def make_boolean_arrays(dBdth, smstat, simtime, starttime, threshold):
run20, t20 = tb.windowMean(dBdth,
time=simtime,
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
predicted_event = np.asarray(run20) >= threshold
Bdoth = np.array([
linalg.norm(smstat['Bdot'][i, :2]) for i in range(len(smstat['Bdot']))
])
obs20, obst20 = tb.windowMean(Bdoth,
time=smstat['time'],
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
obs_event = np.asarray(obs20) >= threshold
minlen = min(len(run20), len(obs20))
predicted_event = predicted_event[:minlen]
obs_event = obs_event[:minlen]
return predicted_event, obs_event
def test_windowMean_outputTimes(self):
'''windowMean should return a known set of output times for a given set of input times and windows'''
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter('always')
wsize = datetime.timedelta(hours=1)
olap = datetime.timedelta(0)
time = [datetime.datetime(2001,1,1) + datetime.timedelta(minutes=n*15) for n in range(48)]
#last time is datetime.datetime(2001, 1, 1, 11, 45), so last hourly bin should
#cover 1100 to 1200 and be centered at 1130
data= [10]*48
outdata, outtime = tb.windowMean(data, time, winsize=wsize, overlap=olap, st_time=datetime.datetime(2001,1,1))
od_ans = [10]*12
ot_ans = [datetime.datetime(2001, 1, 1, 0, 30),
datetime.datetime(2001, 1, 1, 1, 30),
datetime.datetime(2001, 1, 1, 2, 30),
datetime.datetime(2001, 1, 1, 3, 30),
datetime.datetime(2001, 1, 1, 4, 30),
datetime.datetime(2001, 1, 1, 5, 30),
datetime.datetime(2001, 1, 1, 6, 30),
datetime.datetime(2001, 1, 1, 7, 30),
datetime.datetime(2001, 1, 1, 8, 30),
datetime.datetime(2001, 1, 1, 9, 30),
datetime.datetime(2001, 1, 1, 10, 30),
datetime.datetime(2001, 1, 1, 11, 30)]
numpy.testing.assert_almost_equal(od_ans, outdata)
self.assertEqual(ot_ans, outtime)
def test_windowMean3(self):
"""windowMean should give known results 3(regression)"""
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter('always')
wsize = datetime.timedelta(days=1)
olap = datetime.timedelta(hours=12)
data = [10, 20]*50
time = [datetime.datetime(2001,1,1) + datetime.timedelta(hours=n, minutes = 30) for n in range(100)]
time[50:] = [val + datetime.timedelta(days=2) for val in time[50:]]
outdata, outtime = tb.windowMean(data, time, winsize=wsize, overlap=olap, st_time=datetime.datetime(2001,1,1))
od_ans = [ 15., 15., 15., 15., 15., numpy.nan, numpy.nan, 15., 15., 15., 15., 15., 15.]
ot_ans = [datetime.datetime(2001, 1, 1, 12, 0),
datetime.datetime(2001, 1, 2, 0, 0),
datetime.datetime(2001, 1, 2, 12, 0),
datetime.datetime(2001, 1, 3, 0, 0),
datetime.datetime(2001, 1, 3, 12, 0),
datetime.datetime(2001, 1, 4, 0, 0),
datetime.datetime(2001, 1, 4, 12, 0),
datetime.datetime(2001, 1, 5, 0, 0),
datetime.datetime(2001, 1, 5, 12, 0),
datetime.datetime(2001, 1, 6, 0, 0),
datetime.datetime(2001, 1, 6, 12, 0),
datetime.datetime(2001, 1, 7, 0, 0),
datetime.datetime(2001, 1, 7, 12, 0),
]
numpy.testing.assert_almost_equal(od_ans, outdata)
self.assertEqual(ot_ans, outtime)
def test_windowMean_2d(self):
"""windowMean flattens and extends the time var over all the other dims"""
wsize = 2
olap = 0
data = numpy.arange(20).reshape(10,2)
out = tb.windowMean(data, winsize=wsize, overlap=olap)
ansd = [ 1.5, 5.5, 9.5, 13.5]
numpy.testing.assert_almost_equal(ansd, out[0])
anst = [ 1., 3., 5., 7.]
numpy.testing.assert_almost_equal(anst, out[1])
def test_windowMean4(self):
"""windowMean should give known results 4(regression)"""
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter('always')
wsize = datetime.timedelta(days=1)
olap = datetime.timedelta(hours=12)
data = [10, 20]*50
time = [datetime.datetime(2001,1,1) + datetime.timedelta(hours=n, minutes = 30) for n in range(100)]
time[50:] = [val + datetime.timedelta(days=2) for val in time[50:]]
# now test the pointwise
outdata, outtime = tb.windowMean(data, winsize=24, overlap=12)
od_ans = [15.0, 15.0, 15.0, 15.0, 15.0, 15.0, 15.0]
ot_ans = [12.0, 24.0, 36.0, 48.0, 60.0, 72.0, 84.0]
numpy.testing.assert_almost_equal(ot_ans, outtime)
numpy.testing.assert_almost_equal(od_ans, outdata)
def test_windowMean4(self):
"""windowMean should give known results 4(regression)"""
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter('always')
wsize = datetime.timedelta(days=1)
olap = datetime.timedelta(hours=12)
data = [10, 20]*50
time = [datetime.datetime(2001,1,1) + datetime.timedelta(hours=n, minutes = 30) for n in range(100)]
time[50:] = [val + datetime.timedelta(days=2) for val in time[50:]]
# now test the pointwise
outdata, outtime = tb.windowMean(data, winsize=24, overlap=12)
od_ans = [15.0, 15.0, 15.0, 15.0, 15.0, 15.0, 15.0]
ot_ans = [12.0, 24.0, 36.0, 48.0, 60.0, 72.0, 84.0]
numpy.testing.assert_almost_equal(ot_ans, outtime)
numpy.testing.assert_almost_equal(od_ans, outdata)
def test_windowMean_op2(self):
"""windowMean should give expected sums with len as passed function"""
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter('always')
wsize = datetime.timedelta(days=1)
olap = datetime.timedelta(0)
data = [10, 20]*50
time = [datetime.datetime(2001,1,1) + datetime.timedelta(hours=n, minutes = 30) for n in range(100)]
#four points lie on the 5th, so the sum for that day is 4
outdata, outtime = tb.windowMean(data, time, winsize=wsize, overlap=olap, st_time=datetime.datetime(2001,1,1), op=len)
od_ans = [24, 24, 24, 24, 4]
ot_ans = [datetime.datetime(2001, 1, 1, 12, 0),
datetime.datetime(2001, 1, 2, 12, 0),
datetime.datetime(2001, 1, 3, 12, 0),
datetime.datetime(2001, 1, 4, 12, 0),
datetime.datetime(2001, 1, 5, 12, 0)]
numpy.testing.assert_almost_equal(od_ans, outdata)
self.assertEqual(ot_ans, outtime)
def test_windowMean_op(self):
"""windowMean should give known results (regression)"""
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter('always')
wsize = datetime.timedelta(days=1)
olap = datetime.timedelta(hours=12)
data = [10, 20]*50
time = [datetime.datetime(2001,1,1) + datetime.timedelta(hours=n, minutes = 30) for n in range(100)]
outdata, outtime = tb.windowMean(data, time, winsize=wsize, overlap=olap, st_time=datetime.datetime(2001,1,1), op=len)
od_ans = [24, 24, 24, 24, 24, 24, 24]
ot_ans = [datetime.datetime(2001, 1, 1, 12, 0),
datetime.datetime(2001, 1, 2, 0, 0),
datetime.datetime(2001, 1, 2, 12, 0),
datetime.datetime(2001, 1, 3, 0, 0),
datetime.datetime(2001, 1, 3, 12, 0),
datetime.datetime(2001, 1, 4, 0, 0),
datetime.datetime(2001, 1, 4, 12, 0)]
numpy.testing.assert_almost_equal(od_ans, outdata)
self.assertEqual(ot_ans, outtime)
def test_windowMean5(self):
"""windowMean should give known results 5(regression)"""
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter('always')
wsize = datetime.timedelta(days=1)
olap = datetime.timedelta(hours=12)
data = [10, 20]*50
time = [datetime.datetime(2001,1,1) + datetime.timedelta(hours=n, minutes = 30) for n in range(100)]
time[50:] = [val + datetime.timedelta(days=2) for val in time[50:]]
# winsize tests
outdata, outtime = tb.windowMean(data, winsize=24.6, overlap=12)
od_ans, ot_ans = tb.windowMean(data, winsize=24.6, overlap=12)
numpy.testing.assert_almost_equal(ot_ans, outtime)
numpy.testing.assert_almost_equal(od_ans, outdata)
outdata, outtime = tb.windowMean(data, winsize=0.4)
od_ans, ot_ans = tb.windowMean(data, winsize=1.0)
numpy.testing.assert_almost_equal(ot_ans, outtime)
numpy.testing.assert_almost_equal(od_ans, outdata)
outdata, outtime = tb.windowMean(data, winsize=1.0, overlap=2)
od_ans, ot_ans = tb.windowMean(data, winsize=1.0, overlap=0)
numpy.testing.assert_almost_equal(ot_ans, outtime)
numpy.testing.assert_almost_equal(od_ans, outdata)
self.assertEqual(8, len(w))
def test_windowMean5(self):
"""windowMean should give known results 5(regression)"""
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter('always')
wsize = datetime.timedelta(days=1)
olap = datetime.timedelta(hours=12)
data = [10, 20]*50
time = [datetime.datetime(2001,1,1) + datetime.timedelta(hours=n, minutes = 30) for n in range(100)]
time[50:] = [val + datetime.timedelta(days=2) for val in time[50:]]
# winsize tests
outdata, outtime = tb.windowMean(data, winsize=24.6, overlap=12)
od_ans, ot_ans = tb.windowMean(data, winsize=24.6, overlap=12)
numpy.testing.assert_almost_equal(ot_ans, outtime)
numpy.testing.assert_almost_equal(od_ans, outdata)
outdata, outtime = tb.windowMean(data, winsize=0.4)
od_ans, ot_ans = tb.windowMean(data, winsize=1.0)
numpy.testing.assert_almost_equal(ot_ans, outtime)
numpy.testing.assert_almost_equal(od_ans, outdata)
outdata, outtime = tb.windowMean(data, winsize=1.0, overlap=2)
od_ans, ot_ans = tb.windowMean(data, winsize=1.0, overlap=0)
numpy.testing.assert_almost_equal(ot_ans, outtime)
numpy.testing.assert_almost_equal(od_ans, outdata)
self.assertEqual(8, len(w))
def create_dics(thresholds, stations, starttimes):
dic = {}
midlat_sats = [
'BEL', 'CLF', 'FMC', 'HAD', 'MEA', 'OTT', 'SIT', 'THY', 'WNG', 'DOU',
'FUR', 'HLP', 'PIN', 'STJ', 'UPS', 'BFE', 'ESK', 'GIM', 'NEW', 'PBQ',
'SUA', 'VAL', 'FCC', 'IRT', 'NGK', 'RAL', 'TAR', 'VIC'
]
highlat_sats = [
'ABK', 'BLC', 'BRW', 'BJN', 'CBB', 'CMO', 'DNB', 'DOB', 'EAG', 'FSP',
'SMI', 'HRN', 'IQA', 'STF', 'KEV', 'KUV', 'LER', 'LYR', 'NAQ', 'NAL',
'NRD', 'NUR', 'OUJ', 'THL', 'RAN', 'RES', 'SVS', 'TAL', 'AMK', 'TIK',
'YKC'
]
print('Initializing Dictionary')
for threshold in thresholds:
dic[threshold] = {}
for keywrd in ['hour', 'unsmoothed', '30min']:
dic[threshold][keywrd] = {}
dic[threshold][keywrd]['highlat'] = {}
dic[threshold][keywrd]['midlat'] = {}
for i in range(5)[1:]:
dic[threshold][keywrd]['highlat'][str(i)] = {}
dic[threshold][keywrd]['midlat'][str(i)] = {}
dic[threshold][keywrd]['highlat'][str(i)]['obs'] = []
dic[threshold][keywrd]['midlat'][str(i)]['obs'] = []
dic[threshold][keywrd]['highlat'][str(i)]['predicted'] = []
dic[threshold][keywrd]['midlat'][str(i)]['predicted'] = []
proc_dic = {}
for starttime in starttimes:
proc_dic[starttime] = {}
for station in stations:
proc_dic[starttime][station] = {}
for keywrd in ['hour', 'unsmoothed', '30min']:
proc_dic[starttime][station][keywrd] = {}
for threshold in thresholds:
proc_dic[starttime][station][keywrd][threshold] = {}
print('Filling Processing Dictionary')
for keywrd in ['hour', 'unsmoothed', '30min']:
for starttime in starttimes:
strstart = starttime
date = starttime[:8]
print('Date: ', date)
starttime = dt.datetime.strptime(starttime, '%Y%m%d%H%M%S')
if date == '20010831': date = '20010830'
outputdir = './outputs/{}/'.format(date)
blockPrint()
smdata = supermag_parser.supermag_parser(
'./supermag_data/{0}-supermag.txt'.format(date))
enablePrint()
origdata = open_output_data(outputdir, keywrd, starttime)
for stat in stations:
if stat not in smdata.station: continue
else:
smstat = smdata.station[stat]
dBdth, simtime = process_dBdth(origdata, stat)
mlt_len = len(smstat['mlt'])
stat_mlt20 = tb.windowMean(
smstat['mlt'],
time=smstat['time'][:mlt_len],
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
for threshold in thresholds:
predicted_event_sat, obs_event_sat = make_boolean_arrays(
dBdth, smstat, simtime, starttime, threshold)
proc_dic[strstart][station][keywrd][threshold][
'predicted_events'] = predicted_event_sat
proc_dic[strstart][station][keywrd][threshold][
'obs_events'] = obs_event_sat
proc_dic[strstart][station][keywrd][threshold][
'mlt'] = stat_mlt20[0]
proc_dic[strstart][station][keywrd][threshold][
'mlat'] = smstat['mlat']
# proc_dic is filled with boolean arrays for each threshold level, event, and satellite
# next, we need to seciton up the data into mlt sectors
print('Filling Main Dictionary')
for threshold in thresholds:
for keywrd in ['hour', 'unsmoothed', '30min']:
for stat in stations:
if stat in highlat_sats or stat in midlat_sats:
#print(np.where(proc_dic[strstart][station][keywrd][threshold]['obs_events'] == True))
#print(np.where(proc_dic[strstart][station][keywrd][threshold]['predicted_events'] == True))
for i in range(
len(proc_dic[strstart][station][keywrd][threshold]
['mlt'])):
curmlt = proc_dic[strstart][station][keywrd][
threshold]['mlt'][i]
print(curmlt)
flag = ''
if stat in highlat_sats:
flag = 'highlat'
else:
flag = 'midlat'
if 0 <= curmlt < 6:
dic[threshold][keywrd][flag]['1']['obs'] += [
proc_dic[strstart][station][keywrd][threshold]
['obs_events'][i]
]
dic[threshold][keywrd][flag]['1']['predicted'] += [
proc_dic[strstart][station][keywrd][threshold]
['predicted_events'][i]
]
elif 6 <= curmlt < 12:
dic[threshold][keywrd][flag]['2']['obs'] += [
proc_dic[strstart][station][keywrd][threshold]
['obs_events'][i]
]
dic[threshold][keywrd][flag]['2']['predicted'] += [
proc_dic[strstart][station][keywrd][threshold]
['predicted_events'][i]
]
elif 12 <= curmlt < 18:
dic[threshold][keywrd][flag]['3']['obs'] += [
proc_dic[strstart][station][keywrd][threshold]
['obs_events'][i]
]
dic[threshold][keywrd][flag]['3']['predicted'] += [
proc_dic[strstart][station][keywrd][threshold]
['predicted_events'][i]
]
elif 18 <= curmlt < 24:
dic[threshold][keywrd][flag]['4']['obs'] += [
proc_dic[strstart][station][keywrd][threshold]
['obs_events'][i]
]
dic[threshold][keywrd][flag]['4']['predicted'] += [
proc_dic[strstart][station][keywrd][threshold]
['predicted_events'][i]
]
# now calculate tt for each sector to get metrics
print('lets check differences')
print(dic[0.3]['unsmoothed']['highlat']['1']['obs'] == dic[0.3]
['unsmoothed']['midlat']['1']['obs'])
print(dic[0.3]['unsmoothed']['highlat']['2']['obs'] == dic[0.3]
['unsmoothed']['midlat']['2']['obs'])
print(dic[0.3]['unsmoothed']['highlat']['3']['obs'] == dic[0.3]
['unsmoothed']['midlat']['3']['obs'])
print('calculating statistics')
for threshold in thresholds:
for keywrd in ['hour', 'unsmoothed', '30min']:
for i in range(5)[1:]:
for flag in ['highlat', 'midlat']:
obs_events = dic[threshold][keywrd][flag][str(i)]['obs']
predic_events = dic[threshold][keywrd][flag][str(
i)]['predicted']
tt = verify.Contingency2x2.fromBoolean(
predic_events, obs_events)
sector_skill = tt.heidke(ci='bootstrap')
dic[threshold][keywrd][flag][str(
i)]['heidke'] = sector_skill[0]
return proc_dic, dic
stations = ['YKC', 'MEA', 'NEW', 'FRN', 'IQA', 'PBQ', 'OTT', 'FRD', 'VIC']
stations = [key for key in origdata.keys() if len(key)==3]
for stat in stations:
if stat not in smdata.station: continue
subset = origdata[stat]
simtime = subset['time'][::6]
dBdte = decimate(subset['dBdte'], 6)
dBdtn = decimate(subset['dBdtn'], 6)
dBdth = np.array([linalg.norm([dBdtn[i], dBdte[i]]) for i in range(len(dBdtn))])
smstat = smdata.station[stat]
Bdoth = np.array([linalg.norm(smstat['Bdot'][i,:2]) for i in range(len(smstat['Bdot']))])
fig = plt.figure(figsize=(10,4))
ax = fig.add_subplot(111)
ax.plot(simtime, dBdth, 'b-', alpha=0.4)
ax.plot(smstat['time'], Bdoth, 'r-', alpha=0.4)
run20, t20 = tb.windowMean(dBdth, time=simtime, winsize=dt.timedelta(minutes=20), overlap=dt.timedelta(0), st_time=dt.datetime(2000,8,12,1), op=np.max)
ax.plot(t20, run20, marker='o', color='b', linestyle='none', markersize=3, label='SWMF')
obs20, t20 = tb.windowMean(Bdoth, time=smstat['time'], winsize=dt.timedelta(minutes=20), overlap=dt.timedelta(0), st_time=dt.datetime(2000,8,12,1), op=np.max)
ax.plot(t20, obs20, marker='x', color='r', linestyle='none', markersize=3, label='Obs')
ax.set_ylabel('1-min dB/dt$_{H}$ [nT/s]')
ax.set_xlabel('2000-08-12')
splot.applySmartTimeTicks(ax, subset['time'], dolimit=True)
plt.legend()
plt.title(stat)
plt.tight_layout()
#plt.show()
plt.savefig('Aug2000_dBdth_{}.png'.format(stat))
plt.close()
def make_plots(smdata, stations, outputdir, date, starttime):
unsmoothedfiles = glob.glob(outputdir + '/unsmoothed/mag*mag')
hourlyfiles = glob.glob(outputdir + 'hour/mag*mag')
files_30 = glob.glob(outputdir + '30min/mag*mag')
#unsmoothedfiles = ['magnetometers_e20150316-160000.mag','magnetometers_e20150317-160000.mag','magnetometers_e20150318-160000.mag']
#hourlyfiles = ['magnetometers_e20150316-160000_hour.mag','magnetometers_e20150317-160000_hour.mag']
for stat in stations:
fig, (ax1, ax2, ax3) = plt.subplots(figsize=(10, 9),
nrows=3,
sharey=True)
fig2, (b1, b2, b3) = plt.subplots(figsize=(10, 9),
nrows=3,
sharey=True)
count = 0
for fname in unsmoothedfiles:
origdata = spacepy.pybats.bats.MagFile(fname)
origdata.calc_h()
origdata.calc_dbdt()
#if stat not in smdata.station: continue
subset = origdata[stat]
simtime = subset['time'][::6]
dBdte = decimate(subset['dBdte'], 6)
dBdtn = decimate(subset['dBdtn'], 6)
dBdth = np.array(
[linalg.norm([dBdtn[i], dBdte[i]]) for i in range(len(dBdtn))])
dBe = decimate(subset['dBe'], 6)
dBn = decimate(subset['dBn'], 6)
Bh = np.array(
[linalg.norm([dBn[i], dBe[i]]) for i in range(len(dBn))])
b1.plot(simtime, Bh, 'b-', alpha=0.4)
ax1.plot(simtime, dBdth, 'b-', alpha=0.4)
run20, t20 = tb.windowMean(dBdth,
time=simtime,
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
brun20, bt20 = tb.windowMean(Bh,
time=simtime,
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
if count == 0:
ax1.plot(t20,
run20,
marker='o',
color='b',
linestyle='none',
markersize=3,
label='SWMF Unsmoothed')
b1.plot(bt20,
brun20,
marker='o',
color='b',
linestyle='none',
markersize=3,
label='SWMF Unsmoothed')
count = 1
else:
ax1.plot(t20,
run20,
marker='o',
color='b',
linestyle='none',
markersize=3)
b1.plot(bt20,
brun20,
marker='o',
color='b',
linestyle='none',
markersize=3)
count = 0
if len(hourlyfiles) > 0:
for fname in hourlyfiles:
origdata = spacepy.pybats.bats.MagFile(fname)
origdata.calc_h()
origdata.calc_dbdt()
subset = origdata[stat]
simtime = subset['time'][::6]
dBdte = decimate(subset['dBdte'], 6)
dBdtn = decimate(subset['dBdtn'], 6)
dBdth = np.array([
linalg.norm([dBdtn[i], dBdte[i]])
for i in range(len(dBdtn))
])
dBe = decimate(subset['dBe'], 6)
dBn = decimate(subset['dBn'], 6)
Bh = np.array(
[linalg.norm([dBn[i], dBe[i]]) for i in range(len(dBn))])
ax3.plot(simtime, dBdth, 'b-', alpha=0.4)
b3.plot(simtime, Bh, 'b-', alpha=0.4)
run20, t20 = tb.windowMean(dBdth,
time=simtime,
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
brun20, bt20 = tb.windowMean(Bh,
time=simtime,
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
if count == 0:
ax3.plot(t20,
run20,
marker='o',
color='b',
linestyle='none',
markersize=3,
label='SWMF Hourly')
b3.plot(bt20,
brun20,
marker='o',
color='b',
linestyle='none',
markersize=3,
label='SWMF Hourly')
count = 1
else:
ax3.plot(t20,
run20,
marker='o',
color='b',
linestyle='none',
markersize=3)
b3.plot(bt20,
brun20,
marker='o',
color='b',
linestyle='none',
markersize=3)
count = 0
if len(files_30) > 0:
for fname in files_30:
origdata = spacepy.pybats.bats.MagFile(fname)
origdata.calc_h()
origdata.calc_dbdt()
subset = origdata[stat]
simtime = subset['time'][::6]
dBdte = decimate(subset['dBdte'], 6)
dBdtn = decimate(subset['dBdtn'], 6)
dBdth = np.array([
linalg.norm([dBdtn[i], dBdte[i]])
for i in range(len(dBdtn))
])
dBe = decimate(subset['dBe'], 6)
dBn = decimate(subset['dBn'], 6)
Bh = np.array(
[linalg.norm([dBn[i], dBe[i]]) for i in range(len(dBn))])
ax2.plot(simtime, dBdth, 'b-', alpha=0.4)
b2.plot(simtime, Bh, 'b-', alpha=0.4)
run20, t20 = tb.windowMean(dBdth,
time=simtime,
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
brun20, bt20 = tb.windowMean(Bh,
time=simtime,
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
if count == 0:
ax2.plot(t20,
run20,
marker='o',
color='b',
linestyle='none',
markersize=3,
label='SWMF 30min')
b2.plot(bt20,
brun20,
marker='o',
color='b',
linestyle='none',
markersize=3,
label='SWMF 30min')
count = 1
else:
ax2.plot(t20,
run20,
marker='o',
color='b',
linestyle='none',
markersize=3)
b2.plot(bt20,
brun20,
marker='o',
color='b',
linestyle='none',
markersize=3)
# add in observed
if stat not in smdata.station: continue
else:
smstat = smdata.station[stat]
Bdoth = np.array([
linalg.norm(smstat['Bdot'][i, :2])
for i in range(len(smstat['Bdot']))
])
Bh = np.array([
linalg.norm(smstat['B'][i, :2])
for i in range(len(smstat['B']))
])
ax1.plot(smstat['time'], Bdoth, 'r-', alpha=0.4)
ax2.plot(smstat['time'], Bdoth, 'r-', alpha=0.4)
ax3.plot(smstat['time'], Bdoth, 'r-', alpha=0.4)
b1.plot(smstat['time'], Bh, 'r-', alpha=0.4)
b2.plot(smstat['time'], Bh, 'r-', alpha=0.4)
b3.plot(smstat['time'], Bh, 'r-', alpha=0.4)
obs20, t20 = tb.windowMean(Bdoth,
time=smstat['time'],
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
bobs20, bt20 = tb.windowMean(Bh,
time=smstat['time'],
winsize=dt.timedelta(minutes=20),
overlap=dt.timedelta(0),
st_time=starttime,
op=np.max)
# Plot on dBdt plots
ax1.plot(t20,
obs20,
marker='x',
color='r',
linestyle='none',
markersize=3,
label='Obs')
ax2.plot(t20,
obs20,
marker='x',
color='r',
linestyle='none',
markersize=3,
label='Obs')
ax3.plot(t20,
obs20,
marker='x',
color='r',
linestyle='none',
markersize=3,
label='Obs')
# Plot on B plots
b1.plot(bt20,
bobs20,
marker='x',
color='r',
linestyle='none',
markersize=3,
label='Obs')
b2.plot(bt20,
bobs20,
marker='x',
color='r',
linestyle='none',
markersize=3,
label='Obs')
b3.plot(bt20,
bobs20,
marker='x',
color='r',
linestyle='none',
markersize=3,
label='Obs')
adjust_plots(ax1, '1-min dB/dt$_{H}$ [nT/s]', Zero=False, xlab=True)
ax1.set_xlabel('Time')
ax1.set_title('Unsmoothed')
adjust_plots(ax2, '1-min dB/dt$_{H}$ [nT/s]', Zero=False, xlab=True)
ax2.set_xlabel('Time')
ax3.set_title('Hourly Smoothed')
adjust_plots(ax3, '1-min dB/dt$_{H}$ [nT/s]', Zero=False, xlab=True)
ax3.set_xlabel('Time')
ax2.set_title('30min Smoothed')
ax1.legend()
ax2.legend()
ax3.legend()
adjust_plots(b1, '1-min B$_{H}$ [nT]', Zero=False, xlab=True)
b1.set_xlabel('Time')
b1.set_title('Unsmoothed')
adjust_plots(b2, '1-min B$_{H}$ [nT]', Zero=False, xlab=True)
b2.set_xlabel('Time')
b3.set_title('Hourly Smoothed')
adjust_plots(b3, '1-min B$_{H}$ [nT]', Zero=False, xlab=True)
b3.set_xlabel('Time')
b2.set_title('30min Smoothed')
b1.legend()
b2.legend()
b3.legend()
#splot.applySmartTimeTicks(ax, subset['time'], dolimit=True)
fig.suptitle(stat)
fig.tight_layout()
fig.subplots_adjust(top=0.88)
fig2.suptitle(stat)
fig2.tight_layout()
fig2.subplots_adjust(top=0.88)
#plt.show()
fig.savefig(
'/Users/sgraf/Desktop/SWMF_analysis/plots/dBdt/{0}_dBdt_comp_obs_{1}.png'
.format(date, stat))
fig2.savefig(
'/Users/sgraf/Desktop/SWMF_analysis/plots/dBdt/{0}_B_comp_obs_{1}.png'
.format(date, stat))
