The ISM Beyond 3D: Promises & Problems

Slide 1

Slide 1 text

The ISM Beyond 3D: Promises & Problems Joshua E. G. Peek Space Telescope Science Institute July 3, 2017 ISM3D+ Look up here if you get lost! Kirill Tchernyshyov Johns Hopkins Eddie Schlaﬂy LBNL @jegpeek @jegpeek AMA GALFA-HI DR2!

Slide 2

Slide 2 text

Dust & Tomography Emission Maps Beyond 2D The 3rd Dimension Beyond the 3rd Dimension Promises

Slide 3

Slide 3 text

Astronomy has always been a science of catalogs Ptolemy 155

Slide 4

Slide 4 text

1% 4% 23% 72% Dark Energy Dark Matter Plasma & Gas Stars etc. Text Unfortunately, most of the Universe is diﬀuse

Slide 5

Slide 5 text

Text 60s 70s 80s 90s 00s 2% 20% 10% Star & Formation / Star Galaxy & Formation / Galaxy h/t SAO/NASA ADS Everything forms from diﬀuse material, a topic of interest

Slide 6

Slide 6 text

We study the diﬀuse universe through absence in catalogs ESO 99; Lan, Ménard, & Zhu 15 99% 100% 99% 100% 99% 100% Balmer β DIBs in LMZ15 Stellar and Sky Residuals 5150Å 4350Å 5150Å 5950Å Normalized ﬂux Mgb [OI] 5577 NaD Synthetic DIB spectrum 5950 Å 6700 Å [OI] 6300 [OH] Balmer α

Slide 7

Slide 7 text

Dust & Tomography Emission Maps Beyond 2D The 3rd Dimension Beyond the 3rd Dimension Promises

Slide 8

Slide 8 text

Dust & Tomography Emission Maps Beyond 2D The 3rd Dimension Beyond the 3rd Dimension Promises

Slide 9

Slide 9 text

Interstellar metals are in dust, tracing gas and evolution Periodic Elements of Dust 4 Solar Fe Mg Si N Ne C O (99%) + many other trace elements (S,P,Ca,Cl,Ti) Solar Metals slide courtesy L. Corrales

Slide 10

Slide 10 text

1000 10000 wavelength[Å] 6 5 4 2 1 0 3 A O / AV Dust grains extinguish light more towards the blue

Slide 11

Slide 11 text

u g r i z u g r i z W1 W1 F F N N 1000 10000 wavelength[Å] 6 5 4 2 1 0 3 A O / AV MW dust grains can mostly be parameterized by Rv Reddening is often reduced to a single parameter family, Rv

Slide 12

Slide 12 text

Text 2D→3D Classic tomography (medical) uses slices of data

Slide 13

Slide 13 text

Text 2D→3D Classic tomography uses slices of data (CT, PET, MRI)

Slide 14

Slide 14 text

Text 0D→3D ć+ 2008–2012 Stellar tomography ﬁnds density from points

Slide 15

Slide 15 text

Text 1D→3D Vergely+ 2011 Na I Gas tomography uses absence along the line of sight

Slide 16

Slide 16 text

Text 1D→3D Vergely+ 2011 E(B-V) Gas tomography uses absence along the line of sight …so what?

Slide 17

Slide 17 text

“…so what?” “There are a number of specific aims one could adopt for 3d extinction mapping…” Sale 2015 “…The most obvious is to infer the extinction to stars that feature in the ‘input catalogue’. “ “One might also be interested in the estimating the extinction to some arbitrary point or points in space.” Pays the Bills …since extinction aﬀects the light from stars distributed throughout the Galaxy it oﬀers a direct route to studying the elusive 3d structure of the ISM. Is the Money

Slide 18

Slide 18 text

Dust & Tomography Emission Maps Beyond 2D The 3rd Dimension Beyond the 3rd Dimension Promises

Slide 19

Slide 19 text

Dust 90 180 270 90 180 270 0.33 30. MJy/sr Log scale FIG. 8.ÈFull-sky dust map for the NGP (top) and SGP (bottom) Our best high latitude dust maps come from FIR emission Schlegel, Finkbeiner, & Davis 98 Dust 90 180 270 90 180 270 . 0 3 3 3 . 0 MJy/sr Log scale F IG. 8.ÈFull-sky dust map for the NGP ( top) and SGP ( bottom )

Slide 20

Slide 20 text

Dust map errors distort LSS results via sample biasing Huterer, Cunha, and Fang 12 0.1 0.3 1 3 Faint-end slope of luminosity func, s(z) 10-2 10-1 100 101 102 103 bias / error For Peek-Graves 2010 correction to dust map ‘non-Gaussianity’ ‘dw/dt’ target precision For known errors to dust map Redshift of galaxies used for cosmology ~0 0.7 1.9 3.7

Slide 21

Slide 21 text

Standard Candle: near far Standard Crayon: clear obscured “Standard Crayon” methods measure reddening directly Standard Ruler: near far

Slide 22

Slide 22 text

JEGP & Graves 10 Quiescent galaxies can be selected spectroscopically.

Slide 23

Slide 23 text

JEGP & Graves 10 Residual colors are very, very small and unimodal σ δ g-r

Slide 24

Slide 24 text

δE (B − V ) [magnitudes] error[σ] l=0 l=270 l=180 l=90 Cryaons provide corrections to the SFD98 reddening map. JEGP & Graves 10

Slide 25

Slide 25 text

But is FIR actually better than HI? Schlegel, Finkbeiner, & Davis 98

Slide 26

Slide 26 text

The “Yahata Eﬀect” is a fundamental ﬂoor for FIR Yahata+ 2007

Slide 27

Slide 27 text

SFD LAB PG10 PG10-SFD PG10-LAB SFD-LAB >0.1 >0.1 >1021 0 0 0 >0.015 >0.015 >0.015 <-0.015 <-0.015 <-0.015 E (B − V ) E (B − V ) E (B − V ) E (B − V ) E (B − V ) NHI cm−2 HI methods and FIR methods give ~similar precision JEGP 13 Extinction HI FIR Extinction HI FIR

Slide 28

Slide 28 text

HI + FIR method is signiﬁcantly better as errors cancel JEGP 13 Extinction HI FIR cHI cF IR Fraction HI Fraction FIR

Slide 29

Slide 29 text

Making the best HI maps velocity information matters Lenz, Hensley & Doré 2015

Slide 30

Slide 30 text

-90 < HI < 90 , HI < 4 x 1020: HI does a great job Lenz, Hensley & Doré 2015

Slide 31

Slide 31 text

Does 3D, thermal, shape information matter?

Slide 32

Slide 32 text

8 9 10 11 12 13 2005 2010 2015 2020 2025 2030 Huge Galactic HI surveys are on the horizon Log[ # resolution elements] LAB GASS GALFA-HI EBHIS APERTIF/ASKAP SKA surveys? surpasses Planck resolution

Slide 33

Slide 33 text

Dust & Tomography Emission Maps Beyond 2D The 3rd Dimension Beyond the 3rd Dimension Promises

Slide 34

Slide 34 text

Text PS1 APOGEE GAIA “AS4” LSST grizy R~23,000 R ~100 R~23,000 ugrizy 8 x 108 stars 1.5 x 105 stars 2 x 108 stars 5 x 106 stars 140 x 108 stars ~no parallax ~no parallax precise parallax ~no parallax some parallax now now April 2018? 2020s 2020s We are in the era of amazing tomography machines 䢀Available in MAST today!

Slide 35

Slide 35 text

This is big business and growing! Sale 2015

Slide 36

Slide 36 text

Signiﬁcant methodological advances are happening too S. Rezaei Kh. + 2016, Sale 2015 Gaussian Processes Poisson Point Processes

Slide 37

Slide 37 text

Text PS1 dust tomography ﬁts E(B-V), distance, and stellar type Green+ 2014 PS1 Absolute Magnitude

Slide 38

Slide 38 text

Text By combining many PDFs, get an E(B-V) by distance Green+ 2014 ~ 1000 stars / 6’ pixel 100 pc 10 kpc 1 kpc …so what? PS1

Slide 39

Slide 39 text

Text 3D dust maps provide clean distances to molecular clouds Schlafly+ 2014 PS1

Slide 40

Slide 40 text

Dust & Tomography Emission Maps Beyond 2D The 3rd Dimension Beyond the 3rd Dimension Promises

Slide 41

Slide 41 text

Text Majewski+ 2015 APOGEE: R~23,000 of 150,000 red giants, ~in the plane High Resolution Spectroscopy → Very precise stellar parameters → Very precise unreddened colors → Very precisely measured reddening APOGEE

Slide 42

Slide 42 text

Text R(V) variation can be measured in detail with APOGEE Schlafly+JEGP,Tchernyshyov+ 2015 Principal Component 1 Principal Component 2 APOGEE

Slide 43

Slide 43 text

Text R(V) has strange large scale structure related to Planck Beta… Schlafly+JEGP,Tchernyshyov+ 2015 Sagittarius arm Orion-Cygnus / Local Arm I⌫ / ⌫ B⌫ (T)

Slide 44

Slide 44 text

R(V) variation is driven by location, not density! zut! Schlafly, JEGP, Finkbeiner, & Green 2017 ; Reid 2014

Slide 45

Slide 45 text

R(V) variation is driven by location, not density! zut! Schlafly, JEGP, Finkbeiner, & Green 2017 ; Reid 2014

Slide 46

Slide 46 text

Text 60s 70s 80s 90s 00s 2% 20% 10% Star & Formation / Star Galaxy & Formation / Galaxy h/t SAO/NASA ADS But what about formation??

Slide 47

Slide 47 text

t + · ( v) = 0 The continuity equation is how we get at formation

Slide 48

Slide 48 text

Text We add the PPV data in HI+CO, which probes similar gas PS1 PPD HI PPV CO PPV Green+ 2014; McClure-Griﬃths+ 2005; Peek+ 2011; Kalberla+ 2005; Dame+ 2001; Tchernyshyov & Peek 2017 PS1

Slide 49

Slide 49 text

radial velocity distance radio ISM surveys alone radial velocity distance stellar reddening alone radial velocity distance KT: Static Cloud radial velocity KT: Converging Cloud distance radial velocity KT: Diverging Cloud distance The D/V diagram is the killer app for formation PS1

Slide 50

Slide 50 text

Text But how do we associate tomography data with gas? 220 200 180 160 140 Galactic longitude -40 -20 0 20 40 Velocity [km/s LSR] 0.1 1 10 Distance [kpc] CO [Dame et al] Extinction [Schlafly/Green] PS1

Slide 51

Slide 51 text

Text We start with the dust map Tchernyshyov & Peek 2017 PS1

Slide 52

Slide 52 text

Text We start with the dust map PS1 Tchernyshyov & Peek 2017

Slide 53

Slide 53 text

Text We start with the dust map PS1 Tchernyshyov & Peek 2017

Slide 54

Slide 54 text

Text But how do we construct the D/V diagram? PS1 Tchernyshyov & Peek 2017

Slide 55

Slide 55 text

Text The Galactic rotation curve provides a strong prior PS1 Tchernyshyov & Peek 2017

Slide 56

Slide 56 text

Text The Galactic rotation curve provides a strong prior PS1 Tchernyshyov & Peek 2017

Slide 57

Slide 57 text

Text Other rotation curves generate changes in radial velocity PS1 Tchernyshyov & Peek 2017

Slide 58

Slide 58 text

Text We allow each distance deviate in radial velocity to ﬁt PS1 Tchernyshyov & Peek 2017

Slide 59

Slide 59 text

Text We apply Markov “springs”, to add image-plane information PS1 Tchernyshyov & Peek 2017

Slide 60

Slide 60 text

Text We apply Markov “springs”, to add image-plane information Longitude Latitude PS1 Tchernyshyov & Peek 2017

Slide 61

Slide 61 text

Text So we go from this… PS1 Tchernyshyov & Peek 2017

Slide 62

Slide 62 text

Text To our markov spring all-disk ﬁt PS1 Tchernyshyov & Peek 2017

Slide 63

Slide 63 text

Text Masers provide sparse, model-free distances and velocities Reid+ 2014 X (kpc) Y (kpc)

Slide 64

Slide 64 text

KT does an excellent job recovering Masers and PPV Observed Flat PS1 Tchernyshyov & Peek 2017

Slide 65

Slide 65 text

KT does an excellent job recovering Masers and PPV Observed Clemens (1985) Clemens (1985) PS1 Tchernyshyov & Peek 2017

Slide 66

Slide 66 text

KT does an excellent job recovering Masers and PPV Observed KT w/o springs KT w/o springs PS1 Tchernyshyov & Peek 2017

Slide 67

Slide 67 text

KT does an excellent job recovering Masers and PPV Observed KT w/ springs KT w/ springs PS1 Tchernyshyov & Peek 2017

Slide 68

Slide 68 text

KT does an excellent job recovering Masers and PPV PS1 Tchernyshyov & Peek 2017

Slide 69

Slide 69 text

Text 60s 70s 80s 90s 00s 2% 20% 10% Star & Formation / Star Galaxy & Formation / Galaxy h/t SAO/NASA ADS But Josh! what about formation??

Slide 70

Slide 70 text

Are shocks key to forming H2 clouds in the Milky Way? Roberts 1969; Shu 2016 ⇢1v1 = ⇢2v2

Slide 71

Slide 71 text

Shocks do seem to exist in spirals like M81 and M51… Visser 1980 M 81

Slide 72

Slide 72 text

Shocks are also invoked to explain arm feathering Kim & Ostriker 2002 Σ/Σ₀ W [km/s] 26 13 0 0 0 0.5 -0.5 2 4 6 X/L x

Slide 73

Slide 73 text

Text We do not see spiral shocks in the MW… yet Roberts 1972, Tchernyshyov & Peek in prep Two Armed Spiral Shock (seen in M81, M51) Linear Density Wave PS1 ũ = 110 Data from TP17

Slide 74

Slide 74 text

Shock models predict unseen sharp velocity changes PS1 Tchernyshyov & Peek 2017

Slide 75

Slide 75 text

PS1 Shock models predict unseen sharp velocity changes Tchernyshyov & Peek 2017

Slide 76

Slide 76 text

Text Gas emission + dust reddening KT has some ﬂaws… Map tested only at HMSFRs 0.1 1.0 10.0 Aquila-Ophiuchus CO + HI = NH (1021 H/cm2) 0.1 1.0 10.0 AV gas at Av=2 is not CO + HI “Dark Gas” PS1 … to be continued

Slide 77

Slide 77 text

Standard Crayons are a powerful tool for the precision study of dust, crucial for cosmology R(V) is not simply a proxy for gas density; 3D structure influences R(V), for unknown reasons The Perseus arm does not seem to be a spiral shock, although we need better data… Modern tomography uses big data and new models To answer questions of formation we need to measure where gas is and how it moves: D/V diagrams

Slide 78

Slide 78 text

Problems Hypothesis Generation Hypothesis Testing

Slide 79

Slide 79 text

Problems Hypothesis Generation Hypothesis Testing

Slide 80

Slide 80 text

Text The ISM Beyond 3D… has more than three dimensions Schlafly+JEGP,Tchernyshyov+ 2015 I

Slide 81

Slide 81 text

Text Tomographic data can have very complex biases GC 20 15 10 5 -10 -5 0 5 10 350 0 galactic latitude galactic longitude 355 3.1–5.0 kpc 2.0–3.1 kpc 1.3–2.0 kpc ~500pc

Slide 82

Slide 82 text

Both problems gets worse beyond 3D Tchernyshyov & Peek 2017

Slide 83

Slide 83 text

Problems Hypothesis Generation Hypothesis Testing

Slide 84

Slide 84 text

How I was taught to science Observation Pure Theory Interpretation Prediction ✅ ❌ "I know where the information is”

Slide 85

Slide 85 text

How the hip kids science now Observation Pure Theory Data Simulation Prediction More Hip Less Hip

Slide 86

Slide 86 text

The ISM is very, very hard to simulate

Slide 87

Slide 87 text

It’s even hard to interpret: where is the information? 24o 26o 28o 15h00m 14h32m 14h04m 13h36m 24o 26o 28o Right Ascension n o i t a n i l c e D 24o 26o 28o 15h00m 14h32m 14h04m 13h36m 24o 26o 28o Right Ascension n o i t a n i l c e D Clark, JEGP, Putman 14 Declination 24o 26o 28o 15h00m 14h32m 14h04m 13h36m 24o 26o 28o Right Ascension n o i t a n i l c e D

Slide 88

Slide 88 text

The ISM especially hard to interpret in 3D… and beyond Green, Schlafly, & Finkbeiner 2015 …since extinction affects the light from stars distributed throughout the Galaxy it offers a direct route to studying the elusive 3d structure of the ISM. Sale+2015

Slide 89

Slide 89 text

What are our tools? What are we looking for? What are our metrics? What are our models?