def test_carrington_rotation_roundtrip():
t = Time('2010-1-1')
crot = sun.carrington_rotation_number(t)
t_roundtrip = sun.carrington_rotation_time(crot)
dt = t - t_roundtrip
# Stated precision in the docstring is 0.11 seconds
assert_quantity_allclose(dt.to(u.s), 0 * u.s, atol=0.11 * u.s)
def make_afrl_df(source_file, resolution='1D', series='Agg8.25D'):
"""
:param source_file: dir where pickle detection files are saved generated by main.py
:param resolution: time regularization parameter to put everything on a consistent grid
:param series: Which of the aggregation series to use
:return: none
"""
pch_dic = pickle.load(open(source_file, 'rb'))
north = pch_dic[series][0].N_mean_lat
north_std = pch_dic[series][0].N_mean_lat.rolling(series[3:]).std()
north_min = north - north_std
north_min[north_min.lt(np.deg2rad(50))] = np.deg2rad(50)
# Calculating Standard deviation and clipping out of bounds
north_max = north + north_std
north_max[north_max.gt(np.pi / 2)] = np.pi / 2
south = pch_dic[series][1].S_mean_lat
south_std = pch_dic[series][1].S_mean_lat.rolling(series[3:]).std()
south_min = south - south_std
south_min[south_min.lt(-np.pi / 2)] = -np.pi / 2
south_max = south + south_std
south_max[south_max.gt(-np.deg2rad(50))] = -np.deg2rad(50)
# Binning down to the specified resolution
idx = pd.date_range(np.min([north.index[0], south.index[0]]),
np.max([north.index[-1], south.index[-1]]),
freq=resolution)
north = north.resample(resolution).median().reindex(idx, fill_value=np.nan)
north_min = north_min.resample(resolution).median().reindex(
idx, fill_value=np.nan)
north_max = north_max.resample(resolution).median().reindex(
idx, fill_value=np.nan)
south = south.resample(resolution).median().reindex(idx, fill_value=np.nan)
south_min = south_min.resample(resolution).median().reindex(
idx, fill_value=np.nan)
south_max = south_max.resample(resolution).median().reindex(
idx, fill_value=np.nan)
afrl_df = pd.DataFrame(
data={
'north': north,
'north_min': north_min,
'north_max': north_max,
'south': south,
'south_min': south_min,
'south_max': south_max
})
afrl_df['CR'] = carrington_rotation_number(afrl_df.index)
save_name = path.join(path.dirname(path.abspath(source_file)),
'PCH_' + series + '_' + resolution + '_Bin.pkl')
afrl_df.to_pickle(save_name)
def get_sharps(outputpath,
t_start,
t_end,
restart=True,
method='cm',
maxlon=90):
"""
Generate pickle file listing SHARPs within a timeframe.
Set method='cm' to choose frames closest to Central Meridian, or
method='maxflux' to choose frames with the maximum unsigned flux.
Parameter maxlon sets the maximum E-W distance from Central Meridian to include (in degrees).
"""
# Start time:
cr_start = int(carrington_rotation_number(t_start))
# Identify all SHARP regions present within simulation timeframe,
# with at least one frame within +=maxlon of Central Meridian:
# [cache in pickle file for later use]
try:
if (restart):
picklefile = open(outputpath + 'sharps.p', 'rb')
picklefile.close()
print('Reading pre-computed SHARPs from sharps.p')
except:
restart = False
if (restart == False):
sharps, t_rec, t_em = getSHARPs(t_start,
t_end,
method=method,
maxlon=maxlon,
discardfile=outputpath +
'limb_discards.txt')
picklefile1 = open(outputpath + 'sharps.p', 'wb')
pickle.dump(sharps, picklefile1)
pickle.dump(t_rec, picklefile1)
pickle.dump(t_em, picklefile1)
picklefile1.close()
def test_carrington_rotation_number(date, day_fraction, rotation_number):
assert_allclose(sun.carrington_rotation_number(
Time(date, scale='tt') + day_fraction * u.day),
rotation_number,
rtol=0,
atol=2e-4)
def get_pair_overlaps(outputpath,
sharpsfile,
repeat_threshold=1,
outfile='repeatpairs.txt',
plots=True,
bmin=1e-12,
t_restart=''):
"""
Reads data file produced by get_bmrs, identifies pairs of repeat regions, and saves list of their numbers to text file.
"""
print('Identifying repeat regions...')
# Read "good" data (i.e. ones with BMRs):
dat = np.loadtxt(outputpath + sharpsfile,
skiprows=5,
delimiter='\t',
dtype={
'names': ('SHARP', 'NOAA', 'CM time', 'Latitude',
'Carr-Longitude', 'Unsgnd flux',
'Imbalance', 'Good', 'Dipole',
'Bip-Separation', 'Bip-Tilt', 'Bip-Dipole'),
'formats': ('i4', 'i4', 'S10', 'f8', 'f8', 'f8', 'f8',
'i4', 'f8', 'f8', 'f8', 'f8')
})
dat = dat[:][np.where(dat['Good'] == 1)]
times = [
datetime.datetime.strptime(d.decode("utf-8"), '%Y-%m-%d')
for d in dat['CM time']
]
# Prepare new ASCII file:
if (plots):
plt.figure(figsize=(12, 4))
gs = gridspec.GridSpec(2, 3, width_ratios=[1, 1, 1])
bmax = 200
# - Choose plotting boundaries:
pmin = 0
pmax = 360
smin = -1
smax = 1
# - get coordinates:
f = netcdf.netcdf_file(outputpath + 'sharp%5.5i.nc' % dat['SHARP'][0],
'r',
mmap=False)
br = f.variables['br'][:]
f.close()
ns, nph = np.shape(br)
del (br)
ds = 2.0 / ns
dp = 2 * np.pi / nph
sc = np.linspace(-1 + 0.5 * ds, 1 - 0.5 * ds, ns)
pc = np.linspace(0.5 * dp, 2 * np.pi - 0.5 * dp, nph)
cnt = 0
# If not restarting, set initial time and initialise ASCII file:
if (t_restart == ''):
t_restart = np.min(times)
repeatfile = open(outputpath + outfile, 'w')
repeatfile.write(
'List of pairs of repeat SHARPs with repeat_threshold=%g\n' %
repeat_threshold)
repeatfile.write('-- Produced by anthony.yeates[at]durham.ac.uk --\n')
else:
repeatfile = open(outputpath + outfile, 'a')
# Loop through each region:
for k1, sharp1 in enumerate(dat['SHARP']):
print(sharp1)
loaded1 = False
for k2 in range(k1 + 1, len(dat)):
# Select regions later in list with CM passage from 20 to 34 days later:
if ((times[k2] > t_restart) &
(times[k2] - times[k1] >= datetime.timedelta(days=20)) &
(times[k2] - times[k1] <= datetime.timedelta(days=34))):
sharp2 = dat['SHARP'][k2]
# - Load footprints of regions on computational grid:
if (loaded1 == False):
# Load footprint of region on computational grid:
f = netcdf.netcdf_file(outputpath +
'sharp%5.5i.nc' % sharp1,
'r',
mmap=False)
br1 = f.variables['br'][:]
br1b = f.variables['br_bipole'][:]
f.close()
reg1 = (np.abs(br1) > bmin).astype('float')
reg1b = (np.abs(br1b) > bmin).astype('float')
flux1 = np.sum(np.abs(br1))
loaded1 = True
f = netcdf.netcdf_file(outputpath + 'sharp%5.5i.nc' % sharp2,
'r',
mmap=False)
br2 = f.variables['br'][:]
br2b = f.variables['br_bipole'][:]
f.close()
reg2 = (np.abs(br2) > bmin).astype('float')
flux2 = np.sum(np.abs(br2))
if (flux2 > flux1):
continue
# - Rotate back differentially:
dt = (times[k2] - times[k1]).days
br2d = derotate(br2, times[k2] - times[k1])
reg2d = (np.abs(br2d) > bmin).astype('float')
# - Flux of BMR 1 in derotated BMR 2:
ofluxb1 = np.sum(np.abs(br1) * reg2d)
olap = ofluxb1 / flux2
# - Get number of pixels in BMR representations
npix1b = np.sum(reg1)
npix2bd = np.sum(reg2d)
# - If this is greater than overlap_threshold, record and plot:
if ((olap > repeat_threshold) & (npix1b * npix2bd > 0)):
if (plots):
# Load corresponding HMI synoptic maps for context:
crot1 = carrington_rotation_number(times[k1])
brm1, scm1, pcm1 = readmap(crot1)
crot2 = carrington_rotation_number(times[k2])
brm2, scm2, pcm2 = readmap(crot2)
# Compare two regions on grid:
ax = plt.subplot(gs[0])
pm = ax.pcolormesh(np.rad2deg(pcm1),
scm1,
brm1,
cmap='bwr')
plt.contour(np.rad2deg(pc),
sc,
reg1, [0.5],
colors='g',
linewidths=0.75)
pm.set_clim(vmin=-bmax, vmax=bmax)
ax.text(10, 0.8, '(a) CR%4.4i' % crot1)
ax.set_ylabel('Sine Latitude')
ax = plt.subplot(gs[1])
pm = ax.pcolormesh(np.rad2deg(pc), sc, br1, cmap='bwr')
pm.set_clim(vmin=-bmax, vmax=bmax)
ax.text(pmin + 1. / 36 * (pmax - pmin),
smax - 0.1 * (smax - smin),
'(b) SHARP %i' % sharp1)
ax.set_xlim(pmin, pmax)
ax.set_ylim(smin, smax)
ax = plt.subplot(gs[2])
pm = ax.pcolormesh(np.rad2deg(pc),
sc,
br1b,
cmap='bwr')
pm.set_clim(vmin=-bmax, vmax=bmax)
ax.text(
pmin + 1. / 36 * (pmax - pmin),
smax - 0.1 * (smax - smin), r'(c) $\Phi$ = %6.2e' %
(flux1 * ds * dp * 6.96e10**2))
# plt.contour(np.rad2deg(pc), sc, reg2bd, [0.5], colors='k', linewidths=0.75, linestyles='--')
ax.text(pmin + 1. / 36 * (pmax - pmin),
smax - 0.95 * (smax - smin),
r'$R_{12}$ = %4.2f' % olap)
ax.set_xlim(pmin, pmax)
ax.set_ylim(smin, smax)
ax = plt.subplot(gs[3])
pm = ax.pcolormesh(np.rad2deg(pcm2),
scm2,
brm2,
cmap='bwr')
plt.contour(np.rad2deg(pc),
sc,
reg2, [0.5],
colors='k',
linewidths=0.75)
pm.set_clim(vmin=-bmax, vmax=bmax)
ax.text(10, 0.8, '(d) CR%4.4i' % crot2)
ax.set_xlabel('Carrington Longitude')
ax.set_ylabel('Sine Latitude')
ax = plt.subplot(gs[4])
pm = ax.pcolormesh(np.rad2deg(pc),
sc,
br2d,
cmap='bwr')
pm.set_clim(vmin=-bmax, vmax=bmax)
ax.text(pmin + 1. / 36 * (pmax - pmin),
smax - 0.1 * (smax - smin),
'(e) SHARP %i' % sharp2)
ax.set_xlabel('Carrington Longitude')
ax.set_xlim(pmin, pmax)
ax.set_ylim(smin, smax)
ax = plt.subplot(gs[5])
br2bd = derotate(br2b, times[k2] - times[k1])
pm = ax.pcolormesh(np.rad2deg(pc),
sc,
br2bd,
cmap='bwr')
pm.set_clim(vmin=-bmax, vmax=bmax)
ax.text(
pmin + 1. / 36 * (pmax - pmin),
smax - 0.1 * (smax - smin), r'(f) $\Phi$ = %6.2e' %
(flux2 * ds * dp * 6.96e10**2))
ax.set_xlabel('Carrington Longitude')
ax.set_xlim(pmin, pmax)
ax.set_ylim(smin, smax)
plt.savefig(outputpath + 'repeat1_%5.5i.png' % cnt,
bbox_inches='tight')
plt.clf()
cnt += 1
# Output to list of repeat regions:
repeatfile.write('%i\t%i\n' % (sharp1, sharp2))
repeatfile.close()
for jj in range(0,3):
os.chdir(data_directory[jj])
for fitsfile in glob.glob('*.fits'):
# Open the fits file
hdulist_hinode = fits.open(fitsfile)
# Create astropy time objects of the start and end times of the Hinode scan in UTC.
start_time = Time(hdulist_hinode[0].header['TSTART'], format='isot', scale='utc')
end_time = Time(hdulist_hinode[0].header['TEND'], format='isot', scale='utc')
# Determine time in the middle of the scan (still in UTC).
middle_of_scan = start_time+(end_time-start_time)/2
# Determine the Carrington rotation of this scan, round to the first significant
# figure (which corresponds to the stretching of 10 we made above)
this_carr_number = np.round(sun.carrington_rotation_number(t=middle_of_scan), decimals=1)
# Here I am sorting out bad coordinates with quantiles, the example below assumes that a
# maximum of 2% of the coordinates are bad (1% at the lower end, 1% at the upper end).
# That fixes most of the maps, but not all of them. For now, I am just adding/subtracting
# an arcsec on either end to counteract the (probable) overcompensation, i.e.
# throwing away good coordinate values as well. If the HMI cutout needs to be
# *really* precise, this needs to be done more carefully. Right now, the HMI map
# might be a bit too small or too large depending on how many coordinate outliers
# there are.
hinode_xcoords = [np.quantile(hdulist_hinode[38].data,0.01)-1., np.quantile(hdulist_hinode[38].data,0.99)+1.]
hinode_ycoords = [np.quantile(hdulist_hinode[39].data,0.01)-1., np.quantile(hdulist_hinode[39].data,0.99)+1.]
# Take care of some problematic files in the list manually
if fitsfile in ['20170402_210600.fits', '20170403_003405.fits', '20170828_102201.fits', \
'20171009_181600.fits', '20171127_175734.fits', '20180401_215905.fits', \
