def invert_example(axes=[], show=True):
"""Test case for reverse engineering rough solar wind params
We look here at the 1989 storm as studied by Nagatsuma (2015)
and Boteler (2019)
"""
import datetime as dt
import spacepy.toolbox as tb
# fetch doesn't respect the days (it grabs 1 month at a time)
# but we'll give the days we're interested in for reference
st_tup = (1989, 3, 12)
en_tup = (1989, 3, 15)
ky_dat = kyo.fetch('dst', st_tup, en_tup)
st = dt.datetime(*st_tup)
en = dt.datetime(*en_tup)
inds = tb.tOverlapHalf([st, en], ky_dat['time'])
vbz = inverseBurton(ky_dat['dst'][inds], rectify=True)
vbz_OB = inverseOBrienMcPherron(ky_dat['dst'][inds])
if not axes:
fig, ax = plt.subplots(2, sharex=True, figsize=(10, 4.5))
else:
ax = axes
ax[0].plot(ky_dat['time'][inds], vbz / 1e3, 'r-', label='Burton')
ax[0].set_ylabel('v.B$_{south}$ [mV/m]')
ax[0].plot(ky_dat['time'][inds], vbz_OB / 1e3, 'b-', label='O-M')
ax[0].set_ylabel('v.B$_{south}$ [mV/m]')
ax[0].legend()
ax[1].plot(ky_dat['time'][inds], ky_dat['dst'][inds])
ax[1].set_ylabel('Dst [nT]')
ax[1].set_xlabel('1989')
if show:
plt.show()
def initialSW():
"""Construct initial estimate of hourly solar wind parameters"""
st = dt.datetime(1989, 3, 12)
en = dt.datetime(1989, 3, 15)
hourly = getKondrashovSW()
# Keep IMF By and n, set V to Boteler/Nagatsuma
t_ssc1 = dt.datetime(1989, 3, 13, 1, 27)
t_ssc2 = dt.datetime(1989, 3, 13, 7, 43)
t_cme = dt.datetime(1989, 3, 13, 16)
t_turn = dt.datetime(1989, 3, 14, 2)
# Set Bx positive, in accordance with ISEE-3 data, Vy/Vz->0
hourly['Bx'] = dm.dmfilled(hourly['By'].shape, fillval=3)
hourly['Vy'] = dm.dmfilled(hourly['By'].shape, fillval=0)
hourly['Vz'] = dm.dmfilled(hourly['By'].shape, fillval=0)
# Before first SSC
inds_before_1 = tb.tOverlapHalf([dt.datetime(1989, 3, 12), t_ssc1],
hourly['DateTime'])
hourly['V_sw'][inds_before_1] = 400
# Between first and ssecond SSC
inds_between_12 = tb.tOverlapHalf([t_ssc1, t_ssc2], hourly['DateTime'])
hourly['V_sw'][inds_between_12] = 550
# IMF turns north around 1989-03-14T02:00:00 according to inverse Burton and Kondrashov
inds_mainphase = tb.tOverlapHalf([t_ssc2, t_turn], hourly['DateTime'])
hourly['V_sw'][inds_mainphase] = 983
# Then have speed decay towards IMP-8 measurement which is ballpark 820 km/s
inds_rest = tb.tOverlapHalf([t_turn, hourly['DateTime'][-1]],
hourly['DateTime'])
hourly['V_sw'][inds_rest] = np.linspace(983, 820, len(inds_rest))
# Now we have speed, estimate temperature
hourly['Plasma_temp'] = emp.getExpectedSWTemp(hourly['V_sw'],
model='XB15',
units='K')
inds_cme = tb.tOverlapHalf([t_cme, en], hourly['DateTime'])
hourly['Plasma_temp'][
inds_cme] /= 3 # reduce by factor of 3 for CME-like temp
# Get "Kondrashov VBs" using V from Boteler/Nagatsuma
hourly['VBs_K'] = 1e-3 * hourly['V_sw'] * rectify(
-1 * hourly['Bz']) # mV/m
# Now get VBs from inverse Burton
ky_dat = kyo.fetch('dst', (st.year, st.month, st.day),
(en.year, en.month, en.day))
inds = tb.tOverlapHalf([st, en], ky_dat['time'])
# Pressure correct here using n and V
hourly = calc_P(hourly)
hourly['Dst'] = ky_dat['dst'][inds]
dst_star = emp.getDststar(hourly, model='OBrien')
hourly['VBs_OB'] = 1e-3 * inverseOBrienMcPherron(dst_star)
# Make new Bz from VBs_OB
hourly['Bz_OB'] = -1e3 * hourly['VBs_OB'] / hourly['V_sw'] # nT
return hourly
efw_file='*'+date+'*.cdf'
gemf=glob.glob(efw_file)
pyf=pycdf.CDF(gemf[0])
epoch=pd.DatetimeIndex(pyf['epoch'][...])
MLML=pyf['mlt_lshell_mlat'][...]
dMLT=pd.DataFrame({'MLT':np.swapaxes(MLML, 1,0)[0]}, index=epoch)
dL=pd.DataFrame({'L':np.swapaxes(MLML, 1,0)[1]}, index=epoch)
dMLAT=pd.DataFrame({'MLAT':np.swapaxes(MLML,1,0)[2]},index=epoch)
deDens=pd.DataFrame({'eDens':pyf['density'][...]}, index=epoch)
rt = pd.period_range(date,periods=1441, freq='T').to_timestamp()
eDens=np.array(deDens['eDens'].resample('1min', how='median').reindex(index=rt,fill_value=np.nan))
L=np.array(dL['L'].resample('1min', how='median').reindex(index=rt,fill_value=np.nan))
MLT=np.array(dMLT['MLT'].resample('1min', how='median').reindex(index=rt,fill_value=np.nan))
MLAT=np.array(dMLAT['MLAT'].resample('1min', how='median').reindex(index=rt,fill_value=np.nan))
# get Kp
kyoto=spk.fetch('kp', dt1, dt1)
day=datetime.datetime.strftime(dt1,'%d') # need to get day of month
kp_arr=kyoto['kp'][(int(day)-1)*8: int(day)*8]
kpC=0
kpM=0
nKP=np.zeros(1440)
for iKP in range(1440):
nKP[iKP]=kp_arr[kpM]
kpC+=1
if kpC == 180:
kpC=0
kpM+=1
Kp=np.array(nKP)
#
# great, now everything is on same time scale
Created on Thu Jul 9 01:52:44 2020
@author: agnitm
"""
import numpy as np
import spacepy.pybats as pb
import spacepy.pybats.kyoto as kyoto
import matplotlib.pyplot as plt
import datetime as dt
time1 = dt.datetime(2010, 4, 4, 22, 0)
time2 = dt.datetime(2010, 4, 5, 10, 45)
dt = dt.timedelta(minutes=1)
kyoto_ae = kyoto.fetch('ae', time1, time2)
t = time1
aur_data = {}
aur_data['time'] = []
aur_data['AU'] = []
aur_data['AL'] = []
while (t <= time2):
fname = (
'AurIndex_superhi/MagGrid_superhires/mag_grid_e{:}{:0=2d}{:0=2d}' +
'-{:0=2d}{:0=2d}{:0=2d}.out').format(t.year, t.month, t.day, t.hour,
t.minute, t.second)
print(fname)
if fname != 'AurIndex_superhi/MagGrid_superhires/mag_grid_e20100405-083000.out':
data = open(fname, 'r')
# put into pandas arrays
energy=np.swapaxes(pyf['HOPE_ENERGY_Ion'][...],1,0)
epoch=pyf['Epoch_Ion'][...]
MLT=pyf['MLT_Ion'][...]
L=pyf['L_Ion'][...]
EnergyDelta=np.swapaxes(pyf['ENERGY_Ion_DELTA'][...],1,0)[n1:n2]
FLUX=pyf[name_species[iSpe]][...]
FLUXl=np.swapaxes(FLUX,1,0)[n1:n2]
dataFrames={}
MLTdf={}
Ldf={}
Fluxdf={}
Energydf={}
Deltadf={}
Densitydf=np.zeros(1440) # number of minutes in day
kyoto=spk.fetch('kp', DT, DT)
day=datetime.datetime.strftime(DT,'%d') # need to get day of month
kp_arr=kyoto['kp'][(int(day)-1)*8: int(day)*8]
kpC=0
kpM=0
nKP=np.zeros(1440)
for iKP in range(1440):
nKP[iKP]=kp_arr[kpM]
kpC+=1
if kpC == 180:
kpC=0
kpM+=1
else:
continue
#
# get the spacecraft potential
from spacepy.pybats import kyoto as kt
from spacepy.pybats import ImfInput, apply_smart_timeticks
from smooth import smooth
from burton import burton_rk4
start = dt.datetime(1989, 3, 13, 0, 0, 0)
end = dt.datetime(1989, 3, 16, 6, 0, 0)
# LOAD ALL RELEVANT DATA.
# Only load on first call. THIS IS WHY WE USE RUN -I.
# Get observed Kyoto dst:
if ('dst' not in locals()) or ('dst' not in globals()):
print('Fetching DST from Kyoto... please hold.')
dst = kt.fetch('dst', [1989, 3], [1989, 3])
dst['t'] = np.array([(x-start).total_seconds()/60.0 for x in dst['time']])
# Load omni data.
if ('omni' not in locals()) or ('omni' not in globals()):
print('Loading OMNI data... please hold.')
omni = read_ascii('./omni_min_3149.lst')
omni['t'] = np.array([(x-start).total_seconds()/60.0 for x in omni['time']])
# Load SEA data.
sourcedir = '/Users/dwelling/projects/SEA_drivers/CME/SEA_MeanData/'
var = ['vx', 'rho', 'bx', 'by', 'bz', 'temp', 'dst']
fil = ['MeanNvx.mat', 'MeanNn.mat', 'MeanNBx.mat', 'MeanNBy.mat',
'MeanNBz.mat', 'MeanNT.mat', 'MeanNDst.mat']
sea = {}
sea['t'] = np.array(np.arange(6040.))
def makeSW_v2():
"""Construct initial estimate of hourly solar wind parameters"""
st = dt.datetime(1989, 3, 12)
en = dt.datetime(1989, 3, 15)
hourly = getKondrashovSW()
# Keep IMF By and n, set V to Boteler/Nagatsuma
t_ssc1 = dt.datetime(1989, 3, 13, 1, 27)
t_ssc2 = dt.datetime(1989, 3, 13, 7, 43)
t_cme = dt.datetime(1989, 3, 13, 16)
t_turn = dt.datetime(1989, 3, 14, 2)
# Set Bx positive, in accordance with ISEE-3 data, Vy/Vz->0
hourly['Bx'] = dm.dmfilled(hourly['By'].shape, fillval=3)
hourly['Vy'] = dm.dmfilled(hourly['By'].shape, fillval=0)
hourly['Vz'] = dm.dmfilled(hourly['By'].shape, fillval=0)
# Before first SSC
inds_before_1 = tb.tOverlapHalf([dt.datetime(1989, 3, 12), t_ssc1],
hourly['DateTime'])
hourly['V_sw'][inds_before_1] = 400
# Between first and ssecond SSC
inds_between_12 = tb.tOverlapHalf([t_ssc1, t_ssc2], hourly['DateTime'])
hourly['V_sw'][inds_between_12] = 550
# IMF turns north around 1989-03-14T02:00:00 according to inverse Burton and Kondrashov
inds_mainphase = tb.tOverlapHalf([t_ssc2, t_turn], hourly['DateTime'])
hourly['V_sw'][inds_mainphase] = 983
# Then have speed decay towards IMP-8 measurement which is ballpark 820 km/s
inds_rest = tb.tOverlapHalf([t_turn, hourly['DateTime'][-1]],
hourly['DateTime'])
hourly['V_sw'][inds_rest] = np.linspace(983, 820, len(inds_rest))
# Now we have speed, estimate temperature
hourly['Plasma_temp'] = emp.getExpectedSWTemp(hourly['V_sw'],
model='XB15',
units='K')
inds_cme = tb.tOverlapHalf([t_cme, en], hourly['DateTime'])
hourly['Plasma_temp'][
inds_cme] /= 3 # reduce by factor of 3 for CME-like temp
# Get "Kondrashov VBs" using V from Boteler/Nagatsuma
hourly['VBs_K'] = 1e-3 * hourly['V_sw'] * rectify(
-1 * hourly['Bz']) # mV/m
# Now get VBs from inverse Burton
ky_dat = kyo.fetch('dst', (st.year, st.month, st.day),
(en.year, en.month, en.day))
inds = tb.tOverlapHalf([st, en], ky_dat['time'])
# Substitute density curve from Sept. 2017 (double shock)
sep17 = pybats.ImfInput(
filename=
'/home/smorley/projects/github/advect1d/IMF_201709_advect_filt.dat',
load=True)
den_inds = tb.tOverlapHalf(
[t_ssc1 - dt.timedelta(hours=25), hourly['DateTime'][-1]],
hourly['DateTime'])
nhours = len(den_inds)
# Keep the opening 8 hours from Kondrashov (2017 event gets high)
hourly['Den_P'][den_inds[9:]] = 2 + (sep17['rho'][::60][9:nhours] * 2)
# After shocks, ensure number density doesn't drop below 10 (keeps M_A over 2)
after_ssc2 = hourly['DateTime'] > t_ssc2
under_lim = hourly['Den_P'] <= 10
limit_inds = np.logical_and(after_ssc2, under_lim)
hourly['Den_P'][limit_inds] = 10
# Pressure correct here using n and V
hourly = calc_P(hourly)
hourly['Dst'] = ky_dat['dst'][inds]
dst_star = emp.getDststar(hourly, model='OBrien')
hourly['VBs_OB'] = 1e-3 * inverseOBrienMcPherron(dst_star)
# Make new Bz from VBs_OB
hourly['Bz_OB'] = -1e3 * hourly['VBs_OB'] / hourly['V_sw'] # nT
return hourly
dMLAT = pd.DataFrame({'MLAT': np.swapaxes(MLML, 1, 0)[2]},
index=epoch)
deDens = pd.DataFrame({'eDens': pyf['density'][...]},
index=epoch)
rt = pd.period_range(date, periods=1441,
freq='T').to_timestamp()
eDens = np.array(deDens['eDens'].resample(
'1min', how='median').reindex(index=rt, fill_value=np.nan))
L = np.array(dL['L'].resample('1min', how='median').reindex(
index=rt, fill_value=np.nan))
MLT = np.array(dMLT['MLT'].resample(
'1min', how='median').reindex(index=rt, fill_value=np.nan))
MLAT = np.array(dMLAT['MLAT'].resample(
'1min', how='median').reindex(index=rt, fill_value=np.nan))
# get Kp
kyoto = spk.fetch('kp', dt1, dt1)
day = datetime.datetime.strftime(
dt1, '%d') # need to get day of month
kp_arr = kyoto['kp'][(int(day) - 1) * 8:int(day) * 8]
kpC = 0
kpM = 0
nKP = np.zeros(1440)
for iKP in range(1440):
nKP[iKP] = kp_arr[kpM]
kpC += 1
if kpC == 180:
kpC = 0
kpM += 1
Kp = np.array(nKP)
#
for i, l in enumerate(lines):
parts = l.split()
data['time'][i] = parse(' '.join(parts[:2]))
data['symH'][i] = parts[-1]
return data
if __name__ == '__main__':
import matplotlib.pyplot as plt
from spacepy.pybats import kyoto, apply_smart_timeticks, ImfInput
infile = 'imf20000715.dat'
# Obtain actual Dst.
dst = kyoto.fetch('dst', (2000, 7), (2000, 7))
symH= read_symH('WWW_aeasy00006705.dat')
# Simple comparison with 5 minute timestep.
tEul, dEul = burton_euler(infile)
tRk4, dRk4 = burton_rk4(infile)
fig= plt.figure(figsize=(8,5.5))
ax = fig.add_subplot(111)
ax.plot(dst['time'], dst['dst'], 'k--', symH['time'], symH['symH'], 'k-',
tEul, dEul, 'b-', tRk4, dRk4, 'r-')
ax.legend( ['$D_{ST}$', 'Sym-H', 'Burton (Euler)',
'Burton (RK4)'], loc='best')
ax.set_ylabel('Disturbance ($nT$)')
ax.grid()
apply_smart_timeticks(ax, tEul, dolabel=True)