name)[0]
domain_char = ToEdomain1pctCO2vsPiC(model['name'], domain_name)[1]
# -- Average toe
if domain['Atlantic'] != None:
varToEA[i] = np.ma.around(
averageDom(toe_a[i, :, :], 2, domain['Atlantic'], lat,
density))
print('ToE Atlantic:', varToEA[i])
if domain['Pacific'] != None:
varToEP[i] = np.ma.around(
averageDom(toe_p[i, :, :], 2, domain['Pacific'], lat, density))
print('ToE Pacific:', varToEP[i])
if domain['Indian'] != None:
varToEI[i] = np.ma.around(
averageDom(toe_i[i, :, :], 2, domain['Indian'], lat, density))
print('ToE Indian:', varToEI[i])
if method == 'average_signal':
for i, model in enumerate(models):
print('- Reading', model['name'])
varnoise_a = np.ma.max(varstd[:,1,:], axis=0)
varnoise_p = np.ma.max(varstd[:,2,:], axis=0)
varnoise_i = np.ma.max(varstd[:,3,:], axis=0)
# Loop over 5 domains
for j, domain_name in enumerate(domains):
print('- ', j, domains[j])
# Select domain to average
domain = ToEdomainhistvshistNat(model['name'], domain_name)[0]
domain_char = ToEdomainhistvshistNat(model['name'], domain_name)[1]
if method_noise == 'average_std':
# Average noise
if domain['Atlantic'] != None:
varnoise_a[j] = averageDom(varstd_a, 2, domain['Atlantic'], lat, density)
if domain['Pacific'] != None:
varnoise_p[j] = averageDom(varstd_p, 2, domain['Pacific'], lat, density)
if domain['Indian'] != None:
varnoise_i[j] = averageDom(varstd_i, 2, domain['Indian'], lat, density)
# Average noise over projection period if we want to use PiControl
if use_piC == True:
if domain['Atlantic'] != None:
if method_noise == 'average_std':
varnoise2_a[j] = averageDom(stdvarpiC_a, 2, domain['Atlantic'], lat, density)
else:
varnoise2_a[j] = np.ma.std(averageDom(varpiC_a, 3, domain['Atlantic'], lat, density), axis=0)
if domain['Pacific'] != None:
if method_noise == 'average_std':
varnoise2_p[j] = averageDom(stdvarpiC_p, 2, domain['Pacific'], lat, density)
domain_char = ToEdomainhistvshistNat(model['name'], domain_name)[1]
# Loop over number of histNat runs
for khn in range(nruns_hn):
print(' . histNat run number', khn)
# Read file
fhn = open_ncfile(listruns_hn[khn],'r')
# Read var histNat
varhn = fhn.variables[var][tstart:tend,:,:,:]
# Compute std of histNat
varstdhn = np.ma.std(varhn, axis=0)
# Average std
if domain['Atlantic'] != None:
if method_noise == 'average_std':
varnoise_hna[khn,j] = averageDom(varstdhn[1,:,:], 2, domain['Atlantic'], lat, density)
else:
varnoise_hna[khn,j] = np.ma.std(averageDom(varhn[:,1,:,:], 3, domain['Atlantic'], lat, density), axis=0)
if domain['Pacific'] != None:
if method_noise == 'average_std':
varnoise_hnp[khn,j] = averageDom(varstdhn[2,:,:], 2, domain['Pacific'], lat, density)
else:
varnoise_hnp[khn,j] = np.ma.std(averageDom(varhn[:,2,:,:], 3, domain['Pacific'], lat, density), axis=0)
if domain['Indian'] != None:
if method_noise == 'average_std':
varnoise_hni[khn,j] = averageDom(varstdhn[3,:,:], 2, domain['Indian'], lat, density)
else:
varnoise_hni[khn,j] = np.ma.std(averageDom(varhn[:,3,:,:], 3, domain['Indian'], lat, density), axis=0)
# Loop over number of hist runs
for kh in range(nruns_h):
varnoise_a = np.ma.max(varstd[:, 1, :], axis=0)
varnoise_p = np.ma.max(varstd[:, 2, :], axis=0)
varnoise_i = np.ma.max(varstd[:, 3, :], axis=0)
# Loop over 5 domains
for j, domain_name in enumerate(domains):
print('- ', j, domains[j])
# Select domain to average
domain = ToEdomainhistvshistNat(model['name'], domain_name)[0]
domain_char = ToEdomainhistvshistNat(model['name'], domain_name)[1]
if method_noise == 'average_std':
# Average noise
if domain['Atlantic'] != None:
varnoise_a[j] = averageDom(varstd_a, 2, domain['Atlantic'],
lat, density)
if domain['Pacific'] != None:
varnoise_p[j] = averageDom(varstd_p, 2, domain['Pacific'], lat,
density)
if domain['Indian'] != None:
varnoise_i[j] = averageDom(varstd_i, 2, domain['Indian'], lat,
density)
# Loop over number of runs
for k in range(nruns):
run_number = os.path.basename(listruns[k]).split('.')[3]
run_names[k] = run_number
print(' . run number', k, run_number)
# Read file
fh = open_ncfile(listruns[k], 'r')
# Read variables
lat = fh.variables['latitude'][:]
density = fh.variables['lev'][:]
time = fh.variables['time'][:]
varh = fh.variables[var][:]
varhn = fhn.variables[var][:]
# Take difference hist - histNat
var_diff = varh - varhn
# Compute std of histNat
varstd = np.ma.std(varhn, axis=0)
# Average in domains
if SouthST['Atlantic'] != None:
varSouthST = averageDom(var_diff[:, 1, :, :], 3, SouthST['Atlantic'], lat,
density)
noiseSouthST = averageDom(2 * varstd[1, :, :], 2, SouthST['Atlantic'], lat,
density)
if SO['Atlantic'] != None:
varSO = averageDom(var_diff[:, 1, :, :], 3, SO['Atlantic'], lat, density)
noiseSO = averageDom(2 * varstd[1, :, :], 2, SO['Atlantic'], lat, density)
if NA['Atlantic'] != None:
varNA = averageDom(var_diff[:, 1, :, :], 3, NA['Atlantic'], lat, density)
noiseNA = averageDom(2 * varstd[1, :, :], 2, NA['Atlantic'], lat, density)
# ----- Plot ------
fig, ax = plt.subplots(figsize=(10, 5))
ax.set_xlim([tmin + iniyear, tmax + iniyear])
# North Atlantic
print('- Computing ToE of', model['name'], 'with', nMembers[i],
'members')
nruns1 = nruns + nMembers[i]
toe_a[nruns:nruns1, :, :] = toe2read[:, 1, :, :]
toe_p[nruns:nruns1, :, :] = toe2read[:, 2, :, :]
toe_i[nruns:nruns1, :, :] = toe2read[:, 3, :, :]
# -- Select domain to average for each model
domain = ToEdomainhistvshistNat(model['name'], domain_name)[0]
domain_char = ToEdomainhistvshistNat(model['name'], domain_name)[1]
# -- Average toe and determine median of the model
if domain['Atlantic'] != None:
varToEA[nruns:nruns1] = np.ma.around(
averageDom(toe_a[nruns:nruns1, :, :], 3, domain['Atlantic'],
lat, density)) + iniyear
medmodelsToEA[i] = np.ma.around(np.ma.median(
varToEA[nruns:nruns1]))
print('Median ToE Atlantic:', medmodelsToEA[i])
if domain['Pacific'] != None:
varToEP[nruns:nruns1] = np.ma.around(
averageDom(toe_p[nruns:nruns1, :, :], 3, domain['Pacific'],
lat, density)) + iniyear
medmodelsToEP[i] = np.ma.around(np.ma.median(
varToEP[nruns:nruns1]))
print('Median ToE Pacific:', medmodelsToEP[i])
if domain['Indian'] != None:
varToEI[nruns:nruns1] = np.ma.around(
averageDom(toe_i[nruns:nruns1, :, :], 3, domain['Indian'], lat,
density)) + iniyear
medmodelsToEI[i] = np.ma.around(np.ma.median(
lat = fh.variables['latitude'][:]
density = fh.variables['lev'][:]
time = fh.variables['time'][:]
varh = fh.variables[var][:]
varhn = fhn.variables[var][:]
# Take difference hist - histNat
var_diff = varh - varhn
# Compute std of histNat
varstd = np.ma.std(varhn, axis=0)
# Average in domains
if SouthST['Atlantic'] != None :
varSouthST = averageDom(var_diff[:,1,:,:], 3, SouthST['Atlantic'], lat, density)
noiseSouthST = averageDom(2*varstd[1,:,:], 2, SouthST['Atlantic'], lat, density)
if SO['Atlantic'] != None :
varSO = averageDom(var_diff[:,1,:,:], 3, SO['Atlantic'], lat, density)
noiseSO = averageDom(2*varstd[1,:,:], 2, SO['Atlantic'], lat, density)
if NA['Atlantic'] != None :
varNA = averageDom(var_diff[:,1,:,:], 3, NA['Atlantic'], lat, density)
noiseNA = averageDom(2*varstd[1,:,:], 2, NA['Atlantic'], lat, density)
# ----- Plot ------
fig, ax = plt.subplots(figsize=(10,5))
ax.set_xlim([tmin+iniyear,tmax+iniyear])
# North Atlantic
