Hannu Parviainen, Instituto de Astrofísica de Canarias
This notebook works as an appendix to Parviainen et al., Ground based transmission spectroscopy of WASP-80b (2017). The paper covers two analyses: a broadband analysis using three previously published datasets, and a transmission spectroscopy analysis using two GTC-observed spectroscopic time series, and this notebook covers a part of the broadband analysis.
Last (significant) revision: 11.08.2017
The broadband analysis uses three broadband datasets by Triaud et al. (T13, 2013), Fukui et al. (F14, 2014), and Mancini et al. (M14, 2014), containing 27 light curves observed in g', r', i', z', I, J, H, K. The T13 and M14 datasets contain only the time, flux, and flux uncertainty estimates, while the F14 dataset contains also a set of useful covariates, such as the airmass and FWHM estimates.
We start the broadband analysis by homogenising the three datasets into a common format, and store the data in a HDF5-file DDATA/external_lcs.h5. The F14 data is stored as Pandas DataFrames, and the T13, and M14 light curves as Pandas Series. The structure of the file is
/lc/fukui2014/H/irsf_130716 frame (shape->[537,7])
/lc/fukui2014/H/irsf_130822 frame (shape->[576,7])
/lc/fukui2014/H/irsf_131007 frame (shape->[597,7])
/lc/fukui2014/J/irsf_130716 frame (shape->[541,7])
/lc/fukui2014/J/irsf_130822 frame (shape->[566,7])
/lc/fukui2014/J/irsf_131007 frame (shape->[597,7])
/lc/fukui2014/J/isle_130813 frame (shape->[324,7])
/lc/fukui2014/J/isle_130922 frame (shape->[241,7])
/lc/fukui2014/K/irsf_130716 frame (shape->[539,7])
/lc/fukui2014/K/irsf_130822 frame (shape->[582,7])
/lc/fukui2014/K/irsf_131007 frame (shape->[603,7])
/lc/fukui2014/g/oao50_130813 frame (shape->[562,7])
/lc/fukui2014/g/oao50_130922 frame (shape->[268,7])
/lc/fukui2014/i/oao50_130813 frame (shape->[563,7])
/lc/fukui2014/i/oao50_130922 frame (shape->[295,7])
/lc/fukui2014/r/oao50_130813 frame (shape->[566,7])
/lc/fukui2014/r/oao50_130922 frame (shape->[290,7])
/lc/mancini2014/H/grond series (shape->[508])
/lc/mancini2014/J/grond series (shape->[508])
/lc/mancini2014/K/grond series (shape->[508])
/lc/mancini2014/g/grond series (shape->[162])
/lc/mancini2014/i/dfosc series (shape->[156])
/lc/mancini2014/i/grond series (shape->[156])
/lc/mancini2014/r/grond series (shape->[156])
/lc/mancini2014/z/grond series (shape->[157])
/lc/triaud2013/r/eulercam_1 series (shape->[208])
/lc/triaud2013/z/trappist_1 series (shape->[721])
/lc/triaud2013/z/trappist_2 series (shape->[704])
/transmission frame (shape->[2151,7])
Note: We also store the transmission as a function of wavelength for each passband in the HDF5-file, under /transmission, and simplify the analysis a bit by merging the i' and I bands: the average transmission is used for both passbands.
In [1]:
%pylab inline
%run __init__.py
In [2]:
import astropy.io.fits as pf
import pandas as pd
import seaborn as sb
from glob import glob
from os.path import basename, splitext, join
from astropy.table import Table
from exotk.utils.misc import fold
from src.extcore import TC, P, TZERO, DDATA
In [3]:
fnames = sorted(glob(join(DDATA,'fukui_2014/[i,o]*dat')))
hnames = list(map(lambda s: 'lc/fukui2014/{t[1]:s}/{t[0]:s}_{t[2]:s}'.format(t=splitext(basename(s))[0].split('-')), fnames))
hnames = list(map(lambda s: s.replace('/I/','/i/').replace('/G/','/g/').replace('/R/','/r/'), hnames))
In [4]:
dfs = {}
for fname,hname in zip(fnames,hnames):
df = pd.read_csv(fname, sep=' ')
df = pd.DataFrame(df.values[:,:-2], columns=df.columns[1:-1])
stime = pd.Series(df.iloc[:,0].values + 2450000, name='bjd', dtype=np.float64)
df = pd.DataFrame(df.values[:,1:], index=stime, columns=df.columns[1:], dtype=np.float64)
ph = fold(df.index.values, P, TC+TZERO, normalize=True, shift=0.5) - 0.5
df['oe_mask'] = abs(ph) < 0.015
df.flux = df.flux / median(df.flux[~df.oe_mask])
df.to_hdf(join(DDATA,'external_lcs.h5'), hname)
dfs[hname] = df
In [5]:
fig,axs = subplots(6,3,figsize=(14,16), sharey=True, sharex=True)
for k,ax in zip(sorted(dfs.keys()),axs.flat):
df = dfs[k]
ph = fold(df.index.values, P, TC+TZERO, normalize=True, shift=0.5) - 0.5
fl = df.flux.values
ax.plot(ph, fl, '.')
ax.text(0.5,0.9,k.split('2014/')[1].replace('/',' - '), ha='center', transform=ax.transAxes)
[ax.axvline(v, ls='--', c='k', lw=1, alpha=0.5) for v in [-0.015,0.015]]
setp(axs, ylim=(0.93,1.03), xlim=(-0.04,0.04), xticks=linspace(-0.03,0.03,5), yticks=linspace(0.95,1.01,7))
fig.tight_layout()
axs.flat[-1].set_visible(False)
Here we prepare the TRAPPIST and EulerCam light curves by Triaud et al. (2013) (Obtained from Vizier, http://vizier.u-strasbg.fr/viz-bin/VizieR?-source=J/A+A/551/A80&-to=3)
The data contains two TRAPPIST z' light curves and an Euler r' light curve.
In [6]:
df = Table.read(join(DDATA,'triaud_2013_trappist.fits')).to_pandas()
df.columns = map(str.lower, df.columns)
In [7]:
mz1 = df.bjd < 6070
mr1 = (df.bjd < 6150) & (~mz1)
mz2 = (~mz1) & (~mr1)
df.bjd += 2450000
In [8]:
fz1 = pd.Series(df[mz1].flux.values, index=df[mz1].bjd.values, name='z_trappist_1')
fz2 = pd.Series(df[mz2].flux.values, index=df[mz2].bjd.values, name='z_trappist_2')
fr1 = pd.Series(df[mr1].flux.values, index=df[mr1].bjd.values, name='r_eulercam_')
In [9]:
with pd.HDFStore(join(DDATA,'external_lcs.h5')) as f:
f.put('lc/triaud2013/z/trappist_1', fz1)
f.put('lc/triaud2013/z/trappist_2', fz2)
f.put('lc/triaud2013/r/eulercam_1', fr1)
In [10]:
fig, axs = subplots(1,3,figsize=(14,4), sharey=True)
fz1.plot(ax=axs[0], style='.')
fz2.plot(ax=axs[1], style='.')
fr1.plot(ax=axs[2], style='.')
fig.tight_layout()
In [11]:
DFOSC = ['I']
GROND = 'g r i z J H K'.split()
In [12]:
dfm = Table.read(join(DDATA, 'mancini_2014.fits')).to_pandas()
dfm.columns = map(str.lower, dfm.columns)
dfm.bjd += 2400000
dfm['flux'] = 10**(-dfm.mag/2.5)
dfm.drop(['mag','e_mag'], inplace=True, axis=1)
dfm.sort_values(['band','bjd'], inplace=True)
In [13]:
fig,axs = subplots(3,3, figsize=(14,10), sharey=True, sharex=True)
for i,band in enumerate('g r i I z J H K'.split()):
dft = dfm[dfm.band==band]
axs.flat[i].plot(dft.bjd, dft.flux, '.')
setp(axs, xlim=[dfm.bjd.min(),dfm.bjd.max()], ylim=(0.95, 1.03))
fig.tight_layout()
In [14]:
with pd.HDFStore(join(DDATA, 'external_lcs.h5'), 'a') as f:
for band in dfm.band.unique():
inst = 'grond' if band in GROND else 'dfosc'
band = band.replace('I','i')
name = 'lc/mancini2014/{band:s}/{name:s}'.format(band=band,name=inst)
dft = dfm[dfm.band==band]
f.put(name, pd.Series(dft.flux.values, dft.bjd.values, name=name))
In [15]:
def N(a):
return a/a.max()
def favg(filters):
return N(sum(list(map(N, filters)), 0))
dff = pd.read_csv(join(DDATA, 'fukui_2014', 'TM_all.dat'), sep=' ')
dff = pd.DataFrame(dff.values[:,1:-1], columns=dff.columns[2:], index=pd.Series(dff.values[:,0], name='Wavelength'))
dfg = pd.read_csv(join(DDATA, 'GROND_filters.txt'), sep='\t', index_col=0)
dfg.index.name = 'Wavelength'
filter_names = 'g r i z J H K'.split()
transmission = array([favg([dfg.gBand, dff.MITSuME_g]),
favg([dfg.rBand, dff.MITSuME_Rc]),
favg([dfg.iBand, dff.MITSuME_Ic]),
N(dfg.zBand),
favg([dfg.JBand, dff['2MASS_J'], dff.IRSF_J]),
favg([dfg.HBand, dff.IRSF_H]),
favg([dfg.KBand, dff.IRSF_Ks])]).T
ddf = pd.DataFrame(data=transmission, columns=filter_names, index=dfg.index)
ddf.to_hdf(join(DDATA, 'external_lcs.h5'), 'transmission')
ddf.plot();
In [16]:
with pd.HDFStore(join(DDATA, 'external_lcs.h5')) as f:
print(f)