General
Basic Frontier in Exoplanet Characterization Advanced
Hajime Kawahara
RESCEU Online Summer School 2020 PurposeBasic Why Exoplanets?
Universality of Us and Where We Live in
“Exoplanet Characterization”
Like Earth? Water, oxygen? Has ocean? Comfortable?
Like Jupiter? Very different? 1/36 ContentsBasic 1. What kind of exoplanets can we characterize? Landscape of transit/directly-imaged exoplanets
2. How do we characterize exoplanets? Established techniques for exoplanet characterization
3. How will we characterize exoplanets? On-going and future techniques toward exo Earths
Transit exoplanet Direct imaging Basic Exoplanet landscape Planet temperature 1600K 800K Log Stellar Mass (radius) 100K Hot Jupiter Jupiter & Neptune 1Msun (1Rsun) Super Earth
0.1Msun (0.1Rsun)
0.01 AU 0.1 AU 1 AU Distance from a star (log) Basic Exoplanet landscape (solar-type star by Kepler)
1au 2au 5au 0.39+-0.07 planets/star Hot Jupiter (~0.005 p/s) assuming 100% completeness Consistent with RV 10 Long-period planet in Kepler
Super Earth (~ 0.7 p/s) Kepler original
Planet Radius (Earth) Radius Planet 1 E
1 10 100 1000 Orbital period (d) Kepler prime mission Kawahara and Masuda (2019) Basic Exoplanet landscape (solar-type star)
1au 2au 5au 0.39+-0.07 planets/star Hot Jupiter (~0.005 p/s) assuming 100% completeness Consistent with RV 10 Neptune-sized Planets are common at Jupiter position (>~0.5 p/s)
Super Earth (~ 0.7 p/s) Kepler original
Planet Radius (Earth) Radius Planet 1 E
1 10 100 1000 Orbital period (d) Controversial Kepler prime mission Kawahara and Masuda (2019) Basic Exoplanet landscape Planet temperature 1600K 800K Log Stellar Mass Self-Luminous (radius) Planets 100K Hot Jupiter Jupiter & Neptune 1Msun (1Rsun) Super Earth
Earth@low mass star?
0.1Msun (0.1Rsun)
0.01 AU 0.1 AU 1 AU Distance from a star (log) Basic Exoplanet landscape (schematic) Planet temperature 1600K 800K Log Stellar Mass Self-Luminous Planets (radius) 1000K 100K Hot Jupiter Super Earths Neptune & Jupiter Direct Imaging 1Msun from ground (1Rsun) Earth 1995- Kepler (2009-2013) PLATO
TESS(2018-)
TOI-700 system (TESS)
Exo JASMINE
0.1Msun (0.1Rsun) Trappist-1 system (Ground)
0.01 AU 0.1 AU 1 AU Distance from a star (log) Advanced Fill a gap by Exo JASMINE
TOI-700: Bright but small signal (depth=0.05%)
TESS 0.6-1.0 um 10cm all-sky TESS↓ Exo JASMINE 1.1-1.7um 30cm Grounds <0.9um 1m class
Depth=0.3 – 0.1 % Mag = 8 - 11 Exo JASMINE 探査領域
Trappist system Ground+Spitzer→ Large signal (depth=1%) but faint Basic Exoplanet landscape (schematic) Planet temperature 1600K 800K Log Stellar Mass Self-Luminous Planets (radius) 1000K 100K Hot Jupiter Super Earths Neptune & Jupiter Direct Imaging 1Msun from ground (1Rsun) Earth 1995- Kepler (2009-2013) PLATO
TESS(2018-)
TOI-700 system (TESS)
Exo JASMINE
0.1Msun (0.1Rsun) Trappist-1 system (Ground)
0.01 AU 0.1 AU 1 AU Distance from a star (log) Short Summary Exoplanet landscape
- Jupiter in solar system is universal. Super Earth < 1 AU and cool Neptune @ several AU, which do not exist in solar system, are also common.
- Current astronomers are trying to cheat with Earths around low-mass stars or young self-luminous planets because it’s easy to observe.
- Earth not yet
6/36 Basic Transit or directly-imaged planet exists. So next?
You do not care? If we find exo-Earths, we want to know
Molecules in atmosphere Radius/mass H2O or O2 (biosignature) Surface environment Atmosphere, climate Continents/ocean HD XXXXXX b You might care
Exo Earth not yet. Lesson using other exoplanets
Hot Jupiters and self-luminous planets are mainly used in reality directly-imaged planet Basic Beyond mass/radius
Log Stellar Spectroscopy of star+planet Mass Self-Luminous Planets (radius) 1000K
Hot Jupiter Super Earths Neptune & Jupiter 1Msun (1Rsun) Earth
TOI-700 system (TESS)
0.1Msun (0.1Rsun) Trappist-1 system (Ground)
0.01 AU 0.1 AU 1 AU Distance from a star (log) Basic History/Future of Characterization by Transit Exoplanets
Low-Res Spectroscopy (space)
2001 K, H2O 2014 2017 - 2020 ARIEL ~1m HD209458b Clouds/haze Trappist-1 （２０２８） Detection of Na in Hot Planets 7 rocky planetsJWST 6.5m OST・LUVOIR？ HST, 26 ppm （２０２１） Spitzer 15 ppm
Ground 1.2m High-Res Spectroscopy (ground) CO, H2O, TiO, Metals, HCN, CH4
Space 1 m Transit Survey Heng and Winn 2009 2018 Late 2020-2030 Kepler TESS PLATO (CHEOPS) (Exo JASMINE) Basic Low-R Spectroscopy from space (transmission) Hot Jupiters Transmission
Transit depth
difference< 0.1%
tmosphere tmosphere const.) height (+ a
Insensitive to temperature structure
Relative Relative Sensitive to molecules at higher atmosphere Total modeling required Wavelength (nm) Sing+18 Basic Low-R Spectroscopy from space (emission) Hot Jupiters Emission
star
Integrated blackbody Sensitive to temperature structure Sensitive to molecules at mid atmosphere
Parmentier+18 Basic Atoms/Molecules detection using High-R Spectroscopy
R = λ/Δλ ~ 100,000, Δλ ~ linewidth The best one so far. Hot Jupiter KELT9b Fe+ High-Resolution Spectra (HRS) Stellar : Planet
~ 1000 : 1
orbital phase transit wavelength - ＊ In Theoretical spectrum Molecule-by- Molecule Frames(phase) detection Fe+ wavelength velocity Vr (km/s) Hoeijmakers+2018 ＝ Vr = Kp cos(phase) + V0
Planetary wind
-50 0 50 (km/s) (a few km/s) can CCF CCF signal Plant Radial Velocity be detectable (km/s)
Pro: High-frequency signal is stable Kp even for ground-based observation V0 (km/s) Basic HRS Thermal Inversion by Titanium Oxide
WASP33b emission using Subaru/HDS = star : planet ~ 1000 : 1
upper hotter +
lower hotter -
Nugroho, H.K.+ 2017, AJ Intensity at the line center Ishizuka, H.K.+ in prep comes from upper atmosphere
Titanium oxide absorbs blue light and heats the upper atmosphere, like O3 on Earth. Short Summary LRS from space: metals, H2O, haze, scattering, black body Pros: intuitive, can detect continuum (haze, clouds, scattering) Cons: Total modeling required -> weak for systematics Toward exo Earth: JWST will try to detect molecules in terrestrial planets @ late-M Larger telescope is justice (LUVIOR…)
HRS from ground: CO, H2O, TiO, HCN, CH4, metals, ions, T-P structure Pros: molecules even with many lines, no need of total modeling Cons: connection with model is much more difficult Toward exo Earth: Applicable to water detection(many lines). Larger telescope is justice (TMT, E-ELT)
14/36 Basic History and future of direct imaging 1st gen 2nd gen (AO only) (ExAO+coronagraph) Ground 2007 2014 2018 Space HR8799 51 Eri b PDS70 b 2025 2035-45 Roman HabEx or telescope LUVOIR
Planets by reflection light
Self-luminous young planets (T ~ 1000 K) Basic Why is the detection limited so far? Inner Working Angle ∝ λ/D
Telescope Young hot planet size (emission)
Inst dev Almost no signal
Jupiters to be detectable at any
Earths to be detectable at any
Sudden open like gravitational waves (my hope, not yet) Basic 2nd gen: Adaptive Optics, Coronagraph Phase correction @ pupil
by a deformable mirror
푥
휙
푒
푎
=
푥
퐴
Complex amplitude: Complexamplitude:
oronagraph
Extreme AO Extreme
C ClassicalAO Basic 2nd gen: Adaptive Optics, Coronagraph
pupil Focal plane
pupil Focal plane
푥
휙
푒
푎
=
푥
퐴
Complex amplitude: Complexamplitude:
oronagraph
Extreme AO Extreme
C ClassicalAO Basic Why is direct imaging worth?
Greenbaum+2018 Spectrum Molecules, dynamics, atmospheric structure Biosignature in future
現代の天文学（原稿） Photometric variability Planet Spin/Tilt
Surface Composition, Continents, Ocean in future Basic Low-Resolution Spectroscopy
Greenbaum+2018
CO? Lacour+2019
Late L Brown Dwarf model well fitted? (covered by clouds) Advanced High-Dispersion Coronagraphy
Classical HRS for direct imaging HRS + ExAO + Coronagraph (Snellen+14, Hoeijimakers+18) = High-Dispersion Coronagraphy (Kawahara+14, Snellen+15, Wang+17)
REACH on Subaru (Y,J,H) KPIC on Keck (K,L,M) HiRISE on VLT (planned, H,K)
Star: planet = 300 : 1 CCF map of water (Hoeijimakers+18)
Slit or IFU
High-Resolution Single mode Spectrograph Fiber Injection
CCF of CO (Snellen+18) REACH collaboration
Sebastien Vievard
Olivier Guyon Julien Lozi Advanced High-Dispersion Coronagraphy on Subaru
REACH on Subaru telescope
Extreme AO Coronagraph
High-Resolution Spectrograph
IRD
SCExAO Reference laser Photometric monitoring
2014 idea (Kawahara+2014) 2015-2017 Supported by RESCEU 2019 first light (engneering), open use (since S20B) 2020 Kiban A, first science obs (Aug 1-6 2020) Advanced HDC significantly improves S/N
classical AO v.s. ExAO+Coronagraph Comparison of Size on the focal plane Speckle contrast to a host star -1 HR8799 ecd CLASSICAL HRS 10 IRD (MMF) CRIRES+ slit -2 10 0.2’’ 10-3 φ0.48’’ REACH Injection 10-4 e d 10-5 c 0.5” 1” Star-planet separation
ExAO + Coronagraph + Small Injection can significantly increase planet signal in R=100,000 spectrum (0.95-1.75 um) Advanced Power of REACH First Engineering run (2019): Spectroscopic binary 1 (SB1)
Image of a binary before correction
Image of a binary after correction
First science run ( this month! ) HR8799e 5.5 hours pilot data. Now processing, we’ll see…
Contrast = 3 x 10-4 -> planet : star = 1 : 10
S21A proposal welcome! Short Summary Direct imaging Ground: self-luminous planets so far
Pros: cutting-edge technique can be tested by a small group. Cons: It will not reach Earth@solar-type star Toward exo Earth: 30-40 m class telescope will try to detect water/oxygen in terrestrial planets @ late-M
Direct imaging from space: not yet. Roman will open, HabEx, LUVIOR
Pros: No ExAO needed, it can reach Earth@solar-type star, it can detect oxygen, water. Cons: dev by a small group with try-and-error is almost impossible It’s not clear except for low-res spectroscopy -> Methodology for space DI should be explored more.
25/36 Basic Photometric variability as a new probe of exoplanets Space Direct Imaging: Roman 2025-, HabEx/LUVOIR 203X- Reflection light from exoplanets around solar-type stars will be available - Low resultion spectroscopy - Time-series analysis
Variability of 2M1207b by HST (Zhou+ 2016) How do we disentangle spatial and spectral information Basic from time-series data?
Spin-Orbit Tomography given given albedo map (we want to know) 푑 푡 = ∫ 푊 t, Ω a Ω dΩ
Spin-Orbit Unmixing Dynamic Mapping 푑 푡, 휆 = ∫ ∫ 푊 t, Ω A Ω, 푘 X(푘, 휆)d휆dΩ 푑 푡 = ∫ 푊 t, Ω A t, Ω dΩ
k-th component Map of k-th Time-dependent spectrum component albedo Dynamic Unmixing 푑 푡, 휆 = ∫ ∫ 푊 t, Ω 퐴 t, Ω, 푘 X(푘, 휆)d휆dΩ Basic Gallery
2010 Cloudless simulation 2012 full simulation 2019 REAL EARTH Kawahara & Fujii 2010 K&F 2011, F&K 2012 single band (TESS) Luger+2019
2019 REAL EARTH (L2/PCA) 2020 REAL EARTH (NMF) 2020 REAL EARTH (clouds) Fan+2019 Kawahara 2020 Kawahara & Masuda 2020 Basic Linear Inverse Problem
푑 푡 = ∫ 푊 t, Ω a Ω dΩ → 풅 = 푊풂
Singular Value Decomposition 푊 = 푈푇Λ V U,V: orthogonal matrix Moore-Penrose Matrix 푊−푔 = 푉 Λ−1 UT p
Very unstable for small singular value 휿풊
Tikhonov regularization “L2” norm regularization
minimize
The math structure is clear, but needs to determine an arbitrary parameter. Tarantra 2005, Bishop 2006, Farr+2018 AdvancedBasic Bayesian LIP Kawahara and Masuda 2020
Linear Inverse Problem with known covariance matrices likelihood
prior
Posterior:
In general: Nonlinear parameters (NLPs) in 푊 = 푊 품 : axial tilt, spin rate Covarianceare modeled through a Gaussian process with hyperparameters:
Posterior of NLPs given by MCMC: Posterior of a map: Sampled NLPs Advanced Linear Inverse Problem + Matrix Factorization Kawahara 2020
푑 푡, 휆 = ∫ ∫ 푊 t, Ω A Ω, 푘 X(푘, 휆)d휆dΩ
Volume Regularization Nonnegative MF
X(3,:) X(2,:)
X(1,:) Advanced Dynamic Mapping Kawahara and Masuda 2020
흎푖: 푖-th column of 푊෩ : Diagonal matrix from a vector 풚
Remark: We know a posterior of 풂. But, we need to reduce memory size and computational complexity.
The Kronecker-product type kernel and isomorphic transformation enable us to do that.
Example: mean of a posterior of A
Data Model ⊙= element-wise product N~1,000 M~1,000,000 Advanced Dynamic Mapping of Real data
5% credible boundary
Cloud fraction in September 2016 median
95% credible boundary
correlation Kernel correlation Noise level spatial scale intensity time scale SummaryBasic Landscape to characterization Planet temperature 1600K 800K Log Stellar LRS (space):Mass Self-Luminous Planets (radius) simple atoms, 1000K clouds/Haze, 100K Black body Hot Jupiter Super Earths Neptune & Jupiter 1Msun HRS (ground): (1Rsun) Earth molecules, atoms LRS: Complex lines CO, clouds/Haze T-P structure HRS: complex molecules, We’ll see Photometric Variation: TOI-700 system2D image:(TESS) Continents/ocean 2D+time lapse (clouds)
0.1Msun (0.1Rsun) Trappist-1 system (Ground)
0.01 AU 0.1 AU 1 AU Distance from a star (log)