THE LAST SURVEY OF THE ‘OLD’ WSRT: TOOLS AND RESULTS FOR THE FUTURE HI ABSORPTION SURVEYS

F. Maccagni; R. Morganti; T. Oosterloo; K. Geréb; N. Maddox, J. Allison

A SURVEY BEFORE THE ‘BLIND’ SURVEYS

‣ During my PhD, WSRT was upgraded to Apertif: over time antennae went offline

‣ To detect HI absorption we don’t need complete uv-coverage

‣ Great opportunity for HI absorption studies

‣ Observe as many sources as possible before WSRT observations stop

‣ 4/6 hrs observation with variable number of antennae

‣ average noise in the spectra ~ 1 mJy

‣ Set strategy and tools for the ‘blind’ surveys: SHARP, MALS, FLASH

‣ What is the detection rate of HI in the local Universe?

‣ What features of the HI lines relate to the properties of the radio sources?

THE LAST SURVEY OF THE OLD WESTERBORK ‣ 248 sources

‣ 0.02 < z < 0.25

‣ SDSS spectroscopy

‣ SCont ≥ 30 mJy

‣ mostly AGN in ETG

1. 101 sources; SCont ≥ 50 mJy

‣ Stacking experiment [Geréb et al., 2014]

‣ Analysis of the detections [Geréb , Maccagni, et al., 2015]

2. All 248 sources

‣ This Talk [Maccagni et al., 2017]

39 40 41 42 43 44 45 46log10 L22µm [erg s�1]

18

19

20

21

22

23

24

25

26

log 1

0P

1.4G

Hz

[WH

z�1 ]

Jarrett et al. 2013

Dust-poor source

12µm bright source

4.6µm bright source

Interacting source

A3D Dust-poor source

A3D 12µm source

A3D 4.6µm source

ATLAS3D

This SAMPLE

Mid-InfraRed [22 µm] - Radio Power Relation

�10

�5

0

Flu

x[m

Jy]

1.326 1.328 1.330 1.332 1.334 1.336 1.338 1.340Frequency [⇥109 Hz]

0.81.01.21.41.6

Noi

se[m

Jy]

Detection

STRATEGY OF THE SURVEY

Continuum source finder: location continuum sources (down to 10mJy/

5mJy or even lower)

Similar to APERTIF observations Flagging and automatic calibration

Data Cube + Continuum Image

Identify HI absorption detections: Bayesian line-finder [Allison et al 2012]

Extract the spectra @ location continuum sources

beam

�4

�2

0

2

4

Flu

x[m

Jy]

1.306 1.308 1.310 1.312 1.314 1.316 1.318 1.320Frequency [⇥109 Hz]

0.81.01.21.41.61.8

Noi

se[m

Jy]

Non Detection

�10

�5

0

Flu

x[m

Jy]

1.326 1.328 1.330 1.332 1.334 1.336 1.338 1.340Frequency [⇥109 Hz]

0.81.01.21.41.6

Noi

se[m

Jy]

STRATEGY FOR HI ABSORPTION SURVEYS

Characterisation absorption (width, centre, asymmetry etc.

using e.g. busy function)

DATABASE

cross-correlation with LOFAR fields & other archives

(SDSS, WISE)

Intervening or associated?cross-correlation with spectral surveys

Detection �4

�2

0

2

4

Flu

x[m

Jy]

1.306 1.308 1.310 1.312 1.314 1.316 1.318 1.320Frequency [⇥109 Hz]

0.81.01.21.41.61.8

Noi

se[m

Jy]

Non Detection

Properties of radio continuum (extent, spectral

index, …)

Properties of the HI lines

HI ABSORPTION LINES‣ 66 detections

‣ line features measured with the BusyFunction [Westmeier, et al. 2014]

‣ 30 < FWHM < 570 km/s

‣ 70 < FW20 < 640 km/s

‣ 3 main groups:

‣ Narrow lines:

FWHM < 100 km/s

‣ Medium width lines:

100 km/s <FWHM < 200 < km/s

‣ Broad lines:

FWHM > 200 km/s

CHARACTERISATION OF THE SAMPLE

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0FIRST peak / integrated flux

1.0

1.2

1.4

1.6

1.8

2.0

NV

SSm

ajor

/m

inor

axis

Extended non-detection

Compact non-detection

Extended detection

Compact detection

Interacting detection

Interacting non-detection

‣ Compact sources (red):

‣ unresolved by FIRST.

‣ often radio-jets on sub-galactic scales.

‣ many compact sources are young AGN.

‣ Extended sources (blue):

‣ resolved by FIRST.

‣ radio-jets on super-galactic scales.

‣ usually older AGN than compact sources.

CHARACTERISATION OF THE SAMPLE‣ WISE MIR colours

‣ dust in the host galaxy

‣ Dust-poor sources (green)

‣ 12 µm-bright sources (orange)

‣ Emission from PAHs and heated dust.

‣ 4.6 µm-bright sources (black)

‣ The central AGN heats the surrounding circumnuclear dust.

0.0 1.0 2.0 3.0 4.0 5.0W2 – W3 [4.6 - 12 µm]

0.0

0.5

1.0

1.5

W1

–W

2[3

.4-

4.6

µm

]

Dust-poor non-detection

12µm bright non-detection

4.6µm bright non-detection

Dust-poor detection

12µm bright detection

4.6µm bright detection

Interacting non-detection

Interacting detection

DETECTING HI ABSORPTION

‣ 248 sources / 66 Detections

‣ 27 % ± 5.5 % detection rate

‣ Constant in redshift and radio power

‣ Compact sources and MIR bright

‣ HI often detected (~40%).

‣ Extended sources & dust-poor sources

‣ HI is rarely detected (~13%).

0

5

10

15

20

25

Cou

nt

All sources

Detections

0.02 0.05 0.08 0.11 0.14 0.17 0.20 0.23z

0

20

40

60

80

Det

.R

ate

[%]

0

5

10

15

20

25

30

35

Cou

nt

All sources

Detections

23 24 25 26log10 P1.4GHz [W Hz�1]

0

20

40

60

80

Det

.R

ate

[%]

KINEMATICS OF THE HI

22 23 24 25 26log10 P1.4GHz [W Hz�1]

0

200

400

600

800

FW

20[k

ms�

1 ]

Interacting sources

Dust-poor sources

12µm bright sources

4.6µm bright sources

22 23 24 25 26log10 P1.4GHz [W Hz�1]

0

200

400

600

800

FW

20[k

ms�

1 ]

Interacting sources

Extended sources

Compact sources

‣ P1.4GHz< 1024 W Hz-1 ‣ widths ≤ rotational velocity ‣ HI likely in a rotating disk.

‣ P1.4GHz> 1024 W Hz-1

‣ broad asymmetric lines.

‣ Sources with broad lines are: ‣ Compact, i.e. jets within the galaxy. ‣ MIR bright, i.e. rich in heated dust.

22 23 24 25 26log10 P1.4GHz [W Hz�1]

�600

�400

�200

0

200

400

600

v cen

troi

d-

v sys

tem

ic[k

ms�

1 ]

Interacting sources

Extended sources

Compact sources

22 23 24 25 26log10 P1.4GHz [W Hz�1]

�600

�400

�200

0

200

400

600

v cen

troi

d-

v sys

tem

ic[k

ms�

1 ]

Interacting sources

Dust-poor sources

12µm bright sources

4.6µm bright sources

‣ P1.4GHz< 1024 W Hz-1

‣ lines centred at systemic velocity

‣ P1.4GHz> 1024 W Hz-1

‣ lines offset w.r.t. systemic velocity

‣ offset is blue-shifted.

‣ Broad, asymmetric, shifted absorption line

‣ Unsettled kinematics

‣ Powerful radio sources

‣ Compact, i.e. jets within the galaxy.

‣ MIR bright, i.e. rich in heated dust.

KINEMATICS OF THE HI

STACKING EXPERIMENT

�1500 �1000 �500 0 500 1000 1500Velocity [km s�1]

�0.002

0.000

0.002

0.004

0.006

⌧

Compact non-detections

Extended non-detections

�1500 �1000 �500 0 500 1000 1500Velocity [km s�1]

�0.002

0.000

0.002

0.004

0.006

⌧

12µm & 4.6µm-bright non-detections

Dust-poor non-detections

�1500 �1000 �500 0 500 1000 1500Velocity [km s�1]

�0.002

�0.001

0.000

0.001

0.002

0.003

⌧

Non Detections

‣ Non-detections are important!!!!

‣ Stacking of 170 non-detections

‣ NO LINE is detected at ~ 0.0015 (3𝛔)

‣ Stacking of sub-groups of sources

‣ NO LINE is detected at ~ 0.003 (3𝛔)

‣ Not even in compact sources or MIR bright sources.

STACKING THE ATLAS3D NON-DETECTIONS

�1000 �500 0 500 1000Velocity [km s�1]

�6

�4

�2

0

2

4

6

N(H

I)[⇥

1017

cm�

2bea

m�

1 ]

‣ 81 ATLAS3D sources HI is not detected in the centre.

‣ STACKING: 3𝛔 detection of HI emission

‣ N(HI) ~ 3.5 x 1017 (Tspin/cf) cm-2

‣ N(HI) converted in optical depth (Tspin~100 K, cf =1)

‣ 𝝉~ 0.0006 << 0.0015

‣ we need to stack more to detect this gas in absorption.

‣ The HI stacking ATLAS3D is warm?

‣ Tspin ↑ ⇒ 𝝉 ↓ ; Tspin ↓ ⇒ 𝝉 ↑

‣ Stacking in absorption even more difficult

INTERPRETING HI ABSORPTION

‣ Understand the overall distribution of the HI traced by the absorption line ‣ What to can we infer from only the integrated line and the continuum image? ‣ Model the rotating HI disk in front of the radio continuum:

3C 305

Plane of the sky : x,y Side view: z,y From above: x,z

Observation Model

INTERPRETING HI ABSORPTION ‣ 3C 305 ‣ Optical Image (SDSS or other): i, PA of the stellar body ‣ Continuum image: against which radio lobe there is absorption?

‣ i ∊ [0°,180°] ‣ PA ∊ [180°, 360°]

INTERPRETING HI ABSORPTION

247

495

742

989

1237

1484

Cou

nt

0 15 30 45 60 75I [�]

180

210

240

270

300

330

360

PA

[�]

180 210 240 270 300 330 360PA [�]

280

560

840

1119

1399

1679

1959

Cou

nt

‣ MCMC algorithm

‣ find combination of parameters that best fits the observed line the line

‣ i = 45°; PA = 270 °

‣ 3C 305 ‣ Optical Image (SDSS or other): i, PA of the stellar body ‣ Continuum image: against which radio lobe there is absorption?

‣ i ∊ [0°,180°] ‣ PA ∊ [180°, 360°]

INTERPRETING HI ABSORPTION ‣ The bulk of the absorption generated by a rotating disk: i = 45°; PA = 270 ° ‣ Blue-shifted wing not reproduced by the model

�1000 �500 0 500 1000Velocity [km s�1]

�0.010

�0.005

0.000

0.005

Flu

x[m

Jy]

Spectrum

observation

model

-1250 -1000 -750 -500 -250 0 250 500 750 1000 1250�0.004

�0.002

0.000

0.002

0.004

Flu

x[m

Jy]

resitudals

-4.0 -2.0 0.0 2.0 4.0x [kpc]

-4.0

-2.0

0.0

2.0

4.0

y[k

pc]

Plane of the sky

-4.0 -2.0 0.0 2.0 4.0x [kpc]

-4.0

-2.0

0.0

2.0

4.0

z[k

pc]

View from ‘above’

-4.0 -2.0 0.0 2.0 4.0z [kpc]

-4.0

-2.0

0.0

2.0

4.0

y[k

pc]

View from the ‘side’

INTERPRETING HI ABSORPTION

�1000 �500 0 500 1000Velocity [km s�1]

�0.0020

�0.0015

�0.0010

�0.0005

0.0000

0.0005

0.0010

Flu

x[m

Jy]

Spectrum

observation

model

-1250 -1000 -750 -500 -250 0 250 500 750 1000 1250�0.0010

�0.0005

0.0000

0.0005

Flu

x[m

Jy]

resitudals

-800.0 -400.0 0.0 400.0 800x [pc]

-800.0

-400.0

0.0

400.0

800

y[p

c]

Plane of the sky

-800.0 -400.0 0.0 400.0 800x [pc]

-800.0

-400.0

0.0

400.0

800

z[p

c]

View from ‘above’

-800.0 -400.0 0.0 400.0 800z [pc]

-800.0

-400.0

0.0

400.0

800

y[p

c]

View from the ‘side’

320

641

962

1282

1602

1923

Cou

nt

0 20 40 60 80 100 120 140 160 180I [�]

0

60

120

180

240

300

360

PA

[�]

0 60 120 180 240 300 360PA [�]

106.6

213.1

319.7

426.3

532.9

639.4

746.0

Cou

nt

‣ Less information on the source?

‣ i ∊ [0°,180°] ‣ PA ∊ [0°, 360°]

‣ Best-fit solution: i = 55°; PA = 240 °

‣ VLBI high resolution continuum ‣ Likely we can improve the fit

CONCLUSIONS

‣ 27%±5.5% detection rate of HI in absorption

‣ HI detected at all redshifts (0.02 < z < 0.23) and radio powers:

‣ promising for SHARP, MALS, FLASH

‣ Narrow lines

‣ HI mainly in a rotating disk

‣ P1.4GHz< 1024 W Hz-1, Extended sources, dust poor sources

‣ Broad asymmetric shifted lines:

‣ HI has unsettled kinematics

‣ P1.4GHz> 1024 W Hz-1, Compact sources (i.e. often young AGN), MIR bright sources

‣ Stacking experiments:

‣ low optical depth HI is present in the centre of ETGs, warmer HI (Tspin >100 K)?

�10

�5

0

Flu

x[m

Jy]

1.326 1.328 1.330 1.332 1.334 1.336 1.338 1.340Frequency [⇥109 Hz]

0.81.01.21.41.6

Noi

se[m

Jy]

BLIND SURVEYS: AUTOMATIC SEARCH FOR HI ABSORPTION

Characterisation absorption (width, centre, asymmetry etc.

e.g. busy function)

DATABASE

Stackingcross-correlation with LOFAR fields & other archives (SDSS, WISE)

other ideas...Comparison with models

Intervening or associated? cross-correlation with spectral surveys

Detection�4

�2

0

2

4

Flu

x[m

Jy]

1.306 1.308 1.310 1.312 1.314 1.316 1.318 1.320Frequency [⇥109 Hz]

0.81.01.21.41.61.8

Noi

se[m

Jy]

Non Detection

Line-finder: which one? Extract the spectra @ location of every

continuum source in the FOV

Properties of the HI lines

Properties of radio continuum

(size, spectral index, …)