Tip of the Red Giant Branch Distances to NGC 1316, NGC 1380, NGC 1404, & NGC 4457: A Pilot Study of a Parallel Distance Ladder Using Type Ia Supernovae in Early-Type Host Galaxies

TL;DR

This study explores a new method for measuring cosmic distances using Type Ia supernovae in quiescent galaxies, calibrated with TRGB distances, aiming to improve the accuracy of the Hubble constant measurement.

Contribution

It demonstrates the feasibility of a parallel distance ladder using TRGB calibration in early-type galaxies hosting SNe Ia, providing a potentially more consistent approach for cosmological measurements.

Findings

Achieved a Hubble constant of 75.3 ± 2.9 km/s/Mpc.

Established a TRGB calibrator sample of five high-mass, early-type galaxies.

Reduced Hubble residual scatter to under 0.11 mag.

Abstract

Though type-Ia supernovae (SNe Ia) are found in all types of galaxies, recent local Hubble constant measurements have disfavored using SNe Ia in early-type or quiescent galaxies, aiming instead for better consistency with SNe Ia in star-forming, late-type host galaxies calibrated by Cepheid distances. Here we investigate the feasibility of a parallel distance ladder using SNe Ia exclusively in quiescent, massive ($\log M_*/M_{\odot} \geq 10$) host galaxies, calibrated by tip of the red giant branch (TRGB) distances. We present TRGB measurements to four galaxies: three measured from the Hubble Space Telescope with the ACS F814W filter, and one measured from the JWST NIRCam F090W filter. Combined with literature measurements, we define a TRGB calibrator sample of five high-mass, early-type galaxies that hosted well-measured SNe Ia: NGC 1316 (SN 2006dd), NGC 1380 (SN 1992A), NGC 1404 (SN 2007on, SN 2011iv), NGC 4457 (SN 2020nvb), and NGC 4636 (SN 2020ue). We jointly standardize these calibrators with a fiducial sample of 124 Hubble-flow SNe Ia from the Zwicky Transient Facility that are matched in host-galaxy and light-curve properties. Our results with this homogenized subsample show a Hubble residual scatter of under 0.11 mag, lower than usually observed in cosmological samples of the full SN~Ia distribution. We obtain a measurement of the Hubble constant, $H_0 = 75.3 \pm 2.9$ km s$^{-1}$ Mpc$^{-1}$, including statistical and estimated systematic uncertainties, and discuss the potential to further improve the precision of this approach. As calibrator and supernova samples grow, we advocate that future cosmological applications of SNe Ia use subsamples matched in host-galaxy and supernova properties across redshift.

Tables

Table 1. Table 1: Summary of the HST  and JWST  Galaxy Sample Observations

Galaxy	R.A. (J2000)	Decl. (J2000)	Exposure Time (s)		PID	Milky Way $E\left(B-V\right)$   aaReported $E\left(B-V\right)$ values are sampled from the E. F. Schlafly & D. P. Finkbeiner (2011) recalibration of the D. J. Schlegel et al. (1998) dust map at R.A. and Decl. coordinates listed in columns 1 and 2.
	(h:m:s)	(d:m:s)	Filters			(mag)
			F606W	F814W
NGC 1316	03:23:12.69	$-$37:19:19.33	14400	24000	HST-GO-13691	0.021
NGC 1404	03:38:46.68	$-$35:34:04.08	40800	40800	HST-GO-15642	0.012
NGC 4457	12:28:57.59	$+$03:36:56.24	8200	8200	HST-GO-16438	0.022
			F090W	F150W
NGC 1380	03:36:27.59	$-$34:58:34.68	7730	1933	JWST-GO-3055	0.017
NGC 4636	12:42:49.833	$+$02:41:15.95	5154	1074	JWST-GO-3055	0.029

Table 2. Table 2: Summary of Galaxy Structural Parameters

Galaxy	R.A. (J2000)	Decl. (J2000)	P.A.	$e$
	(deg)	(deg)	(deg)
NGC 1316	50.67	$-$37.08	$+$55.4	0\lx@alignment@align.3
NGC 1380	54.11	$-$34.98	$-$84.1	0\lx@alignment@align.6
NGC 1404	54.72	$+$35.59	$-$10.3	0\lx@alignment@align.2
NGC 4457	187.25	$+$03.57	$+$70.5	0\lx@alignment@align.8

Table 3. Table 3: Summary of High-fidelity Photometric Source Catalog Properties

Galaxy	Crowding (mag)	Sharpness2	SNR (Blue, Red)	$N_{\ast }$	SMA${}_{\text{inner}}$	$N_{\ast ,\text{SMA}\ge {\text{SMA}}_{\text{inner}}}$
NGC 1316	$0.15$	$0.03$	$2.0$, $4.0$	12,035	$\cdots$	$\cdots$
NGC 1380	$0.15$	$0.02$	$4.0,4.0$	110,556	${201.81}^{\prime \prime }$	47,743
NGC 1404	$0.43$	$0.03$	$2.0$, $4.0$	52,091	${138.69}^{\prime \prime }$	39,486
NGC 4457	$0.19$	$0.02$	$2.0$, $4.0$	11,445	${156.51}^{\prime \prime }$	6,661

Table 4. Table 4: Summary of the TRGB Distance Measurements

Galaxy	$m_{\text{F814W}}^{\text{TRGB}}$ (mag)	${\mu }_{\text{F814W}}^{\text{TRGB}}$ (mag)	D${}_{\text{F814W}}{}^{}{}_{}{}^{\text{TRGB}}$ (Mpc)
NGC 1316 (SN 2006dd)	${27.45}_{-0.01}^{+0.06}\text{ (stat)}$	${31.50}_{-0.03}^{+0.06}\text{ (stat) }\pm 0.04\text{ (sys)}$	${19.95}_{-0.29}^{+0.58}\text{ (stat) }\pm 0.35\text{ (sys)}$
NGC 1404 (SN 2007on, SN 2011iv)	${27.18}_{-0.07}^{+0.09}\text{ (stat)}$	${31.23}_{-0.07}^{+0.09}\text{ (stat) }\pm 0.04\text{ (sys)}$	${17.62}_{-0.59}^{+0.74}\text{ (stat) }\pm 0.31\text{ (sys)}$
NGC 4457 (SN 2020nvb)	${27.07}_{-0.03}^{+0.01}\text{ (stat)}$	${31.12}_{-0.03}^{+0.01}\text{ (stat) }\pm 0.04\text{ (sys)}$	${16.75}_{-0.23}^{+0.08}\text{ (stat) }\pm 0.29\text{ (sys)}$
	$m_{\text{F090W}}^{\text{TRGB}}$ (mag)	${\mu }_{\text{F090W}}^{\text{TRGB}}$ (mag)	D ${}_{\text{F090W}}{}^{}{}_{}{}^{\text{TRGB}}$ (Mpc)
NGC 1380 (SN 1992A)	$27.00\pm 0.01\text{ (stat)}$	$31.32\pm 0.03\text{ (stat) }\pm 0.04\text{ (sys)}$	$18.37\pm 0.27\text{ (stat) }\pm 0.34\text{ (sys)}$
NGC 4636aaThe values reported here for NGC 4636 are not measured in this study and are adopted from G. S. Anand et al. (2025). We reproduce their results which are provided separately for NIRCam detectors A1 and A2 (NRCA1 and NRCA2, respectively). We note that G. S. Anand et al. (2025) use a different flux calibration for their JWST  photometry than was used in this study. The flux calibration results in a known offset in the recovered magnitudes of stars (G. S. Anand et al., 2024a, b). The flux calibration is accounted for in the distance modulus with the proper choice of TRGB zero-point. (SN 2020ue)	$26.813\pm 0.035\text{ (stat) }$ (NRCA1)	$31.12\pm 0.07$	$16.8\pm 0.6$
$\dots$	$26.608\pm 0.035\text{ (stat) }$ (NRCA2)	$\dots$	$\dots$

Table 5. Table 5: SALT2 and BayeSN Supernova Light Curve Fit Parameters and Host-galaxy Stellar Masses.

	SALT2				BayeSN		Host Stellar Mass
SN Ia	$m_B$ (mag)	$M_B^i$ (mag)	$x_1$	$c$	$A_V$ (mag)	$\theta$	$\log M_{\star }/M_{\odot }$
SN 1992A	$12.436\pm 0.029$	$-18.884\pm 0.049$	$-1.600\pm 0.040$	$-0.028\pm 0.025$	$0.06\pm 0.04$	$0.95\pm 0.14$	$10.88\pm 0.15$
SN 2006dd	$12.183\pm 0.030$	$-19.317\pm 0.067$	$-0.475\pm 0.053$	$-0.014\pm 0.027$	$0.15\pm 0.06$	$0.08\pm 0.17$	$11.56\pm 0.15$
SN 2007on	$12.922\pm 0.023$	$-18.308\pm 0.093$	$-2.061\pm 0.043$	$+0.022\pm 0.022$	$0.30\pm 0.05$	$1.65\pm 0.07$	$10.96\pm 0.15$
SN 2011iv	$12.348\pm 0.023$	$-18.882\pm 0.093$	$-1.745\pm 0.033$	$-0.024\pm 0.021$	$0.36\pm 0.05$	$1.13\pm 0.13$	$10.96\pm 0.15$
SN 2020ue	$12.024\pm 0.022$	$-19.096\pm 0.073$	$-1.839\pm 0.031$	$-0.025\pm 0.019$	$0.34\pm 0.05$	$1.68\pm 0.04$	$10.98\pm 0.15$
SN 2020nvb	$12.140\pm 0.026$	$-18.980\pm 0.048$	$-1.419\pm 0.065$	$-0.055\pm 0.024$	$0.23\pm 0.06$	$1.67\pm 0.12$	$10.49\pm 0.15$

Table 6. Table 6: Model Fit Results for the Fiducial Sample and Analysis Variants.

	Calibrators		Hubble flow
Sample	$N$	$\sigma$	$N$	$\sigma$	$H_0$	$M_B$	$-5a_B$	${\sigma }_{\mathrm{int}}$	$\alpha$	$\beta$	$\gamma$
		(mag)		(mag)	(km s-1 Mpc-1)	(mag)	(mag)	(mag)			(mag dex-1)
Fiducial: $z>0.023$, $x_1$:[$-$2,0], $c$:[$-$0.2,+0.1]	5	0.096	124	0.106	75.3 $\pm$ 1.7 $\pm$ 2.4	$-19.152\pm 0.059$	$-3.536\pm 0.027$	$0.079\pm 0.009$	$0.209\pm 0.016$	$2.11\pm 0.17$	$0.111\pm 0.029$
Include SN 2007on, $x_1$:[$-$2.2,0]	6	0.192	134	0.108	77.0 $\pm$ 1.7 $\pm$ 2.5	$-19.119\pm 0.058$	$-3.551\pm 0.026$	$0.083\pm 0.010$	$0.229\pm 0.016$	$2.25\pm 0.17$	$0.111\pm 0.029$
Strict, fast-decliners: $x_1$:[$-$2,$-$1], $c$:[$-$0.1,0.0]	4	0.106	38	0.080	76.1 $\pm$ 1.6 $\pm$ 2.4	$-19.253\pm 0.088$	$-3.659\pm 0.078$	$0.058\pm 0.015$	$0.228\pm 0.045$	$1.22\pm 0.54$	$0.040\pm 0.045$
Loose: $z>0.015$, $x_1$:[$-$3,+1], $c$:[$-$0.3,+0.5]	6	0.190	413	0.170	77.5 $\pm$ 2.4 $\pm$ 2.5	$-19.001\pm 0.071$	$-3.448\pm 0.016$	$0.136\pm 0.007$	$0.189\pm 0.009$	$2.87\pm 0.06$	$0.150\pm 0.026$
All Hubble-flow SNe Ia and hosts:	6	0.200	873	0.211	78.0 $\pm$ 2.7 $\pm$ 2.5	$-19.001\pm 0.076$	$-3.462\pm 0.008$	$0.162\pm 0.006$	$0.161\pm 0.007$	$3.05\pm 0.04$	$0.095\pm 0.010$
No ZTF photometry corrections:	5	0.097	125	0.104	76.1 $\pm$ 1.8 $\pm$ 2.4	$-19.151\pm 0.059$	$-3.558\pm 0.024$	$0.084\pm 0.009$	$0.204\pm 0.016$	$2.08\pm 0.16$	$0.106\pm 0.030$
No host stellar mass correction:	5	0.117	124	0.111	74.0 $\pm$ 1.7 $\pm$ 2.4	$-19.256\pm 0.055$	$-3.602\pm 0.021$	$0.087\pm 0.009$	$0.208\pm 0.017$	$2.11\pm 0.17$	0 (fixed)
Add SN 1994D in NGC 4526:	6	0.105	124	0.106	74.4 $\pm$ 1.6 $\pm$ 2.4	$-19.179\pm 0.056$	$-3.537\pm 0.026$	$0.079\pm 0.010$	$0.209\pm 0.016$	$2.14\pm 0.17$

Table 7. Table 7: LCO observations of SN 2020ue. All photometry is from template-subtracted images.

Telescope	MJD	Filter	Mag	Mag Err
1m0-04	58861.30	B	14.779	0.022
1m0-04	58861.30	B	14.789	0.021
1m0-04	58861.30	V	14.660	0.024
1m0-04	58861.30	V	14.648	0.024
1m0-04	58861.31	g	14.585	0.008
1m0-04	58861.31	g	14.571	0.007
1m0-04	58861.31	r	14.592	0.008
1m0-04	58861.31	r	14.593	0.007
1m0-04	58861.31	i	14.820	0.007
1m0-04	58861.32	i	14.839	0.008
1m0-10	58862.02	B	14.252	0.022
1m0-10	58862.02	B	14.239	0.022
1m0-10	58862.02	V	14.144	0.024
1m0-10	58862.02	V	14.141	0.024
1m0-10	58862.03	g	14.051	0.008
1m0-10	58862.03	g	14.035	0.008
1m0-10	58862.03	r	14.105	0.009
1m0-10	58862.03	r	14.103	0.010
1m0-10	58862.03	i	14.325	0.008
1m0-10	58862.04	i	14.328	0.008
1m0-11	58862.66	B	13.954	0.024
1m0-11	58862.66	V	13.948	0.030
1m0-11	58862.66	V	18.340	0.155
1m0-11	58862.66	g	13.654	0.010
1m0-11	58862.67	g	13.671	0.009
1m0-11	58862.67	r	13.816	0.007
1m0-11	58862.67	r	13.914	0.008
1m0-11	58862.67	i	14.013	0.008
1m0-11	58862.68	i	14.136	0.010
1m0-08	58866.38	B	12.890	0.021
1m0-08	58866.38	B	12.912	0.021
1m0-08	58866.39	V	12.795	0.024
1m0-08	58866.39	V	12.817	0.024
1m0-08	58866.39	r	12.715	0.006
1m0-08	58866.39	r	12.730	0.007
1m0-08	58866.39	i	12.999	0.006
1m0-08	58866.39	i	12.992	0.006
1m0-05	58870.28	i	12.800	0.007
1m0-13	58874.98	B	12.256	0.022
1m0-13	58874.98	V	12.088	0.024
1m0-13	58874.98	V	12.156	0.024
1m0-13	58874.99	i	12.839	0.007
1m0-13	58874.99	i	12.779	0.011
1m0-05	58879.24	i	13.235	0.008
1m0-05	58879.24	i	13.272	0.008
1m0-10	58882.03	i	13.490	0.021
1m0-12	58883.06	i	13.559	0.017
1m0-12	58883.06	i	13.339	0.021
1m0-10	58884.07	i	13.338	0.014
1m0-10	58884.07	i	13.353	0.016
1m0-04	58885.26	B	13.438	0.022
1m0-04	58885.26	B	13.443	0.022
1m0-04	58885.27	i	13.510	0.007
1m0-04	58885.27	i	13.528	0.008
1m0-09	58887.25	B	13.536	0.026
1m0-09	58887.25	B	13.511	0.025
1m0-13	58888.98	B	13.979	0.024
1m0-13	58888.98	B	13.851	0.059
1m0-13	58888.98	V	13.044	0.028
1m0-13	58888.98	V	13.039	0.027
1m0-13	58888.98	g	13.372	0.010
1m0-13	58888.98	r	12.874	0.011
1m0-13	58888.98	i	13.278	0.014
1m0-13	58888.98	i	13.032	0.036
1m0-09	58890.22	B	13.961	0.025
1m0-09	58890.22	B	13.950	0.027
1m0-13	58893.05	B	14.658	0.029
1m0-13	58893.05	B	14.667	0.029
1m0-13	58893.06	V	13.471	0.032
1m0-13	58893.06	V	13.459	0.032
1m0-13	58893.06	g	13.933	0.012
1m0-12	58893.10	B	14.544	0.025
1m0-12	58893.10	B	14.558	0.024
1m0-12	58895.92	B	14.736	0.023
1m0-12	58895.92	B	14.754	0.024
1m0-12	58895.93	V	13.574	0.025
1m0-12	58895.93	V	13.565	0.025
1m0-12	58895.93	g	14.227	0.009
1m0-12	58895.93	g	14.216	0.010
1m0-12	58895.93	r	13.259	0.010
1m0-12	58895.93	r	13.277	0.013
1m0-12	58895.93	i	13.401	0.009
1m0-12	58895.93	i	13.324	0.014
1m0-11	58900.71	g	14.959	0.050
1m0-11	58900.71	r	13.828	0.016
1m0-11	58900.71	r	13.829	0.018
1m0-11	58900.71	i	13.454	0.050
1m0-11	58900.71	i	13.345	0.053
1m0-04	58904.30	B	15.231	0.025
1m0-04	58904.30	B	15.368	0.022
1m0-04	58904.30	V	14.140	0.032
1m0-04	58904.30	V	14.103	0.030
1m0-04	58904.30	g	14.825	0.009
1m0-04	58904.30	g	14.837	0.010
1m0-04	58904.30	r	14.005	0.009
1m0-04	58904.30	r	14.049	0.011
1m0-04	58904.30	i	13.992	0.017
1m0-04	58904.30	i	13.915	0.015
1m0-11	58907.68	B	15.391	0.022
1m0-11	58907.68	B	15.406	0.022
1m0-11	58907.68	V	14.341	0.025
1m0-11	58907.68	V	14.336	0.024
1m0-11	58907.68	g	14.877	0.009
1m0-11	58907.68	g	14.884	0.008
1m0-11	58907.68	r	14.087	0.009
1m0-11	58907.68	r	14.070	0.009
1m0-11	58907.68	i	14.203	0.008
1m0-11	58907.69	i	14.231	0.010
1m0-04	58910.17	B	15.405	0.023
1m0-04	58910.17	B	15.437	0.023
1m0-04	58910.17	V	14.405	0.025
1m0-04	58910.17	V	14.440	0.025
1m0-04	58910.18	g	14.921	0.009
1m0-04	58910.18	g	14.894	0.011
1m0-04	58910.18	r	14.218	0.010
1m0-04	58910.18	r	14.208	0.011
1m0-04	58910.18	i	14.306	0.018
1m0-04	58910.18	i	14.264	0.022
1m0-05	58913.38	B	15.477	0.022
1m0-05	58913.38	V	14.524	0.025
1m0-05	58913.39	V	14.518	0.025
1m0-05	58913.39	g	15.013	0.010
1m0-05	58913.39	r	14.346	0.010
1m0-05	58913.39	i	14.554	0.013
1m0-09	58916.38	B	15.147	0.025
1m0-09	58916.38	V	14.214	0.030
1m0-09	58916.38	V	14.696	0.027
1m0-09	58916.38	g	15.238	0.013
1m0-09	58916.38	g	15.217	0.013
1m0-09	58916.38	r	14.358	0.019
1m0-09	58916.38	r	14.318	0.031
1m0-09	58916.38	i	14.305	0.026
1m0-09	58916.38	i	14.311	0.015
1m0-04	58921.14	B	15.524	0.025
1m0-04	58921.14	B	15.508	0.024
1m0-04	58921.14	V	14.679	0.031
1m0-04	58921.14	V	14.665	0.028
1m0-04	58921.14	g	15.136	0.009
1m0-04	58921.14	g	15.143	0.010
1m0-04	58921.14	r	14.634	0.009
1m0-04	58921.15	r	14.638	0.009
1m0-04	58921.15	i	14.751	0.020
1m0-04	58921.15	i	14.869	0.010

Table 8. Table 8: LCO observations of SN 2020nvb. All photometry is from template-subtracted images.

Telescope	MJD	Filter	Mag	Mag Err
1m0-13	59031.73	B	13.094	0.020
1m0-13	59031.73	V	12.977	0.022
1m0-13	59031.73	g	12.938	0.006
1m0-13	59031.74	r	12.942	0.005
1m0-13	59031.74	i	13.209	0.008
1m0-13	59032.79	B	12.791	0.021
1m0-13	59032.79	V	12.760	0.024
1m0-13	59032.79	g	12.517	0.008
1m0-13	59032.79	r	12.688	0.010
1m0-13	59032.79	i	13.073	0.017
1m0-12	59033.80	B	12.635	0.020
1m0-12	59033.80	V	12.574	0.022
1m0-12	59033.80	i	12.938	0.006
1m0-11	59035.41	B	12.522	0.020
1m0-11	59035.41	V	12.513	0.022
1m0-11	59037.40	V	12.364	0.022
1m0-11	59037.41	i	12.886	0.005
1m0-10	59038.70	B	12.417	0.020
1m0-10	59038.70	V	12.306	0.022
1m0-10	59038.71	i	12.910	0.006
1m0-12	59047.76	B	12.954	0.020
1m0-12	59047.76	V	12.560	0.022
1m0-12	59047.76	r	12.756	0.005
1m0-12	59047.76	i	13.492	0.009
1m0-11	59051.36	B	13.554	0.021
1m0-11	59051.36	V	12.957	0.022
1m0-11	59051.36	g	13.036	0.006
1m0-11	59051.36	r	12.934	0.006
1m0-11	59051.36	i	13.456	0.010
1m0-13	59053.73	V	12.956	0.022
1m0-13	59053.73	g	13.308	0.006
1m0-13	59053.73	r	13.052	0.018
1m0-13	59053.73	i	13.512	0.007
0m4-07	59057.71	V	13.368	0.049
0m4-07	59057.71	r	13.231	0.062
0m4-04	59060.25	B	15.047	0.050
0m4-04	59060.25	V	13.528	0.030
0m4-04	59060.25	g	14.015	0.019
1m0-10	59060.72	V	13.452	0.022
1m0-10	59060.72	g	14.166	0.005
1m0-10	59060.72	r	13.276	0.005
1m0-10	59060.72	i	13.416	0.006
1m0-10	59063.72	B	14.839	0.021
1m0-10	59063.72	V	13.710	0.022
1m0-10	59063.72	g	14.459	0.006
1m0-10	59063.72	r	13.508	0.005
1m0-10	59063.72	i	13.624	0.008
0m4-03	59066.35	V	14.024	0.035
0m4-03	59066.35	g	14.534	0.028
0m4-03	59066.35	r	13.611	0.030
0m4-03	59066.35	i	13.658	0.041
1m0-13	59066.72	B	15.192	0.026
1m0-13	59066.72	V	14.028	0.025
1m0-13	59066.72	g	14.704	0.009
1m0-13	59066.72	r	13.810	0.011
1m0-13	59066.72	i	13.833	0.012
1m0-11	59071.38	B	15.154	0.024
1m0-11	59071.38	V	14.167	0.024
1m0-11	59071.38	g	14.714	0.010
1m0-11	59071.38	r	13.978	0.008
1m0-11	59071.38	i	14.012	0.012

Equations

$$P=\{\begin{array}{ll}10^{\left(A\ast \left(m-m_{\text{TRGB}}\right)+B\right)},& \text{if }m-m_{\text{TRGB}}\ge 0\\ 10^{\left(C\ast (m-m_{\text{TRGB}})\right)},& \text{if }m-m_{\text{TRGB}}<0\end{array}$$

$${\text{F814W}}_{\text{corrected}}={\text{F814W}}_{\text{fiducial}}$$

$$+b_{\text{linear}}\left[\left(\text{F606W-F814W}\right)-1.1\right]$$

$${\mu }_{\text{SALT2}}=m_B+\alpha x_1-\beta c-M_B$$

$$\begin{array}{rl}\text{calibrators:}& M_B=m_B^j-{\mu }_{\mathrm{TRGB}}^j\\ & +\alpha x_1^j-\beta c^j+\gamma (\log M_{\ast }^j-10)\\ \text{Hubble-flow:}& 5\log d_L(H_0,{\Omega }_M,{\Omega }_{\Lambda })=m_B^k-M_B\\ & +\alpha x_1^k-\beta c^k+\gamma (\log M_{\ast }^k-10)-25\end{array}$$

$$\begin{array}{rl}\text{calibrators:}& {\sigma }_j^2={\sigma }_{m_B,j}^2+{\sigma }_{{\mu }_{\mathrm{TRGB}},j}^2\\ & +{\alpha }^2{\sigma }_{x_1,j}^2+{\beta }^2{\sigma }_{c,j}^2+2\alpha {\mathrm{cov}}_j(m_B,x_1)\\ & -2\beta {\mathrm{cov}}_j(m_B,c)-2\alpha \beta {\mathrm{cov}}_j(x_1,c)\\ & +{\gamma }^2{\sigma }_{\log M_{\ast },j}^2+{\sigma }_{\mathrm{int}}^2\\ \text{Hubble-flow:}& {\sigma }_k^2={\sigma }_{m_B,k}^2+{\sigma }_{z,\mathrm{mag},k}^2+{\sigma }_{\mathrm{pec},\mathrm{mag},k}^2\\ & +{\alpha }^2{\sigma }_{x_1,k}^2+{\beta }^2{\sigma }_{c,k}^2+2\alpha {\mathrm{cov}}_k(m_B,x_1)\\ & -2\beta {\mathrm{cov}}_k(m_B,c)-2\alpha \beta {\mathrm{cov}}_k(x_1,c)\\ & +{\gamma }^2{\sigma }_{\log M_{\ast },k}^2+{\sigma }_{\mathrm{int}}^2\end{array}$$

Full text

\uatDistance Indicators394 — \uatGalaxy Distances590 — — \uatHertzsprung Russell diagram725 — \uatHubble constant758 — \uatHubble Space Telescope761 — \uatRed giant tip1371 — \uatStandard Candles1563 — \uatStellar Astronomy1583 — \uatType Ia supernovae1728

††journal: ApJ

show]mjbnewman25astro@gmail.com

1 Introduction

At present, the most precise measurements of the current expansion rate of the Universe use a three-rung local distance ladder in which type-Ia supernovae (SNe Ia) play a central role (e.g., A. G. Riess et al., 2016, 2022; S. Dhawan et al., 2018; W. L. Freedman et al., 2019, 2025). The high and standardizable luminosities of SNe Ia make them the best choice to extend the distance ladder into the smooth Hubble flow, where cosmological redshifts dominate observed recession velocities. Because SNe Ia occur only once every few centuries in galaxies like the Milky Way, their luminosities cannot be calibrated directly through primary (geometric) distance indicators, requiring instead a middle rung to measure the Hubble constant. The best established of these secondary distance indicators are Cepheid variable stars (e.g., H. S. Leavitt & E. C. Pickering, 1912; M. G. Lee et al., 1993; L. Ferrarese et al., 2000; A. G. Riess et al., 2016, 2024) and the tip of the red-giant branch (TRGB; e.g., M. G. Lee et al., 1993; S. Sakai et al., 1997; M. Bellazzini et al., 2001; D. Makarov et al., 2006; L. Rizzi et al., 2007; I. S. Jang & M. G. Lee, 2017; W. L. Freedman et al., 2019; W. L. Freedman, 2021; W. L. Freedman et al., 2025; D. Scolnic et al., 2023; S. Li et al., 2025).

Normal SNe Ia demonstrate broad homogeneity in their light curve properties (e.g., Z.-W. Liu et al., 2023) with quantifiable variety that allows them to be standardized for use as precise cosmological distance indicators (e.g., M. M. Phillips, 1993; M. Hamuy et al., 1996; R. Tripp, 1998; J. Guy et al., 2007; S. Jha et al., 2007; A. G. Riess et al., 2016; W. D. Kenworthy et al., 2021; K. S. Mandel et al., 2022). SN Ia light curves and luminosities (before and after standardization) are also correlated with host-galaxy properties, including host mass, star formation rate, and larger-scale environment (e.g., M. Hamuy et al., 1995; M. Sullivan et al., 2010; P. L. Kelly et al., 2010; M. Sullivan et al., 2010; C. R. Burns et al., 2018; M. Rigault et al., 2020; S. A. Uddin et al., 2020; C. Larison et al., 2024; M. Ginolin et al., 2025a, b; R. Senzel et al., 2025).

The quest for increasing precision and accuracy in measuring $H_{0}$ has made it imperative to limit systematic uncertainties in the distance ladder (W. L. Freedman, 2021; A. G. Riess et al., 2022). For SNe Ia, this means ensuring that samples are consistent across the second and third rungs in all properties that could be correlated with luminosity. However, by necessity, the chosen secondary distance indicator determines the type(s) of SN Ia host galaxies. Specifically classical Cepheids, being young stars, are only found in sufficient numbers in star-forming galaxies.

The TRGB method, conversely, can be applied to any system hosting stellar populations older than a few Gyr, requiring only a well-populated (and well-observed) red-giant branch (RGB). Specifically, it can be used to calibrate SNe Ia in star-forming and quiescent host galaxies of all morphologies (e.g., J. Mould & S. Sakai, 2008; R. B. Tully, 2023). However, the TRGB method has been limited by past observing facilities in its effective distance range ($D\leq 20$ Mpc), restricting its application to more nearby galaxies that either can, or have already been, calibrated with Cepheids (e.g., A. G. Riess et al., 2016; R. L. Beaton et al., 2019; A. G. Riess et al., 2024; W. L. Freedman et al., 2025). Until recently, the Hubble Space Telescope (HST ) was the preferred observatory for TRGB-based distance measurements. The JWST is now rapidly becoming the primary facility for the TRGB method based on its high angular resolution and increased sensitivity relative to the HST . In particular, precise TRGB calibrations are already available for JWST NIRCam wide filters (G. S. Anand et al., 2024a; M. J. B. Newman et al., 2024a, b; T. J. Hoyt et al., 2025; S. Li et al., 2025). Soon, the JWST in combination with the upcoming Nancy Grace Roman Space Telescope (Roman) holds the potential to drastically increase the number of galaxies with TRGB-based distance measurements and ultimately improve the calibration of SN Ia luminosities (e.g., G. S. Anand et al., 2021; K. Kraemer et al., 2023).

TRGB stars are evolved (4–12 Gyr), low-mass (0.8–2 $M_{\odot}$) stars that mark the end of the (first-ascent) red giant branch (RGB) stellar evolutionary phase and the onset of the helium flash. Due to the tight scaling relations between core mass, core radius, and core temperature of RGB stars, the bolometric luminosity at the TRGB is nearly uniform across stellar populations (A. Serenelli et al., 2017), but the luminosity at specific wavelengths can vary with stellar properties. Observationally, the TRGB appears as a sharp discontinuity and, given sufficient star counts, can be readily identified in a color-magnitude diagram (CMD; e.g., G. S. Da Costa & T. E. Armandroff, 1990; S. Sakai et al., 1996; B. F. Madore & W. L. Freedman, 1998). In practice, the TRGB method is typically applied at I-band equivalent wavelengths where the brightness of the TRGB is only modestly dependent on mass, metallicity, and age (M. G. Lee et al., 1993; S. Sakai et al., 1997; R. L. Beaton et al., 2019). The HST F814W filter, on both ACS or WFC3/UVIS, has been the standard TRGB filter choice for decades with continual improvements throughout, including updated zero-points, color-based metallicity corrections to the F814W luminosity function (LF), and more sophisticated methods to identify the TRGB in CMDs (e.g., L. Ferrarese et al., 2000; B. Méndez et al., 2002; M. Bellazzini et al., 2001; D. Makarov et al., 2006; L. Rizzi et al., 2007; V. A. Mager et al., 2008; W. L. Freedman, 2021, see § 3.1 for details;). Since the launch of the JWST , the TRGB method has been calibrated over a range of CMD combinations from the NIRCam F090W to the F444W filter. The TRGB feature in the HST F814W and JWST F090W filters exhibits a high degree of similarity in trends between the brightness versus metallicity and age (K. B. W. McQuinn et al., 2019; G. S. Anand et al., 2024a; M. J. B. Newman et al., 2024b). Thus, the JWST F090W filter is currently preferred for calibrating rungs of the local distance ladder. There are significant gains in employing the TRGB method at NIR wavelengths where the TRGB stars appear up to 2 mags brighter relative to the F814W/F090W filters. These observations can extend the feasible distance range of the TRGB method by at least double and dramatically increase the volume, and thus the number of galaxies with secure TRGB distances (see M. J. B. Newman et al., 2024b, for NIR JWST TRGB calibrations).

Because of the tension between the locally measured $H_{0}$ and the inference derived from the cosmic microwave background (CMB) assuming standard $\Lambda$CDM, there is a compelling interest in checking each rung of the local distance ladder. Studies have compared TRGB and Cepheid distances to the same galaxies, particularly those that are SN Ia hosts (e.g., W. L. Freedman, 2021; S. Dhawan et al., 2023; T. Araucaria Project et al., 2023; A. G. Riess et al., 2024). Such a comparison inherently restricts the sample to SN Ia host galaxies in the intersection of both methods, specifically including only star-forming hosts and their subset of the overall SN Ia population. This leaves out useful data, namely SNe Ia found in non-star-forming host galaxies, calibrated through the TRGB. Here we focus on a parallel distance ladder using TRGB and SNe Ia in massive, quiescent galaxies.

Approximately 40% of SNe Ia in the low-redshift Universe are found in early-type host galaxies (e.g., W. Li et al., 2011; O. Graur et al., 2017; R. Senzel et al., 2025) and the majority of these host galaxies have stellar masses on the high side of the “mass step” seen in SN Ia standardization, with $\log M_{*}/M_{\odot}\gtrsim 10$ (e.g., P. L. Kelly et al., 2010; S. A. Uddin et al., 2020; C. Larison et al., 2024; M. Ginolin et al., 2025a). However, only two out of 22 SN Ia host galaxies in the TRGB-calibrated second rung of the Carnegie-Chicago Hubble Program (CCHP R. L. Beaton et al., 2019; W. L. Freedman et al., 2019, 2025) sample are massive, early-type hosts (NGC 1316 and NGC 1404). Separately, the TRGB-surface brightness fluctuation (SBF) program (G. S. Anand et al., 2024a, 2025; J. B. Jensen et al., 2025), designed to anchor the SBF distance technique to the JWST TRGB and to measure $H_{0}$  is incrementally increasing the number of massive, early-type galaxies through ongoing JWST  programs at distances beyond the reach of HST. This includes the SN Ia host NGC 1380 (SN 1992A). Recently, two SNe Ia, SN 2020nvb and SN 2020ue, were observed in the massive, early-type galaxies NGC 4457 and NGC 4636, respectively. Both NGC 4457 and NGC 4636 are Virgo cluster galaxies located at a relatively large angular separation from the cluster center ($8$∘ and $10.8$∘, respectively, B. Vollmer et al., 2013; H. S. Park et al., 2010), and are morphologically classified as S0 and E/S0, respectively (B. Vollmer et al., 2013; M. P. Véron-Cetty & P. Véron, 2006).

Here, we present a pilot study to measure TRGB distances to four nearby ($D\leq 20$ Mpc), massive early-type SN Ia host galaxies – NGC 1316, NGC 1380, NGC 1404, and NGC 4457. We combine this with a TRGB distance for a similar SN Ia host galaxy NGC 4636 (G. S. Anand et al., 2025, taken from the literature as the data were still proprietary at the time we unblinded our analysis). We then apply this TRGB calibration to a sample of Hubble-flow SNe Ia with matched supernova light-curve and host-galaxy properties from the ZTF SN Ia Data Release 2 (M. Rigault et al., 2025) producing a “proof-of-concept” measurement of the Hubble constant.

In § 2, we present new HST observations for NGC 4457, archival HST data for NGC 1404 and NGC 1316, and archival JWST data for NGC 1380,introduce our data reduction methods, including image alignment, photometry, artificial star tests, and high-fidelity photometric catalog culling, and describe our SN Ia observations and data reduction process. In § 3, we detail and apply our methodology for TRGB-based distances, while the SN Ia light curve fitting and standardization methodology is described in § 4. In § 5, we discuss our fiducial calibrator and Hubble-flow SN Ia sample selection, describe our joint standardization and cosmological model fitting routine, and present the fiducial $H_{0}$, along with results from several sample selection variants. Finally, in § 6, we discuss and summarize our findings, including our recommendations for increasing the sample with JWST.

2 TRGB Observations and Data Reduction

Our galaxy sample includes the few early-type galaxies that host at least one SN Ia with well-sampled light curves and have the requisite observations for a precise TRGB measurement: NGC 1316, NGC 1380, NGC 1404, NGC 4457, and NGC 4636.

2.1 Reduction of Imaging Data

New observations of NGC 4457 were acquired with the HST Advanced Camera for Surveys Wide Field Camera (ACS/WFC; hereafter ACS) as part of HST -GO-16453 (PI McQuinn). We optimized the observations for TRGB measurements by imaging the outer stellar fields of the galaxy with the F814W and F606W filters. Coordinated parallel observations were obtained with the Wide Field Camera 3 imager ultraviolet and visible light channel (WFC3/UVIS; hereafter UVIS) F606W and F814W filters. Shown in the right panel of Figure 1 are the locations for our ACS (light blue) and UVIS (orange) observations.

Both NGC 1316 and NGC 1404 have deep archival ACS observations in F606W and F814W obtained for the CCHP (see D. Hatt et al., 2018a; T. J. Hoyt et al., 2021, respectively). The left and middle panels of Figure 1 show the ACS pointings for NGC 1316 and NGC 1404, respectively. We obtained the ACS exposures for NGC 1316 and NGC 1404 from the Mikulski Archive for Space Telescopes (MAST;https://dx.doi.org/10.17909/T9RP4V (catalog doi:10.17909/T9RP4V)).

NGC 1380 and NGC 4636 were both observed with the JWST in the Near Infrared Camera (NIRCam) detector imaging mode in the program JWST-GO-3055 (The TRGB-SBF Project; G. S. Anand et al., 2024a, 2025). For optimal TRGB distance measurements, the observing strategy employed the F090W filter and imaged galaxy fields beginning outside the dense galactic stellar discs and extending out to the galactic halos. At the time of this study, only NGC 1380 has publicly available archival data on MAST; NGC 4636 observations were still in their exclusive access period at the time of unblinding. We therefore check the consistency between the distance scale presented in G. S. Anand et al. (2024a) and this study by independently reducing NGC 1380 observations and measuring a TRGB distance with our own methods. We then adopt the TRGB distance for NGC 4636 presented in G. S. Anand et al. (2025, see § 3.1 for details). In Table 1, we provide several summary statistics for observations of the five galaxies.

2.1.1 Image Alignment

Precise alignment of HST observations is essential to measuring robust photometric properties of sources in the images. We use the tool TweakReg included in the DrizzlePac 3.0 Python package to perform image alignment and correct world coordinate system (WCS) solutions for individual exposures. First, we generate separate, matched source catalogs for each filter. Second, we run TweakReg on the combined source catalog to update all WCS solutions to a single reference frame with the constraint that, at minimum, 10 sources are cross-matched in all images. WCS solutions are considered satisfactory when the alignment’s root mean square (RMS) is better than 0.1 pixels.

Reference images for the ACS (or UVIS where applicable) F606W and F814W filters were generated using the AstroDrizzle software included in DrizzlePac 3.0. Here, all exposures in a given filter are combined into a single, deep image (i.e., a drizzled image). We use the F606W image as our final reference frame in the photometry processing (§ 2.1.2).

Alignment solutions for the JWST NIRCam observations of NGC 1380 were taken directly from the pipeline as they were sufficient for this work. We selected as a reference image the mosaiced i2d extension image in the F090W filter.

2.1.2 Photometry

We performed point-spread function (PSF) photometry on the well-aligned HST images to generate the data for our analysis. We use the software DOLPHOT, a modified version of the WFPC2-specific package HSTPhotwhich provides ACS, UVIS, and JWST NIRCam-specific modules (A. E. Dolphin, 2002; A. Dolphin, 2016; D. R. Weisz et al., 2024). As mentioned above for HST (see § 2.1.1), we set the drizzled F606W image as our reference frame for source identification owing to its high angular resolution. DOLPHOT also requires a parameter file with global and image-specific values. We adopted the values from the Panchromatic Hubble Andromeda Treasury (PHAT) and Panchromatic Hubble Andromeda: Triangulum Extended Region (PHATTER) surveys (see B. F. Williams et al., 2014, 2021). We use the parameter values for the short wavelength photometry from the JWST Early Release Science Resolved Stellar Populations Program (D. R. Weisz et al., 2024).

2.1.3 Artificial Star Tests

We use artificial star tests (ASTs) to quantify photometric completeness (recovery fraction) and photometric errors due to blending in our stellar catalogs. We injected $\sim 100,000$ artificial stars into each image. The spatial distribution of these artificial stars spans the full range of the sources determined in the photometry (§ 2.1.2). Fake star coverage of the CMDs is uniformly distributed to span the entire parameter space where real sources appear. We then run photometry on all fake stars in each image using DOLPHOT with identical parameter files to the original runs. The results of the ASTs are used directly in our TRGB fitting method (see § 3.1).

2.1.4 High-fidelity Photometric Catalogs

To optimize our stellar catalogs for robust TRGB measurements, we apply two different methods to cull the initial photometry: thresholds on how well an individual source is recovered and spatial cuts that follow each galaxy’s structural parameters. We find this combination highly effective in generating high-fidelity stellar catalogs for our galaxies.

We start by applying conservative, per filter cuts to the full photometric output. DOLPHOT provides important quality metrics for every source it identifies in the image. In particular, we focus on the crowding metric, a measure in magnitudes of how much brighter a star would be in the absence of nearby sources; the sharpness metric, a measure of how point-like or extended an object is (negative values are typically cosmic rays, a value of zero is a perfectly fit source, and positive values correspond to extended sources); and the signal-to-noise ratios (SNRs) (A. E. Dolphin, 2002; A. Dolphin, 2016; D. R. Weisz et al., 2024).

We first retain from the catalog only sources with lenient quality metric thresholds: crowding values of $\leq 1$ mag; sharpness between $-0.316$ and $0.316$ (sharpness${}^{2}\leq 0.1$); and SNRs of $\geq 2$ (non-TRGB magnitude filter, F606W or F150W) and F814W or F090W $\geq 4$. We then examine the remaining sources, iteratively tightening the crowding and sharpness2 cuts until no visual diffraction spike artifacts remain in the catalogs. These initial cuts are used to remove non-stellar sources and/or poorly recovered sources.

Next, we consider how the spatial distribution of sources in our targets impacts the quality of their photometry. The targeted fields in our galaxy sample’s HST observations are not uniformly placed relative to the galaxy’s center. We targeted two different fields around NGC 4457, with the ACS field as the primary. Part of the ACS field covers the outer stellar field, while the other part is nearer to the dense central nucleus. The UVIS field is almost entirely positioned in the outer stellar field. The HST observations for NGC 1316 are firmly in the outer stellar disk. The HST observations for NGC 1404 are placed similarly to the ACS field in NGC 4457; however, the NGC 1404 field overlaps directly with the galaxy’s center. For NGC 1380 the JWST  NIRCam instrument pointing covers the central region and an outer stellar disk region in NIRCam A and NIRCam B, respectively. The observing design is identical for NGC 4636.

For each galaxy, we follow the same procedure. First, we divide our initial source catalog into concentric elliptical annuli about the center of a galaxy. We adopt the structural parameters, namely ellipticity, position angle, and semi-major and minor axes, from the literature. These parameters and their references are provided in Table 2 for each galaxy. Concentric annuli are generated iteratively with approximately equal numbers of sources in each annulus. In Figure 2, left panel, we present an example of our culling method applied to the ACS field of NGC 1404. Concentric annuli (solid black curves) are over-plotted on the x/y locations of the source catalog in pixel coordinates. Second, we calculate the mean crowding value within each annulus. The sources are color-coded by their mean crowding value (see the color bar). Third, we apply a discrete derivative filter to the mean crowding values as a function of the annulus number. In Figure 2, right panel, we show the mean crowding versus annulus number (identical color-coding as in the left panel) and the derivative response (blue curve). We identify the peak in the derivative as the location of our spatial and crowding value cut-off.

Finally, in the left panels of Figure 3 to Figure 6, we show the spatial distribution of high-fidelity sources (black points) and sources excluded from further consideration (blue points). The final DOLPHOT quality metric thresholds are summarized in Table 3. These thresholds include the number of stars, first with only the strict quality metric cuts applied ($N_{*}$) and second, including both the quality metric and spatial cuts ($N_{*,\text{SMA}\geq\text{SMA}_{\text{inner}}}$). We note two additional spatial considerations we made in creating our high-fidelity catalogs. First, as shown in the left panel of Figure 3, we excluded from the NGC 1316 catalog sources in the region around the background galaxy 2MASS-J03231562-3719444. Second, shown in the left panel of Figure 5, we excluded from the NGC 1404 catalog sources about the nearby background dwarf galaxy FCC B1281. Sources excluded from the catalog are also shown in CMD space in Figure 3 through Figure 6.

3 TRGB Distance Measurements: Methods & Results

This section describes the TRGB distance measurement technique and summarizes our findings.

3.1 TRGB Methodology

Since its first modern application by M. G. Lee et al. (1993), the techniques used to measure the location of the TRGB feature in CMDs have increased in quantity and sophistication. Early in its history, the apparent magnitude of the TRGB was often identified visually in a CMD. M. G. Lee et al. (1993) introduced a formalized method for identifying the TRGB by first generating an I band LF (i.e., a 1-D histogram) marginalized over the color of the stars in a CMD and second convolving an edge-detection filter, a zero-sum Sobel kernel of the form $[-2,0,+2]$. The Sobel kernel has since been revised in several ways to account for sources of uncertainty that can act to shift the location of the TRGB in a LF (e.g., various forms of Gaussian smoothing, S. Sakai et al., 1996; D. Hatt et al., 2018a; R. L. Beaton et al., 2019).

As an alternative to the Sobel-kernel approach, a more sophisticated Bayesian maximum likelihood -based technique was developed to measure the TRGB. This technique includes a parametric form for the RGB LF, which it uses to fit the observed LF (see B. Méndez et al., 2002; D. Makarov et al., 2006). The probability distribution includes terms that account for the photometric uncertainty distribution and completeness function from ASTs (see D. Makarov et al., 2006). For our TRGB measurements, we adopt the same theoretical LF form used in D. Makarov et al. (2006):

[TABLE]

where A characterizes the RGB slope with a normal prior of $\mu_{\text{A}}=0.3$ and uncertainty of $\sigma_{\text{A}}=0.07$, B represents the RGB jump, and C is the AGB slope with a normal prior of $\mu_{\text{C}}=0.3$ and an uncertainty of $\sigma_{\text{C}}=0.2$. Parameters A, B, and C are free parameters in our application. The maximum likelihood measurement uncertainty is based on the range of solutions returning the log of the probability, $\log{(P)}$, within 0.5 of the maximum.

In addition, we consider and account for the extended stellar metallicity/age in our galaxy sample. It is well-established that the TRGB brightness is approximately constant in F814W for only the oldest and most metal-poor RGB stars. At higher metallicities and/or younger ages, the TRGB deviates from a constant brightness (e.g., M. G. Lee et al., 1993; L. Rizzi et al., 2007). In practice, this metallicity/age effect is traced by the color of the stars. We correct for this color-based metallicity dependence and adopt the calibration from I. S. Jang & M. G. Lee (2017), the quadratic transformation (QT) system, derived for the HST filters (see also T. J. Hoyt et al., 2021). We reproduce the calibration here, over the applicable color range F606W$-$F814W$>1.5$, for convenience:

[TABLE]

where F814W${}_{\text{fiducial}}$ is the original apparent magnitude of the stars, F814W${}_{\text{corrected}}$ is the color-corrected magnitude (i.e., the rectified magnitude), and $a_{\text{quadratic}}=0.159\pm 0.025$ and $b_{\text{linear}}=-0.047\pm 0.039$ are the best-fit parameters for the quadratic and linear terms, respectively. The TRGB is found to be constant in brightness for colors F606W-F814W$\leq 1.5$ mag; therefore, we apply no correction for stars with colors blueward of 1.5 mag. The stars with colors F606W-F814W$>1.5$ mag are color-corrected to the pivot point of F606W-F814W$=1.1$ mag where the TRGB is still flat, and thus, the apparent magnitude is constant. I. S. Jang & M. G. Lee (2017) note that their pivot point accounts for differences between the color of the stars in the observed galaxy and the color used to anchor the zero-point in their calibration. We apply this correction to F814W magnitudes in the high-fidelity photometry catalogs on a per-star basis. We apply the color-correction before fitting for the TRGB to increase the contrast of the TRGB discontinuity in the LF.

A similar metallicity effect appears for the stars near the TRGB in JWST filters. When observed in the $F090W$ filter, the TRGB is found to be approximately constant as a function of color up to a red color limit that depends on the second filter paired with the $F090W$ filter. Stars redder than the characteristic color appear fainter, an effect dominated by higher metallicity in the atmospheres of RGB stars (G. S. Anand et al., 2021; M. J. B. Newman et al., 2024b). In the middle panel of Figure 4 we demonstrate the color dependent TRGB brightness in the JWST  $F090W$ and $F150W$ filters. (M. J. B. Newman et al., 2024b) calibrated the JWST  TRGB color dependence and zeropoint for 18 combinations of NIRCam wide filters, including $F090W-F150W$ vs $F090W$. They report that the TRGB brightness is consistent with being constant in $F090W$ over the color range $1.15\text{ mag }\leq F090W-F150W\leq 1.68\text{ mag}$ The stars we observe in NGC 1380 span a color range of $1.4\text{ mag }\leq F090W-F150W\leq 2.8\text{ mag}$. At present there is no empirical calibration for TRGB stars beyond the color of $F090W-F150W=1.68$ mag in the literature, though this is an active area of study. Therefore, we consider only stars for the TRGB brightness measurement in NGC 1380 within the calibrated color range from (M. J. B. Newman et al., 2024b).

For all CMDs, HST  or JWST   we impose color and apparent magnitude constraints to select stars that are consistent with RGB stars. Stars blueward of the RGB (e.g., main sequence or helium-burning stars) can reduce the strength of the discontinuity corresponding to the TRGB in the LF and bias the final measurement. All photometry is foreground-extinction-corrected before fitting (see column 7 in Table 1).

3.2 TRGB Fitting Results

We present our final TRGB measurements for the targets NGC 1316, NGC 1380, NGC 1404, and NGC 4457 in Table 4 and compare to results published in the literature. The table includes the TRGB magnitudes with statistical uncertainties, and the distance moduli and distances with separated statistical and systematic uncertainties. For galaxies observed with HST we use the F814W TRGB zero-point from W. L. Freedman (2021) of $M^{TRGB}_{F814W}=-4.049\pm 0.015\text{ (stat) }\pm 0.035\text{ (sys) }$ mag. For NGC 1380 we use the JWST F090W TRGB zero-point from M. J. B. Newman et al. (2024b) of $M^{TRGB}_{F090W}=-4.32\pm 0.02\text{ (stat) }\pm 0.04\text{ (sys) }$ mag.

In the middle panels of Figure 3 through Figure 6, we present CMDs containing only high-fidelity sources for the three galaxies. The TRGB (dashed black lines) and statistical uncertainties (shaded orange regions) are marked in each CMD. We also include CMDs of only sources rejected/excluded from the high-fidelity catalogs. We include the final TRGB fits in these CMDs as a reference point. The number of rejected sources varies between the targets.

We measure an HST TRGB magnitude and distance modulus to NGC 1316 $m^{TRGB}_{F814W}=27.45^{+0.06}_{-0.01}\text{ (stat)}$ mag and $\mu=31.50^{+0.06}_{-0.03}\text{ (stat) }\pm 0.04\text{ (sys) }$ mag. This result is in excellent agreement with the result from D. Hatt et al. (2018a) who reported a TRGB magnitude of $m^{TRGB}_{F814W}=27.40\pm 0.04$ mag and, after applying the zeropoint adopted in this study, a corresponding extinction-corrected distance modulus $\mu=31.42\pm 0.04\text{ (stat) }\pm 0.04\text{ (sys) }$ mag.

For NGC 1404 we compare our HST TRGB results of $m^{TRGB}_{F814W}=27.18^{+0.09}_{-0.07}$ mag and $\mu=31.23^{+0.09}_{-0.07}\text{ (stat) }\pm 0.04\text{ (sys) }$ mag to two studies in the literature. We note that while all studies use the same HST observations (GO-15642), their approaches to PSF-fitting photometry, TRGB fitting methodology (including the adopted zeropoint) may differ. First, we find broad agreement with T. J. Hoyt et al. (2021) who report $m^{TRGB}_{F814W}=27.32\pm 0.03$ mag and extinction-corrected and zeropoint updated $\mu=31.35\pm{0.04}\text{ (stat) }\pm{0.04}\text{ (sys) }$ mag. G. S. Anand et al. (2024a) reported an HST TRGB distance modulus of $\mu=31.28\pm{0.10}$ mag (reproduced from the Extragalactic Distance Database with an updated zeropoint replacing $M^{TRGB}_{F814W}=-4.04$ mag; G. S. Anand et al., 2021) and measured a new JWST TRGB distance modulus of $\mu=31.364\pm{0.067}$ mag. While the three reported HST -based distance moduli show broad agreement, the best-fit distance moduli show a marginal preference towards brighter (lower/closer) values. However, including the latest JWST -based TRGB result from G. S. Anand et al. (2024a) which skews toward a fainter magnitude (more distant) may hint at the need for additional observations of NGC 1404 that span a range of distance from the center to help reduce uncertainty.

For NGC 4457 there is only one TRGB-based distance modulus reported in the literature (S. Li et al., 2025). Compared to the this study, where we find $m^{TRGB}_{F814W}=27.07^{+0.01}_{-0.03}$ mag and $\mu=31.12^{+0.01}_{-0.03}\text{ (stat) }\pm 0.04\text{ (sys) }$ mag, S. Li et al. (2025) report a brighter TRGB magnitude, $m^{TRGB}_{F814W}=27.00\pm 0.10$ mag, and lower distance modulus, $\mu=31.0\pm 0.1$ mag. Within the uncertainties we find broad agreement between the two measures.

Finally, we compare to the single reported JWST TRGB-based distance to NGC 1380 in the literature G. S. Anand et al. (2024b). We find that our TRGB value of $m^{TRGB}_{F090W}=27.00\pm 0.01\text{ (stat) }$ mag is in good agreement with the extinction-corrected TRGB magnitude reported in G. S. Anand et al. (2024a). In particular, we find excellent agreement with their reported value measured in NIRCam module A detector 1 (A1) of $m^{TRGB}_{F090W}=27.04$ mag, while the value for NIRCam A2 is less consistent toward fainter magnitudes (i.e., NGC 1380 is further away). We use this agreement as a justification to add the galaxy NGC 4636, reported in G. S. Anand et al. (2025), to our sample bringing the final sample size to five SN Ia host galaxies. We adopt the distance modulus for NGC 4636 from G. S. Anand et al. (2025) of $\mu^{TRGB}_{F090W}=31.12\pm 0.07$ and include the information in Table 4.

We note that all TRGB values reported in G. S. Anand et al. (2025) are based on magnitudes measured in the Vega-Vega flux calibration system, while in this study we use the current Sirius-Vega calibration. Nevertheless, the TRGB zero-points are internally consistent between the two studies.

4 Type Ia Supernova Observations and Standardization

The five early-type galaxies in our TRGB sample were chosen because they hosted well-measured SNe Ia that can provide standardized distances. Three galaxies hosted four SNe Ia with already published photometry. NGC 1316 was the host of SN 2006dd, for which we use data from M. Stritzinger et al. (2010). NGC 1380 hosted SN 1992A, for which we use the photometry presented by N. B. Suntzeff (1996) and also compiled by G. Altavilla et al. (2004). NGC 1404 was the host of two SNe Ia, SN 2007on, and SN 2011iv; we use the data presented by C. Gall et al. (2018) to make distance measurements using these two sibling SNe Ia.

For the two SNe Ia in our remaining TRGB galaxies, SN 2020ue in NGC 4636 and SN 2020nvb in NGC 4457, we present BVgri photometry here. These data were obtained via the Sinistro cameras on Las Cumbres Observatory’s network of robotic telescopes (T. M. Brown et al., 2013), through the Global Supernova Project (GSP) collaboration. The data were reduced in the standard method with lcogtsnpipe111https://github.com/LCOGT/lcogtsnpipe (S. Valenti et al., 2016), a PyRAF-based image reduction pipeline. Instrumental magnitudes were attained with PSF photometry after template subtraction and calibrated with external standards catalogs. Due to the brightness of both SNe and the proximity of SN 2020nvb to the center of its host, some of the observations were saturated and unusable; these epochs were removed from the analysis. The light curves of SN 2020ue and SN 2020nvb are shown in the lower panels of Figure 7 and the photometry is given in Appendix B (Table 7 and Table 8).

Cosmological application of SN Ia photometry in the optical requires standardization of their absolute magnitudes, for which many methods exist (e.g., M. M. Phillips, 1993; M. Hamuy et al., 1996; A. G. Riess et al., 1996; S. Perlmutter et al., 1997; R. Tripp, 1998; J. Guy et al., 2005; S. Jha et al., 2007; C. R. Burns et al., 2011; W. D. Kenworthy et al., 2021). Here we focus on two approaches to SN Ia light curve standardization, SALT2 (J. Guy et al., 2007) and BayeSN (K. S. Mandel et al., 2022).

SALT2 is the most commonly used methodology for SN Ia cosmology; it fits a multicolor SN Ia light curve with three parameters: $m_{B}$, which represents the apparent magnitude in the B band222Technically, SALT2 fits in flux space using an amplitude parameter $x_{0}$ to scale the model flux to the observations. The model is constructed such that the peak $B$-band apparent magnitude is given by $m_{B}=-2.5\log x_{0}+10.5$ (e.g., G. Taylor et al., 2023).; $x_{1}$, which parameterizes the decline rate; and $c$, which reflects the SN color (corresponding roughly to B$-$V at peak). A lower $x_{1}$ indicates a faster-evolving light curve, and a higher c denotes a redder color. The fiducial “standard” SN Ia with $x_{1}=0$ and $c=0$ is established during training of the model; here we adopt the SALT2 training by G. Taylor et al. (2023). The light-curve shape parameter $x_{1}$ is further normalized so that its standard deviation over the cosmological SN Ia sample is $\sigma(x_{1})\approx 1$. The color parameter $c$ combines the effects of intrinsic variations in SN Ia color with dust reddening in the SN host galaxy; we separately correct for Milky Way dust assuming fidelity of the D. J. Schlegel et al. (1998) dust maps and E. F. Schlafly & D. P. Finkbeiner (2011) recalibration.

SALT2 light-curve fit parameters for our TRGB-calibrated SN Ia are given in Table 5 and the fits are shown in Figure 7. We can immediately recognize that our sample is a biased draw from the SN Ia population: all of the objects have $x_{1}<0$; indeed five of the six calibrators have $x_{1}<-1.4$. A selection of early-type host galaxies with TRGB distances clearly leads to a fast-declining sample of SNe Ia. This correlation of SN Ia decline rate with host-galaxy environment is well known (e.g., D. Branch et al., 1996; M. Hamuy et al., 2000; D. A. Howell, 2001) and has clear import in cosmological applications (see § 5).

For the SALT2 parameterization, standardization is based on the R. Tripp (1998) method, with linear corrections based on $x_{1}$ and $c$:

[TABLE]

where $\mu_{\text{SALT2}}$ represents the inferred distance modulus, and $\alpha$, $\beta$, and $M_{B}$ are global parameters determined from a fit for a given sample. Depending on the cosmological application, $M_{B}$ is either calibrated by other distances (as we will do here with TRGB distances to infer $H_{0}$) or, if only relative distances are needed (e.g., in high-redshift SN Ia cosmology), $M_{B}$ can be marginalized over, assuming a value of $H_{0}$.

Large cosmological SN Ia samples show distance modulus residuals that are correlated with environmental properties, e.g., the “mass-step” in which SN Ia from low (stellar) mass host galaxies and high-mass host galaxies standardize to slightly different absolute magnitudes (P. L. Kelly et al., 2010; M. Sullivan et al., 2010; H. Lampeitl et al., 2010). Precision cosmology with such samples thus often includes an additional empirical correction to Equation 3 based on each SN Ia host galaxy stellar mass.

Importantly, because we are deliberately restricting our SN Ia sample to those hosted in massive, early-type galaxies, we cannot reliably use previous estimates of $\alpha$, $\beta$, and the mass step that were derived from larger SN Ia samples from a broader range of hosts. SN Ia properties correlate with host-galaxy environment before standardization (e.g., in their light-curve shape or $x_{1}$ distributions) as well as after standardization (e.g., the mass step). Indeed, it has recently been shown that the best-fit $\alpha$ varies depending on the decline rates ($x_{1}$) of the fit SNe Ia, with a larger $\alpha$ for faster-declining SNe Ia (P. Garnavich et al., 2023; C. Larison et al., 2024; M. Ginolin et al., 2025a). Because our sample of early-type hosts preferentially selects for such fast-declining SNe Ia, we must take care in consistently deriving and applying the empirical light-curve corrections.

In addition, because SALT2 is trained on cosmological samples where the majority of SNe Ia have larger $x_{1}$ values than our objects, it is worthwhile to explore light-curve models that may be better suited to faster-declining objects. Models with conventional decline-rate indicators like $\Delta m_{15}$ and SALT $x_{1}$ can run into difficulty in fitting fast-declining objects, and maximum light color (P. M. Garnavich et al., 2004) or the “color-stretch” $s_{BV}$ (C. R. Burns et al., 2014) may better parameterize these objects. Here, we also explore using a newer framework, BayeSN, a hierarchical Bayesian SN Ia light curve fitting method (K. S. Mandel et al., 2022). BayeSN separates SN Ia colors into components based on both host galaxy dust and intrinsic SN spectral energy distribution (SED) variations. We use the version of BayeSN that was trained by S. M. Ward et al. (2023) and present the best-fit model parameters ($A_{V}$, $\theta$) in Table 5. The BayeSN light curve fits to the photometry are shown in Figure 7; they do better at capturing the i-band behavior of our objects. Nevertheless, to best compare with other supernova cosmology analyses, we continue to use SALT2 in our distance measurements.

5 Measuring the Hubble Constant with a Parallel Distance Ladder: Proof of Concept

The SALT2 light curve fits of the TRGB-calibrated SN Ia in massive quiescent galaxies show that these objects comprise only a subset of the total SN Ia population. Because the statistical precision on $H_{0}$ measurements has been typically limited by the number of calibrator SNe Ia, interest in a “parallel” distance ladder using TRGB (rather than Cepheids) has generally nonetheless included all galaxies that can be measured, to get the largest sample. This means including TRGB distances measured in the halos of star-forming SN Ia host galaxies that also were used as Cepheid calibrators. The fraction of overlapping galaxies is relatively high: 19 out of 22 CCHP TRGB galaxies in Table 3 of W. L. Freedman et al. (2025) have Cepheid observations. Similarly, 23 out of 30 TRGB galaxies in the SH0ES compilation by S. Li et al. (2025, Table 3) have Cepheid data333In these counts, we include NGC 3627 (M66) and NGC 3368 (M96) as galaxies with Cepheid observations, even though their SNe Ia (SN 1989B and SN 1998bu, respectively) were too reddened to be considered for the A. G. Riess et al. (2022) Cepheid sample. From the S. Li et al. (2025) compilation, we also count NGC 4414 (host of SN 1974G and SN 2021J) as a Cepheid galaxy. Two additional galaxies were not counted, but may have Cepheid data: NGC 4666 (host of ASASSN-14lp), and the edge-on spiral NGC 7814 (SN 2021rhu). Including these in the count would further increase the overlap between TRGB and Cepheid hosts.. This overlap is scientifically valuable, because it enables direct comparison of TRGB and Cepheid distances to the same galaxies (A. G. Riess et al., 2024; W. L. Freedman et al., 2025), but it also limits the independence of the two approaches, using many of the same host galaxies and the same supernovae in the second and third rungs.

In this paper we are exploring a more truly parallel distance ladder, using TRGB distances to massive, quiescent SN Ia host galaxies only, for which Cepheid distances are not possible. Because the SNe Ia found in these environments are a special subsample, we aim to check for systematic differences in the inferred distance scale using these objects in both the calibrator and Hubble-flow samples, rather than combining them together with SNe Ia in star-forming hosts (whether calibrated by Cepheids or TRGB).

One downside of this approach is clearly the limited sample size of the calibrators and potentially even Hubble flow objects. Samples grow with time, however, and in this case new large surveys like the Zwicky Transient Facility (ZTF; E. C. Bellm et al., 2019a, b; M. J. Graham et al., 2019) are providing a wealth of well-measured SNe Ia that can be used not only for better statistical precision, but perhaps more importantly, to allow subsample selection to test for systematic uncertainties. We view our analysis in this section as a “proof of concept” for future SN cosmology. Rather than generically treating all SNe Ia in one group, we imagine an approach with better-matched subsamples across redshifts, e.g., calibrators and the Hubble flow, with reduced or more-tailored SN standardization corrections and different sensitivities to systematics.

Recently, M. Rigault et al. (2025) present $gri$ light curves of over 3000 SNe Ia with $z<0.3$ as part of ZTF SN Ia DR2, massively increasing the low-redshift SN Ia sample, with numerous cosmological and astrophysical applications. For our purposes, we select from the large ZTF DR2 SN Ia sample objects that are matched to the environmental and light-curve properties of our TRGB-calibrated SN Ia in massive, quiescent galaxies.

5.1 Fiducial Sample Selection

Here we define “fiducial” sample cuts that we can apply uniformly to the calibrators and the Hubble flow, in the attempt to make these samples as similar as possible. Importantly, we defined the fiducial sample (and variants) independently of the inferred distances or the value of $H_{0}$; we blinded our analysis using a hidden, random offset to the TRGB distances when choosing our sample cuts, and then fixed the fiducial sample and analysis choices before unblinding.

We begin by limiting the ZTF DR2 SN Ia sample to what M. Rigault et al. (2025) define as the “complete” volume-limited sample with $z_{\rm helio}<0.06$ (see also C. Larison et al., 2024), comprising approximately 1000 SNe Ia. This is particularly helpful in our application where we are preferentially selecting faster-declining, lower-luminosity SNe Ia that are less likely to be represented at larger distances in the full ZTF DR2 SN Ia sample.

We further restrict the sample to include only the “cosmological” SNe Ia (sn_type = snia-cosmo). In addition to enforcing light-curve quality cuts, this selection excludes objects spectroscopically classified as peculiar. In particular, we exclude 91bg-like or 86G-like objects that are also found preferentially in the massive, quiescent galaxies (e.g., S. Taubenberger et al., 2008; J. S. Gallagher et al., 2008; F. H. Panther et al., 2019) that we are calibrating with TRGB. However, none of our calibrator SNe Ia are explicitly spectroscopically typed as 91bg-like or 86G-like, so we match this requirement in the Hubble-flow sample. With a larger calibrator sample, it may be possible to expand our analysis to include and standardize 86G-like and 91bg-like objects (e.g., O. Graur, 2024). A small number of Hubble-flow objects (SNe 2018ccl, 2019etc, 2020nef, 2020pwn, 2020sii, 2020acua, and 2020adii) are clear outliers compared to the rest of the fiducial sample, and are also excluded from our analysis.

The ZTF SN Ia DR2 compilation includes host-galaxy information for the sample, as well as SALT2 light curve fits. To match the calibrator galaxies, in our fiducial sample we select for massive, quiescent host galaxies with $\log(M_{*}/M_{\odot})>10$ and rest-frame host color $g-z>1$ (M. Ginolin et al., 2025a). Similarly, to match the calibrator SN Ia light curve properties, we restrict our fiducial sample to objects with SALT2 parameters $-2.0<x_{1}<0$ and $-0.2<c<+0.1$. Finally, in our fiducial sample, we require the SNe Ia to have well-measured spectroscopic redshifts ($\sigma_{z}\leq 0.001$) and to be in the smooth Hubble flow ($z_{\rm cosmo}>0.023$) in the CMB frame, after applying a flow correction444https://github.com/KSaid-1/pvhub (A. Carr et al., 2022). Even with these relatively strict cuts, the large overall ZTF cosmological sample still provides 124 Hubble-flow SNe Ia that are matched in their properties to the calibrators.

M. Rigault et al. (2025) caution against direct cosmological use of the ZTF SN Ia DR2 sample because of uncertain absolute photometric calibration of their forced-photometry methodology at the 0.05 mag level (though relative photometry, including across filters, is accurate to 0.01 mag). The ZTF data taken after November 2019 also suffer from a detector “pocket effect” causing a few-percent non-linearity in the measured flux. Both of these issues will be addressed in a future ZTF SN Ia data release. For our “proof-of-concept” demonstration here, we establish the ZTF zeropoints through cross-calibration against a sample of 28 SNe Ia measured in common with Las Cumbres Observatory $gri$ photometry (reduced as described in § 4 for SN 2020ue and SN 2020nvb). We compare individual photometric measurements as well as derived SALT2 parameters, and find consistency between both approaches. For simplicity, we apply zeropoint corrections to SALT2 $m_{B}$ directly. For light curves peaking before November 2019, based on the SALT2 fits to the Las Cumbres photometry, we adjust the ZTF-derived peak magnitudes as follows: $m_{B}=m_{B,{\rm ZTF}}+0.024\pm 0.035$ mag. For SNe Ia peaking after November 2019 (with the pocket effect), we find a negligible mean offset, but increase the peak magnitude uncertainty as $m_{B}=m_{B,{\rm ZTF}}\pm 0.026$ mag. In addition, for these objects we correct the pocket effect on the shape of the light curves with $x_{1}=x_{1,{\rm ZTF}}+0.14$ (M. Rigault et al., 2025). We also explore the effect of not making these corrections as a sample variant in our analysis. Because our cross-calibration is based on a relatively small number of objects in common, and because changes to the ZTF zeropoint would systematically move the Hubble-flow sample relative to our calibrators, we add an overall 2.5% systematic uncertainty in our derived $H_{0}$, corresponding to the estimated 0.05 mag zeropoint uncertainty.

5.2 Standardization and Model Fit

With our fiducial sample defined and ZTF photometry corrections applied, we proceed to standardize the Hubble-flow and calibrator SNe Ia, based on the R. Tripp (1998) formula above, but with the inclusion of a linear host-mass correction (see below). Following the approach of S. Dhawan et al. (2018), we jointly fit the TRGB-calibrated SNe Ia that determine $M_{B}$ (indexed by $j$) and the Hubble-flow SNe Ia (indexed by $k$), with

[TABLE]

where $M_{*}$ is the host-galaxy stellar mass (in solar masses), $d_{L}$ is the luminosity distance in Mpc, and $\Omega_{M}=1-\Omega_{\Lambda}=0.3$ is assumed. The total uncertainty for each object is given by the quadrature sum of various component uncertainties and we also include covariances in the SALT2 fit parameters (J. Marriner et al., 2011):

[TABLE]

where $\sigma_{m_{B}}$, $\sigma_{x_{1}}$, and $\sigma_{c}$ are the SALT2 fit uncertainties, cov() denotes their covariances, $\sigma_{\mu_{\rm TRGB}}$ is TRGB distance-modulus uncertainty (calibrators only), $\sigma_{z,{\rm mag}}$ and $\sigma_{\rm pec,mag}$ are redshift uncertainty and peculiar-velocity uncertainty (assumed to be $\pm$150 km s*-1*), converted to magnitudes (Hubble-flow objects only), $\sigma_{{\rm log}M_{*}}$ is the logarithmic host mass uncertainty, and $\sigma_{\rm int}$ is a fit parameter for the intrinsic (unmodeled) scatter in our combined SN sample.

We use the emcee package D. Foreman-Mackey et al. (2013) to perform a joint MCMC fit with six model parameters: $H_{0}$ (in conventional units of km s*-1* Mpc*-1*), $M_{B}$ (mag), $\alpha$, $\beta$, $\gamma$, and $\sigma_{\rm int}$ (mag)555Because our fiducial sample restricts $-2<x_{1}<0$, it would be better to apply the light-curve shape correction as $\alpha(x_{1}+1)$, so that there was no correction in the middle of the $x_{1}$ range. This would shift the definition of $M_{B}$ to its value for an $x_{1}=-1$ supernova and help to decorrelate $M_{B}$ and $\alpha$ (P. Garnavich et al., 2023). However, to ease comparison with looser sample cuts on $x_{1}$ as well as with literature results, we retain the traditional centering of $M_{B}$ at $x_{1}=0$. Following a convention in low-redshift supernova cosmology, we also report a version of the “intercept of the ridge line” that is well constrained by the Hubble-flow SNe Ia, in magnitude units: $-5\,a_{B}=M_{B}-5\log H_{0}+25$. Our Bayesian analysis uses uniform priors on all model parameters, enforcing $H_{0}>0$ and $\sigma_{\rm int}>0$.

5.3 Results

Figure 8 shows the high-quality Hubble diagram based on the 124 ZTF objects in the fiducial sample, after standardization, with five removed outliers (§ 5.1) displayed with lower opacity. Remarkably the RMS scatter about the best-fit model is just 0.106 mag, comparable to the best results in modern supernova cosmology, with an unmodeled scatter of just $\sigma_{\rm int}=0.079\pm 0.009$ mag. This is a testament to the high-quality ZTF photometry (M. Rigault et al., 2025) and clearly demonstrates that we have identified a homogeneous subsample of the SN Ia population.

Further validation of our approach is seen in our results for the SN correction coefficients $\alpha$, $\beta$, and $\gamma$, as illustrated in Figure 9. We isolate each of the correlations with light-curve shape ($x_{1}$), color ($c$), and host-galaxy stellar mass ($\log M_{*}/M_{\odot}$) in the figure panels. This figure also shows in low opacity the remainder of the full ZTF cosmological, volume-limited Hubble-flow sample, as well as our TRGB-calibrated SNe Ia, demonstrating our fiducial sample cut choices to best match the calibrator and Hubble-flow objects. The data points from our fiducial sample show significantly less scatter in all panels compared to the overall population.

Importantly, our model fit coefficients from the massive, quiescent host-galaxy SN Ia subsample are derived for our specific sample, and differ significantly from fits to the overall SN Ia population. Namely, we find $\alpha=0.209\pm 0.016$, a steeper $x_{1}$ correction coefficient than in typical analyses (e.g., $\alpha\approx 0.15$; D. Brout et al., 2022; DES Collaboration et al., 2024; D. Rubin et al., 2025). This is in accord with the steeper $x_{1}$ correction seen for fast-declining objects in other recent analyses (P. Garnavich et al., 2023; C. Larison et al., 2024; M. Ginolin et al., 2025a), and can also be seen in the top left panel of Figure 9, where including objects with larger $x_{1}>0$ would push the best-fit slope shallower.

Similarly, we find $\beta=2.11\pm 0.17$, a somewhat weaker color correction compared to full-sample analyses ($\beta\approx 3$), but consistent with SNe Ia in luminous red galaxies at higher redshift (R. Chen et al., 2022). Because the SALT2 approach uses a single color parameter that combines the effects of intrinsic color variations and host-galaxy dust reddening, it is not unexpected that our massive, quiescent host-galaxy SN Ia subsample would show differences here (likely with less dust and potentially different dust properties; e.g., D. Brout & D. Scolnic, 2021). The top right panel of Figure 9 suggests that widening the color distribution of the sample would indeed favor a steeper color correction.

The lower panel of Figure 9 explains our inclusion of a host-galaxy mass correction. Even though we have already restricted the fiducial sample to hosts with high stellar mass ($\log M_{*}/M_{\odot}>10$), we nevertheless find a best-fit $\gamma=0.111\pm 0.029$ mag dex*-1*, inconsistent with zero at 3.8$\sigma$ significance. Visually, a linear correction seems most appropriate for our restricted subsample, though were we to expand the sample in SN light curve properties (i.e., wider $x_{1}$ or $c$ ranges) or host-galaxy stellar mass, a “step” correction might then be preferred. We do not speculate on the cause of this host-galaxy mass trend within our fiducial subsample; rather, we are content that in our empirical approach, its inclusion is preferred and slightly reduces the Hubble diagram scatter. We further explore this correction in analysis variants below. Understanding this issue in more detail can be complicated, based on whether host-galaxy correlations are identified after light-curve and color standardization or if they are simultaneously fit with the supernova corrections, as we do here (M. Ginolin et al., 2025a; Y. S. Murakami & D. Scolnic, 2025).

Figure 9 highlights the comparison of the calibrator sample with the Hubble-flow objects. Choices in defining the fiducial sample cuts to best “match” the calibrator and Hubble-flow samples are somewhat subjective, and so we reiterate that these were made with the cosmological inferences blinded. Our fiducial cut on light-curve shape, with $-2<x_{1}<0$ excludes the calibrator SN 2007on in NGC 1404, substantially reducing the number of fiducial calibrators from 6 to 5. We choose the threshold $x_{1}>-2$ for two main reasons: first, the Hubble-flow sample becomes quite sparse below $x_{1}<-2$ (Figure 9, top left) and second, this is also the approximate threshold where SALT2 light curve fits begin to poorly differentiate different kinds of fast-declining SNe Ia (e.g., see Figure 4 right panel of C. R. Burns et al., 2014). Indeed, C. Gall et al. (2018) note significant luminosity differences between the more extreme SN 2007on and its “sibling” in NGC 1404, SN 2011iv, that we retain in our calibrator sample. In our analysis, SN 2007on is more clearly an outlier compared to the other calibrators and the Hubble flow objects seen in Figure 9. We consider analysis variants including SN 2007on in the next section.

Our TRGB-calibrated SNe Ia, and their individual implications for $H_{0}$, are shown in Figure 10. With SN 2007on excluded, the scatter in the calibrators ($\sigma=0.096$ mag) is consistent with the scatter seen in the Hubble-flow sample (Figure 8) and also consistent with the model (including intrinsic scatter), with the calibrators giving $\chi^{2}=4.0$ for effectively 4 degrees of freedom (5 calibrators minus 1 model parameter, $M_{B}$, that they constrain; the other model parameters are chiefly determined from the much larger Hubble-flow sample).

A corner plot (D. Foreman-Mackey, 2016) of our full model fit samples is shown in Figure 11. All model parameters are well constrained; when we quote point estimates of any of these, we use the posterior sample medians (50th percentile) with the 16th and 84th percentiles defining the approximate 1$\sigma$ confidence region. The quite symmetric posterior marginal distributions mean these estimates differ negligibly from estimates based on the sample means and standard deviations. Our full model analysis code is publicly available666https://github.com/mjbnewman/ETG-TRGB-SNIa.

Marginalizing over all nuisance parameters, our fiducial estimate of the Hubble constant using this parallel distance ladder with TRGB-calibrated distances to fast-declining SNe Ia, exclusively in massive, quiescent host galaxies, is $H_{0}=75.3\pm 1.7$ (stat) $\pm$ 2.4 km s*-1* Mpc*-1*. As described above, we include a 2.5% systematic uncertainty to account for the uncertain ZTF photometric zeropoint (and subsume into this any systematic offset between the ZTF Hubble-flow light curves versus the non-ZTF calibrator light curves). We also include a 2% (0.04 mag) correlated systematic uncertainty in the TRGB absolute calibrations and potential zeropoint offset between HST  F814W and JWST  F090W.

Our $H_{0}$ results are consistent with many other “high” values of the local expansion rate (e.g., E. Di Valentino et al., 2025). A simplistic comparison of our fiducial value (in this admittedly proof-of-concept approach), $H_{0}=75.3\pm 2.9$ km s*-1* Mpc*-1* with the CMB-derived $H_{0}=67.4\pm 0.5$ km s*-1* Mpc*-1* (Planck Collaboration et al., 2020) corresponds to a 2.7$\sigma$ indication of the Hubble tension. We stress that our underlying data are not independent of previous TRGB+SNe Ia analyses. In particular, we share in common with CCHP (W. L. Freedman et al., 2025) the HST TRGB data and SN light curves for NGC 1316/SN 2006dd and NGC 1404/SN 2007on+SN 2011iv, and we adopt their F814W TRGB zero point, but derive slightly different TRGB distances compared to T. J. Hoyt et al. (2021) as described above. Similarly, compared to the SH0ES analyses (A. G. Riess et al., 2024; S. Li et al., 2025), we share all HST TRGB data in common, excepting NGC 4636 (adopted from G. S. Anand et al., 2025), and share SN light curves except for the SN 2020ue and SN 2020nvb photometry newly presented here. Indeed, our HST TRGB results agree well with the SH0ES distances presented by S. Li et al. (2025) and our combined TRGB+SN inferences for $H_{0}$ are also consistent with those shown by S. Li et al. (2025, their Figure 3) for the same objects.

5.4 Analysis Variants