1 of 38

secrets of star formation histories with laduma and SIMBA

Leyya Stockenstroom (phd, uct)

Supervised by dr. jacinta delhaize (UCT)

Co-supervised by DR. ros skelton (uct, saao)

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

2 of 38

galaxy evolution

2.

figure source: Walter et al., 2020a

PHISCC 2025, Cagliari

3 of 38

overview

Master’s work

Investigate the relationship between the HI and star formation history properties of low-redshift LADUMA galaxies

3.

HI properties

SFH properties

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

4 of 38

key properties	Value
Central Pointing (J2000)	03h32m30s – 28d07m57s
Frequency range	1304-1420 MHz
Redshift range	0-0.088
Synthesised beam
Position angle of beam	34°
Per channel flux sensitivity
Channel width	104.52 kHz

The laduma survey

4.

PHISCC 2025, Cagliari

5 of 38

hi sample

5.

Ran SoFiA-2 [Westmeier et al. 2021] on the low redshift LADUMA data cube

Obtained the HI properties of 193 galaxies

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

6 of 38

hi sample

5.

Ran SoFiA-2 [Westmeier et al. 2021] on the low redshift LADUMA data cube

Obtained the HI properties of 193 galaxies

PHISCC 2025, Cagliari

7 of 38

hi sample

5.

Ran SoFiA-2 [Westmeier et al. 2021] on the low redshift LADUMA data cube

Obtained the HI properties of 193 galaxies

Cross-matched with the DEVILS photometric catalogue for multiband photometry

Leaving a sample of 135 galaxies

	Facility	Survey	Filters
UV	GALEX	Multiple	FUV, NUV
Optical	VST	VOICE	u, g, r, i
Nir	VISTA	VIDEO	Y, J, H, Ks
MIR	Spitzer	SERVES, SWIRE	S36, S45, S58, S80, MIPS24, MIPS0
Fir	Herschel	HerMES	P100, P160, S250, S350, S500

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

8 of 38

SED fitting

6.

HI redshift

+

photometry

SED fitting

SFH

Performed SED fitting for all 135 HI galaxies

SED fitting code ProSpect [Robotham et al., 2020]
Incorporates an evolving metallicity
Utilises a parametric SFH model

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

9 of 38

star formation history

7.

Lookback time

Star formation rate

The star formation history (SFH)

Reflects the timeline of a galaxy’s star formation activity
SFR vs lookback time
i.e. traces the SFR of a galaxy through its history

Modelled by a truncated Gaussian in ProSpect
Assumes one defined peak SFR in a galaxy’s history

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

10 of 38

star formation history

8.

VS

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

11 of 38

star formation history

9.

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

12 of 38

star formation history

10.

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

13 of 38

star formation history

11.

VS

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

14 of 38

star formation history

12.

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

15 of 38

star formation history

13.

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

16 of 38

The Gaps

14.

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

17 of 38

The Gaps

14.

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

18 of 38

The Gaps

14.

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

19 of 38

overview

Phd work

Use SIMBA simulation to confirm the SFR peak vs HI gas fraction relation is not a result of selection bias or SED modelling assumptions

15.

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

20 of 38

SIMBA simulations

16.

GIZMO: Gravity + meshless finite mass hydrodynamics

H₂-based star formation

Grackle cooling
On-the-fly self-shielding for HI fraction

Star formation feedback: Kinetic two-phase decoupled winds (FIRE scalings)

Two-mode BH accretion, separated at T~10⁵K:

Torque-limited (cool): Limited by angular mom loss to acc disk (~HERGs).
Bondi (hot): Limited by gravitational capture from hot medium (~LERGS).

Kinetic BH feedback, 3 modes: Radiative (bipolar v~10³, high-f_Edd), Jet (bipolar v~10⁴, low-f_Edd), X-ray (v~10², low-f_Edd+low-f_gas, spherical).

Dust production & destruction on the fly.

Dave et al., 2019

21 of 38

Simba simulations

17.

	Facility	Filters
UV	GALEX	FUV, NUV
Optical	Suprimecam	g, r, i, z
Nir	VISTA	Y, J, H, Ks
IR	WISE	W1,W2

With the flagship SIMBA run [Dave et al., 2019]:

HI masses
Stellar masses
SFR
FUV-IR colours

Meaning we can run ProSpect on SIMBA galaxies

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

22 of 38

Simba simulations

17.

With the flagship SIMBA run [Dave et al., 2019]:

HI masses
Stellar masses
SFR
FUV-IR colours

Meaning we can run ProSpect on SIMBA galaxies

We can also use z=0-20 snapshots to extract the SFH directly from the simulation

Avoiding restrictive models

Allowing us to compare properties from Prospect to true properties from SIMBA

PHISCC 2025, Cagliari

23 of 38

Stellar mass & SFR

18.

0.4 dex offset in SFR

Offset roughly agrees with Lower et al., (2020)

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

24 of 38

SFH prospect vs simba

19.

The properties of the SFHs from ProSpect are not very reliable

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

25 of 38

SFH relations from SIMBA

20.

PHISCC 2025, Cagliari

Email: STCLEY001@myuct.ac.za

26 of 38

Summary and future work

20.

PHISCC 2025, Cagliari

Thank You ☺

Email: STCLEY001@myuct.ac.za

27 of 38

Thank You ☺

Leyya Stockenstroom

Email: STCLEY001@myuct.ac.za

28 of 38

Additional slides

29 of 38

sofia parameter name	set value
the flagging parameters
flag.region	3533, 3578, 974, 1016, 0, 1102
the S+C parameters
scfind.kernelsXY	0, 3, 5
scfind.kernelsZ	0, 3, 5, 7
scfind.threshold	4.0
the Linker parameters
linker.radiusXY	3
linker.radiusZ	2
the reliability parameters
reliability.threshold	0.5
reliability.scaleKernel	0.25
reliability.minSNR	12.0

Hi source finding

SoFiA-2 was used to perform blind source finding

Ran iteratively to optimise parameters
11 parameters adjusted
The remaining parameters were set to default

19.

30 of 38

Hi source finding

Source finding done with SoFiA-2

232 candidate sources found

193 are confirmed HI detections (found by SoFiA-2 only)
39 detections deemed as false

19.

“False”

detection

Confirmed

detection

31 of 38

Hi source finding

Source finding done with SoFiA-2

This catalogue was then compared to other methods of source finding (manual and semi-automated)

232 candidate sources found

193 are confirmed HI detections (found by SoFiA-2 only)
13 were detected by SoFiA-2 only
12 sources found only by manual source finding
17 sources found only by matched filtering (semi-automated)

SoFiA-2 found 82% of the sources found by other methods

19.

32 of 38

SED fitting

12.

33 of 38

Priors catagory	priors	Description	units	scale	Fitting range
SFH	mSFR	Peak SFR	Myr	log	[-3, 4]
	mpeak	Lookback time of peak SFR	Gyr	linear	[-2, 13.38]
	mperiod	Width of normal distribution	Gyr	log	[log(0.3), 2]
	mskew	Skewness of normal distribution	Gyr	linear	[-0.5, 1]
Metalitcty	Zfinal	Final matilicity	-	log	[-4, 1.3]
Dust		Column density of dust in birth clouds	-	log	[-2.5, 1.5]
		Column density of dust in ISM	-	log	[-5, 1]
		Tempreture of dust in birth clouds	-	linear	[0, 4]
		Temperature of dust in ISM	-	linear	[0, 4]

SED fitting

9.

SED fitting done with ProSpect [Robotham et al., 2020]

Incorporates an evolving metallicity
Multiple algorithms for processing SFHs
Utilises an MCMC algorithm to fit 9 free priors
Performed SED fitting to our HI sample of 135 galaxies

34 of 38

Photometric catalogue

7.

The Deep Extragalactic Visible Legacy Survey (DEVILS)

An ongoing spectroscopic survey at z<1
Compiled a photometric catalogue as part of their survey
Not the entirity of the LADUMA field is observed by DEVILS
135 galaxies with multiwavelength coverage

35 of 38

scaling relations

13.

36 of 38

scaling relations

21.

37 of 38

SFH relations from Prospect

25.

38 of 38

Robustness of SFH

22.

A similar test has been done by Lower et al., 2020
Who looked at how parametric models of SFH compare to non-parametric models

0.4 dex offset with parametic models
0.1 dex offset with non-parametric models

However, this was tested with PROSPECTOR
How does ProSpect compare?