try:
Ahat = np.vstack([Ahat, snap['A'][:]])
F = np.vstack([F, snap['F'][:]])
except:
Ahat = snap['A'][:]
F = snap['F'][:]
t = np.array(t)
it0 = np.where(t / 86400 > 260)[0][0]
grad = (np.gradient(F[it0:], axis=1))
fmask = ((grad - np.mean(grad)) / np.std(grad) > 5) * 1.0
ict, dmask = dg.count_blocks(fmask, 10, 10)
ts = t[it0:][np.where(dmask > 0.5)[0]] / 86400
xs = x[np.where(dmask > 0.5)[1]] / 1e3
nsig = np.sum(fmask)
result = {
"forcing phase": xc * n / Lx,
"onset": (xs, ts),
"delay": (xs, ts - initc.tc),
"nblocks": ict,
"integrated blocking": nsig
}
ensemble.append(result)
def blocking_iteration(cxpeak=1.0, fpeak=1.8, beta=60, Y=10, noisy_initc=None):
if not noisy_initc:
noisy_initc = noisyconditions(cfunc=noisybackground,
cxpeak=cxpeak,
peak=fpeak,
background=True,
forcing=True,
beta=beta,
Y=Y)
else:
noisy_initc.beta = beta
noisy_initc.Y = Y
noisy_initc.cxpeak = cxpeak
noisy_initc.peak = fpeak
cleanup = True
if cleanup:
os.system("rm -rf output/")
cond = noisy_initc
wavenum = 2
model = AtmosphericBlocking.Model(nx=1024,
Lx=28000e3,
dt=.005 * 86400,
alpha=cond.alpha,
tmax=3.5 * 86400,
D=3.26e5,
tau=cond.tau,
sfunc=cond.forcing,
cfunc=cond.getcx,
forcingpeak=cond.peak,
injection=cond.inject,
save_to_disk=True,
overwrite=True,
tsave_snapshots=50,
path='output/')
model.tmax = 450 * 86400
model.run()
setup = h5py.File("output/setup.h5")
x = setup['grid/x'][:]
fnis = np.array(sorted(glob.glob("output/snapshots/*.h5")))
Ahat, F, S, C = 0, 0, 0, 0
t = []
for fni in fnis[0::2]:
snap = h5py.File(fni)
t.append(snap['t'][()])
try:
Ahat = np.vstack([Ahat, snap['A'][:]])
F = np.vstack([F, snap['F'][:]])
S = np.vstack([S, snap['S'][:]])
C = np.vstack([C, snap['C'][:]])
except:
Ahat = snap['A'][:]
F = snap['F'][:]
S = snap['S'][:]
C = snap['C'][:]
t = np.array(t)
it0 = np.where(t / 86400 > 260)[0][0]
ua = 40 - cond.alpha * (Ahat + model.A0[np.newaxis, :])
gamma = 4 * cond.alpha * F / C**2
masked = gamma.copy()
masked[np.where(gamma < 0.99999)] = 0
masked[np.where(gamma >= 0.99999)] = 1
threshu = 5
ugrad = -(np.gradient(ua[it0:], axis=1))
umask = ((ugrad - np.mean(ugrad)) / np.std(ugrad) > threshu) * 1.0
uct, ucmask = dg.count_blocks(umask, 80 - 2 * wavenum, 10)
uts = t[it0:][np.where(ucmask > 0.5)[0]] / 86400.0
uxs = x[np.where(ucmask > 0.5)[1]] / 1.0e3
shock_size = np.sum(umask)
smask = (S[it0:] > 0.6e-4) * 1.0
sct, scmask = dg.count_blocks(smask, 5, 5)
sts = t[it0:][np.where(scmask > 0.5)[0]] / 86400
sxs = x[np.where(scmask > 0.5)[1]] / 1e3
return cond, uct, shock_size, uxs, uts, sct, sts, sxs