Future AO Legacy HI line surveys: Future AO Legacy HI line surveys: Synergies with other surveysSynergies with other surveys

Martha HaynesMartha HaynesCornell UniversityCornell University

Frontiers of Astronomy with the Frontiers of Astronomy with the World’s Largest Radio TelescopeWorld’s Largest Radio Telescope

September 12-13, 2007September 12-13, 2007

Thanks to many people whose slides/diagrams/ideas have been borrowed for

this presentation

Topics in this talkTopics in this talk

•What are the main questions in cosmology and galaxy evolution that HI science can address?

•What is the current state of extragalactic HI science of relevance to those topics?

•How do Exgal/Cosmo HI surveys fit in with surveys at other wavelengths?

•What other radio facilities are/will be available?

•How might AO contribute uniquely to advancing the field?

2nd+ generation HI surveys2nd+ generation HI surveys• In comparison with opt/IR, the HI view is largely immature• HIMF based only only few thousand objects (HIPASS; SFI++),

whereas O/IR LF is based on hundreds of thousands to millions of objects!

• “Missing Satellite Problem”: • HIMF at low masses• Halo occupation number

• Clustering of gas-rich galaxies:• Correlation functions: HI-HI, HI-opt, HI-IR• Bias parameter• “Void problem”• Dark energy constraints from baryon acoustic

oscillation peaks determined with redshifts• Mass assembly history of galaxies: HIMF across the

masses

Four published results1. Eisenstein et al 2005 3D map from SDSS 46,000 galaxies in 0.72 (h-1Gpc)3

2. Cole et al 2005 3D map from 2dFGRS at AAO 221,000 galaxies in 0.2 (h-1Gpc)3

3. Padmanabhan et al 2007 Set of 2D maps from SDSS 600,000 galaxies in 1.5 (h-1Gpc)3

4. Blake et al 2007 (Same data as above)

Current State of the Art in Baryon Acoustic Oscillations (BAO)

Thanks to Pat McDonald (CITA)

AAO 4-m telescope at Siding Spring, Australia

SDSS 2.5-m telescope, Apache Point, NM

(spectro-z)5%

(spectro-z)3%

(photo-z)5%

HI surveys are woefully behind in

numbers of detections

a.k.a.: SDSS-3

JDEM: ADEPT conceptJDEM: ADEPT concept

Warren Moos: presentation to BEPAC

OIR Spectroscopic BAO surveysOIR Spectroscopic BAO surveys

Warren Moos: presentation to BEPAC

How and when do galaxies form?How and when do galaxies form?Numerical

simulations are used to trace the

gravitational collapse of matter (dark+luminous)

across cosmic time

The “Missing Satellite Problem”The “Missing Satellite Problem” • Models/simulations predict large

numbers of satellites => Logarithmic slope of the faint end of the CDM mass function ~ -1.8 (Press-Schechter value)

• Kauffmann et al. (1993)• Klypin et al. (1999)

• But the current census does not count them (light not mass):

• Faint end slope of the optical LF• Faint end slope of the HIMF

• But, is there anything we can detect?• Baryon loss during reionization ( e.g.,

Efstathiou 1992; Barkana & Loeb 1999; Shaviv & Dekel 2003)

• Can they (ever) form stars? (Verde et al. 2002)

The HI Mass FunctionThe HI Mass Function

N=1000

?

Parkes HIPASS survey: Zwaan et al. 2003

•Previous surveys have included few (if any) objects with HI masses less than 108 M.

•At lowest masses, differ by 10X:

Rosenberg & Schneider (2000)

versus

Zwaan et al. (1997)

•Statistics •Systematics

HIMF @ z=0 Challenges Challenges •Need better statistics: larger, more sensitive surveys

•At the faint end, all the galaxies are nearby• Redshift distances are highly unreliable• LSS affects accuracy of flow models

Masters, H & G 2004, ApJ 607 L115

•Need a “fair sample” to overcome (and allow study of) cosmic variance

• Σ(1/Vmax) corrections must account for LSS• Not just that space density varies with distance• Fractional volume of space occupied by regions of a

particular density do tooSpringob, H & G 2005, ApJ 621, 215

• Other methods (e.g. 2DSWML) do not give normalization

Statistics, statistics, statisticsStatistics, statistics, statistics

Springob et al. 2005(optically selected)

N=2800

N=265

Rosenberg & Schneider 2002

Cosmic varianceCosmic variance

Must sample enough volume to acquire a “fair sample”

If we covered a similar slice in the opposite part of the sky (coming….) we would see a very DIFFERENT redshift distribution => LSS!!!

At these distances, 540 square degrees is not enough.

Environment & the HIMFEnvironment & the HIMFPrevious studies based only on Virgo have suggested that the HIMF in Virgo is flatter than in the field

• Only a single cluster• Very small number statistics/systematics vs

comparison• Is this just HI deficiency?• Watch out for morphological biases

Kovač, Oosterloo & van der Hulst (2005): CanVen• Similar to Virgo (low mass slow flatter)

BUT……..

Zwaan et al. (2005): HIPASS• Higher density regions => more low masses

Inconsistency:Symptom of inadequate

volume?

Springob, Haynes & Giovanelli (2005)• Much larger sample, optically targeted• HI flux and diameter limited

subsample (N = 2200 objects)• PSCz density field out to 6000 km/s• Low mass end of HIMF in high density

regions flatter and M* lower• Cannot be just morphology or HI

deficiency

Environment & the HIMFEnvironment & the HIMF

• Agreement between optically selected and HI blind HIMFs no worse than internal agreement among HI blind surveys

• Need larger sample to discriminate whether HIMF shape is dependent on morphology and environment separately (as done for 2dFGRS LF, e.g. Croton et al 2006)

Springob et al 2005 ApJ 621 215

HI and the “missing satellite” problemHI and the “missing satellite” problem HIPASS result: no cosmologically significant population of result: no cosmologically significant population of

HI-rich dark galaxies: HI-rich dark galaxies: ALFALFA agrees… butagrees… but HIPASS MMHIHI > > 101088 M M

ALFALFA is specifically designed (wide area, high velocity is specifically designed (wide area, high velocity resolution) to detect hundreds of objects with resolution) to detect hundreds of objects with MMHIHI < 10 < 107.57.5 M M

• Low HI mass Low HI mass • Narrow HI line width + exclude face-on objectsNarrow HI line width + exclude face-on objects• Will Will only be detected nearby => need to sample be detected nearby => need to sample

cosmologically significant volume

Future studies will focus on extending• Deeper in HI Mass than ALFALFA in Local Supercluster • Much larger volume than AGES

ALFALFA has already detected more objects with log MHI < 7.5 than all other previous blind HI surveys combined

Lowest HI mass objectsLowest HI mass objects

log MHI < 7.2

ALFALFA has already detected more objects with log MHI < 7.5 than all other previous blind HI surveys combined

The “Void Problem”The “Void Problem”Peebles (2000)

ApJ 597, 495

•Cosmic voids are filled with low mass dark matter haloesMare Nostrum simulation

vrot>55km/s

• ~1000 haloes with M < 109M and vrot< 20 km/s in a 20 h-1

Mpc void are predicted

Halo mass function in voids Gottlöber et al

(2003)

Luminosity function of void galaxiesLuminosity function of void galaxies

• Void LF has a faint M* but a similar faint-end slope, compared to the overall LF

• Void galaxies are blue, disk-like and have high H equivalent width => good HI targets

Void galaxies in the SDSS: Hoyle et al (2005)

1000 galaxies in lowest density cells of total 155,000 galaxies (SDSS 2005)

Clustering of HI galaxiesClustering of HI galaxies

ξ(r) for HIPASS•Meyer et al (2007):

•HI rich galaxies extremely weakly clustered

•Clustering scale depends on Vrot

•Basilakos et al (2007):•Massive HIPASS galaxies show same clustering as optically-selected sample

•Low mass systems (MHI < 109 M) show nearly uniform distribution

Inconsistency: Symptom of inadequate volume?

(Very) preliminary ALFALFA result(Very) preliminary ALFALFA result

•460 Mpc-3 in PPS foreground void at v~2200 km/s

•Simulations of Gottlöber et al. (2003) with dark:HI = 10:1 predict 38 HI sources

•ALFALFA finds no objectsThis makes Jim Peebles very excited…But only 2% of ALFALFA volume

STAY TUNED…..

Amélie Saintonge, Ph.D. thesis, Cornell U.

Saintonge et al. 2007, submitted

ALFALFA: ALFALFA: HI Cosmology at z=0HI Cosmology at z=0

• HI Mass function sampled over a fair volume• Low mass slope• Highest masses• Variation with environment• Halo occupation number• The “void problem”

• Correlation function over a fair volume• HI-HI• HI-optical/IR selected• Bias parameter (do galaxies trace mass)

• TF relation => peculiar velocities

•3rd generation surveys need to improve by ~10X in mass sensitivity on ALFALFA in local universe.

•3rd generation surveys must probe deeper in z and still cover cosmologically significant volume

How and when do galaxies acquire their How and when do galaxies acquire their gas?gas?

Kereš et al. (2005)

How and when do galaxies acquire their How and when do galaxies acquire their gas?gas?

Kereš et al. (2005)

AUDS: Arecibo Ultra-Deep Survey• Results (as reported by M.

Zwaan)• 53 hours during commissioning• 50 microJy rms• 14 HI detections + 9

candidates• 0.07 < z < 0.15

N=23

Highest mass objects: future SKAHighest mass objects: future SKA

• A prime science driver of the SKA is a HI “billion galaxy” survey .

(Abdalla & Rawlings 2004)

• Previous HI surveys detect very few objects with MHI > 1010 M; HIMF not well constrained at highest masses either.

• No theoretical expectation for massive galaxies with no stellar counterpart => targeted surveys

ALFALFA has already detected more than twice as many objects with log MHI > 10.4 than all other

previous blind HI surveys combined

Highest mass objects: future SKAHighest mass objects: future SKA

ALFALFA has already detected more than twice as many objects with log MHI > 10.4 than all other

previous blind HI surveys combined

SDSS color distributionSDSS color distribution

Ivan Baldry et al.

SDSS color distributionSDSS color distribution

Ivan Baldry et al.

Blue and red sequence LF’sBlue and red sequence LF’s

Baldry et al. (2004)

AGN: The Missing Link?AGN: The Missing Link?

Di Matteo, Springel & Hernquist 2005

• Tight observed relation between Mbulge and MBH

• Many of the transition objects in the so-called “green valley” with 3 < (NUV-r) < 5 appear to have AGN

• HI provides measure of cool gas, potential for star formation, unavailable otherwise => key clue to models of galaxy assembly

Highest HI mass ALFALFA detections show a range of morphologies/optical surface brightnesses• (Most) appear to be luminous disk systems

• Some have MHI/L > 2 => Mgas ~ M*

• Some have M* > 3 x 1010 M (“transition mass”)

• Fraction of AGN (TBD)

Direct measure of gas content in z ~ 0 “transition objects” • High mass objects • Many in “green valley” (NUV-r: (3-5)) have AGN• GASS (GALEX-Arecibo-SDSS survey: Schminovich et al.)

1000 galaxies, chosen by (NUV-r) colors, spectra 0.025 < z < 0.06 (matches ALFALFA range) Low gas mass fraction Mgas/M* ~ 0.01

Mass assembly: cool gas contentMass assembly: cool gas content

Probing HI galaxies at z ~ 0.2Probing HI galaxies at z ~ 0.2

There current are NO constraints on the

HIMF at cosmological z’s

SDSS-selected sample of spirals at z ~ 0.2+ Demonstrator of AO capability for HI studies over cosmic

timescales => evolution of Tully-Fisher relation, evolution of HI disks, constraints on evolution of HI Mass function

Future: development of focal/phased array for 900-1200 MHz

Redshifted HI emission

Barbara Catinella, NAIC (now MPA)

SDSS image

SKA precursor science with AOSKA precursor science with AO

Lister Staveley-Smith (Spineto, 2007)

Lister Staveley-Smith (Spineto, 2007)

Lister Staveley-Smith (Spineto, 2007)

But mapping may not be the right approach for Arecibo at intermediate z!

•Legacy applications of extragalactic HI line surveys pre-SKA•Deeper local blind HI surveys to explore faint end of HIMF

over adequate volume for cosmology => wide area•Deeper targeted HI surveys to explore the astrophysics of

mass assembly of galaxies=> push as far in z as practical•Absorption line studies (not discussed here)

•For redshifted HI line, need capability in 900-1200 MHz range•Phase I: 3-beam array, low noise (1 beam on target; 2 RFI)•Phase II: Phased array

•RFI issues must be tackled head-on!

Future HI line surveys with AOFuture HI line surveys with AO

•Need well defined survey science requirements•What surveys are needed to do what science?

• Design surveys to optimize science• Mechanisms to prioritize/coordinate

•Need to understand resource requirements of surveys •Survey results must be delivered on optimal schedule•Observing team support & software development•Automation + quality assurance•Data management, archiving, data products, access

Data volumes are huge!

•R&D for RFI mitigation/identification/excision

•Timely and convenient delivery of data products to public archive

•Permanent curation/delivery•Public access tools; visualization tools

•How to engage and energize the community at large?!

Practical challenges to “US” Practical challenges to “US”