WorksPhysics and Astronomy
A DOLPHOT/Euclid Study of the Resolved, Faint Outer Milky Way Globular Cluster AM-1
Abstract
We present a resolved star study of AM-1, a distant ( kpc), faint () Milky Way (MW) globular cluster, based on Euclid Quick Data Release 1 (Q1) imaging. To perform photometry of the Euclid VIS and NISP imaging, we develop a Euclid-specific module for the crowded field stellar photometry package DOLPHOT. We construct and present empirical point spread function (PSF) models for VIS and NISP, along with a photometric calibration, entirely from the public Q1 images. Our photometry reaches point source depths of , 1-1.5 mag deeper than the public Q1 catalog, as expected for dedicated resolved star PSF photometry. We resolve the cluster to 1 mag below its main sequence turnoff in all bands. The Euclid photometry agrees with color-magnitude diagram (CMD) modeling of archival HST/WFPC2 imaging and with the 2MASS-anchored VISTA Hemisphere Survey to mag in , , and , with a 0.07 mag offset in . We find that AM-1 hosts an old (10.6 Gyr), intermediate-metallicity ([Fe/H] ) population, with a present-day stellar mass of and , and has a half-light radius of 12-16 pc with a round, low-concentration profile. These affirm earlier studies placing AM-1 among the most extended, diffuse outer-halo clusters, which dynamical work identifies as accreted. We describe and publicly release DOLPHOT v2.1, the Euclid module, and its empirical PSFs, which together enable Euclid resolved star studies ranging from faint MW systems such as AM-1 to more luminous galaxies throughout the Local Volume. We anticipate further refinements, including spatially varying PSFs, with Euclid Data Release 1.
Introduction
Faint dwarf galaxies and low-mass star clusters that populate the halo of the Milky Way (MW) are among our most sensitive probes of dark matter, reionization, and the hierarchical assembly of galaxies, and they have long anchored our understanding of CDM on small scales (e.g., J. S. Bullock & M. Boylan-Kolchin [11]; J. D. Simon [80]). Realizing the scientific potential of these low-luminosity systems requires finding them and characterizing their stellar populations, both of which are limited by the depth, area, and angular resolution of the available imaging.
The Sloan Digital Sky Survey (SDSS) opened this faint regime, with the discovery of ultra-faint dwarf galaxies (UFDs; e.g., B. Willman et al. [94]; D. B. Zucker et al. [96]; V. Belokurov et al. [8]). Since that time, deeper imaging from programs such as the Dark Energy Survey (DES; Dark Energy Survey Collaboration et al. [21]), Pan-STARRS (K. C. Chambers et al. [14]), the DECam Local Volume Exploration survey (DELVE; A. Drlica-Wagner et al. 2021), and the Ultraviolet Near-Infrared Optical Northern Survey (UNIONS; S. Gwyn et al. [42]) has roughly doubled the
Corresponding author: Daniel R. Weisz Email: [email protected]
NASA Hubble Fellow
satellite census again while pushing discovery to lower surface brightness and larger distances (e.g., K. Bechtol et al. [7]; S. E. Koposov et al. [50]; A. Drlica-Wagner et al. [27]; B. P. M. Laevens et al. [53]; S. Mau et al. [62]; W. Cerny et al. [13]; S. E. T. Smith et al. [83]). The Vera C. Rubin Observatory and its Legacy Survey of Space and Time (LSST) are expected to further expand the census of faint systems around the MW (e.g., Ž. Ivezić et al. [48]; B. Mutlu-Pakdil et al. [67]; K. Tsiane et al. [87]; E. E. Shread et al. [79]; C. Y. Tan et al. [86]).
Wide-area, space-based imaging offers a further advance by providing the depth, angular resolution, and point spread function (PSF) stability needed to resolve these systems into individual stars and construct clean, deep color–magnitude diagrams (CMDs). The Nancy Grace Roman Space Telescope (Roman), Euclid, and the Chinese Space Station Telescope (CSST) will each image large fractions of the sky with image quality that approaches what has only been available to date over the small fields of the Hubble Space Telescope (HST) and the James Webb Space Telescope (JWST) (e.g., R. Akeson et al. [1]; Euclid Collaboration et al. [29]; CSST Collaboration et al. [16]). This combination is ideally suited to discovering and characterizing low-luminosity systems around the MW and, more broadly, resolved stellar systems of all luminosities throughout the Local Volume ( Mpc).
Here, we demonstrate the potential of Euclid for the study of resolved stellar systems by focusing on the low-luminosity MW globular cluster (GC) AM-1. AM-1 is the sole known Galactic GC contained in the Euclid Quick Data Release 1 (Q1), which makes it the natural testbed for resolved star photometry with the publicly available data. First identified on photographic survey plates by B. F. Madore & H. C. Arp [59], AM-1 is among the most distant MW GCs ( kpc; e.g., N. Suntzeff et al. [85]; A. Dotter et al. [26]). This large distance, along with its sparse stellar population, has long made AM-1 a challenging target. We summarize the properties of AM-1 in Table 1.
| Quantity | Value |
| Observational | |
| R.A. (J2000) | 03:55:02.9 |
| Decl. (J2000) | −49:36:50 |
| Euclid obs. date | 2024 Sep 6 |
| Euclid exp. () | 1861 s |
| Euclid exp. (, , ) | 349 s each |
| WFPC2 obs. date | 1999 Jun |
| WFPC2 exp. (F555W, F814W) | 16,120 s / 19,120 s1 |
| limit (, , , ) | 27.1, 25.3, 25.5, 25.3 |
| 50% compl. (, , , ) | 26.6, 25.1, 25.1, 24.9 |
| Stars in Euclid catalog () | 41132 |
| CMD-selected structural candidates | 3063 |
| Physical (WFPC2, this work) | |
| 20.38 0.01 | |
| (kpc) | 119 1 |
| Age (Gyr) | 10.6 0.6 |
| 10.025 0.025 | |
| [Fe/H] | 4 |
| 0.0225 | |
| 6 | |
| () | |
| () | |
| () | |
| Structural (Euclid, this work) | |
| (arcsec) | |
| (pc) | |
| (arcsec) | |
| (pc) | |
| (arcsec) | |
| (pc) | |
| (King) | |
| (90%) |
Table 1. Observational and Structural Properties of AM-1
The most detailed view of AM-1 to date comes from deep HST/WFPC2 (Wide Field Planetary Camera 2) imaging in F555W and F814W by A. Dotter et al. [26] that reaches well below the main sequence turnoff (MSTO). They found that AM-1 is an old, metal-poor ([Fe/H] = ) cluster that features a prominent red horizontal branch (HB). In this respect AM-1 resembles other distant, outer halo GCs such as Pal 3, Pal 4, Pal 14, and Eridanus, which also have red HBs and comparable ages (P. B. Stetson et al. [84]; A. Dotter et al. [26]).
To enable studies of resolved stellar systems such as AM-1, we develop a Euclid-specific module for DOLPHOT7, a crowded field stellar PSF photometry package that is widely used for studies of nearby galaxies (e.g., A. E. Dolphin [23]; J. J. Dalcanton et al. [18]; D. J. Radburn-Smith et al. [72]; B. F. Williams et al. [93]; A. Savino et al. [77]). The module provides a robust, well-tested, and publicly available route to PSF photometry of Euclid VIS and NISP imaging within a framework that includes other telescope-specific modules for HST, JWST, and Roman data (e.g., J. J. Dalcanton et al. [19]; D. R. Weisz et al. [92]). The current Euclid module relies only on publicly released Q1 data products and requires no consortium-internal tools or PSF models. Early Euclid data, notably the Early Release Observations (ERO) of nearby galaxies (L. K. Hunt et al. [46]) and MW GCs (D. Massari et al. [61]; M. Griggio et al. [41]), are already delivering high-precision resolved star science, including empirical PSF modeling and astrometry of Galactic GCs (M. Libralato et al. [56, 57]), multiple population diagnostics in Euclid colors (A. P. Milone et al. [65]), and deep photometry of nearby galaxies to faint surface brightness (F. Annibali et al. [3]).
However, these analyses have largely relied on data products internal to the Euclid Consortium, i.e., individual calibrated frames rather than the public survey releases, and, in most cases, on specialized photometry codes developed within that community (e.g., KS2, euclid1pass). A main goal of our Euclid Guest Investigator Program (24-EGIP24-0005, PI Savino) is the public release of this DOLPHOT module for Euclid, which will enable the broader community to maximize the scientific return of Euclid starting with the first major data release (DR1).
This paper is organized as follows. In § 2 we summarize the Euclid VIS and NISP imaging of AM-1 and the supporting HST/WFPC2 dataset. In § 3 we detail our process of empirical PSF model construction for VIS and NISP. In § 4 we describe the Euclid DOLPHOT module, including new features of DOLPHOT v2.1, and our preprocessing of the VIS and NISP images. In § 5 we present our DOLPHOT photometry of AM-1. In § 6 we compare it with the Q1 MER catalog and ground-based near-infrared photometry, assess the stellar populations of AM-1 using Euclid and WFPC2 CMDs, and measure the cluster’s structural parameters. Finally, we summarize our results and describe the public release of the module in § 7.
Data
Euclid
Euclid (R. Laureijs et al. [54]) observed AM-1 with both the VIS (Euclid Collaboration et al. [30]) and NISP (Euclid Collaboration et al. [31]) in September 2024. We retrieved the data from the NASA/IPAC Infrared Science Archive (IRSA) using the Python Irsa interface, selecting the calibrated frames (for VIS, quadrant frames, i.e., the region read out by one of a CCD’s four amplifiers) of Observation ID 3590, which contains AM-1. The full set of exposures is listed in Table 2. Our analysis of AM-1 uses only the three VIS quadrants and single NISP detector that cover the cluster (Table 2), rather than the full field of view of Observation 3590. However, we build the PSF models (§ 3) from all detectors in all 124 Q1 observations.
| Image | Start Time | Detector | Filter | Exposure |
| 3590-0 | 03:35:27 | NISP DET21 | 87.2448 | |
| 3590-0 | 03:37:41 | NISP DET21 | 87.2448 | |
| 3590-0 | 03:39:55 | NISP DET21 | 87.2448 | |
| 3590-0-2 | 03:39:58 | VIS 5-2.E | 89.52 | |
| 3590-1-1 | 03:43:01 | VIS 4-2.H | 560.52 | |
| 3590-1 | 03:53:13 | NISP DET21 | 87.2448 | |
| 3590-1 | 03:55:27 | NISP DET21 | 87.2448 | |
| 3590-1 | 03:57:41 | NISP DET21 | 87.2448 | |
| 3590-1-2 | 03:57:44 | VIS 4-2.H | 89.52 | |
| 3590-2-1 | 04:00:47 | VIS 4-2.H | 560.52 | |
| 3590-2 | 04:10:59 | NISP DET21 | 87.2448 | |
| 3590-2 | 04:13:13 | NISP DET21 | 87.2448 | |
| 3590-2 | 04:15:27 | NISP DET21 | 87.2448 | |
| 3590-3-1 | 04:18:33 | VIS 4-2.E,H | 560.52 | |
| 3590-3 | 04:28:45 | NISP DET21 | 87.2448 | |
| 3590-3 | 04:30:59 | NISP DET21 | 87.2448 | |
| 3590-3 | 04:33:13 | NISP DET21 | 87.2448 |
Table 2. Euclid observations of AM-1, all taken September 6 2024.
The cluster was imaged at four dither pointings. Each NISP pointing consists of a single 87 s exposure in each of , , and . The nominal Euclid observing sequence acquires one 561 s VIS exposure at each of the four dithers, along with a 90 s VIS exposure at the first two (Euclid Collaboration et al. [32]), but the AM-1 VIS data comprise two 90 s and three 561 s exposures in . The deep exposure from the first dither is absent from Q1, and we have confirmed that it will also be absent from DR1, presumably because it did not pass the validation checks required for release (IRSA Helpdesk, private communication). The Euclid pipeline flags cosmic rays in the calibrated frames, and we therefore required no additional preprocessing before running DOLPHOT. Figure 1 shows a color image of AM-1 constructed from the VIS and NISP mosaics. The cluster is cleanly resolved into individual stars across its full extent.
Figure 1. Euclid color image of AM-1, constructed from the Q1 VIS and NISP mosaics ( blue, green, red). The field is on a side and is centered on the cluster. The cluster is resolved into individual stars across its full extent, illustrating the angular resolution and sensitivity that Euclid brings to bear on distant, sparse stellar systems.
Hubble Space Telescope
We complement the Euclid imaging with archival HST/WFPC2 observations of AM-1, which we use as an external check on our Euclid photometry (see § 6.3). The observations were obtained in June 1999 in F555W and F814W, comprising four short (180–230 s) and 14 long (900–1400 s) exposures in each band. We refer the reader to A. Dotter et al. [26] for full details.
We retrieved the WFPC2 images from the Mikulski Archive for Space Telescopes (MAST). The cluster falls on the WF3 chip, which was dithered with integer and half-pixel offsets. We therefore cleaned cosmic rays only within groups of images sharing a common filter and an integer-pixel shift. This yields four groups of 4–5 images per filter, which we find provides adequate cosmic-ray rejection prior to photometry. As a final step, we combine any deep images taken at a common dither position to speed up the photometry.
Empirical Point Spread Function Models
Accurate stellar photometry with DOLPHOT requires a library of model PSFs for each detector and filter. For example, PSFs for the DOLPHOT JWST module (D. R. Weisz et al. [92]) are provided by WebbPSF (now called STPSF) (e.g., M. D. Perrin et al. [70]). No comparable, public resource exists for Euclid. There is no tool to generate the models, and the PSF products available from the Euclid repository are not suitable for precision resolved star photometry (e.g., the publicly released VIS PSF is not oversampled; Euclid Collaboration et al. [33]). The released models are also not uniformly available: the Q1 PSF file is missing for 10 of the 124 observations, including the observation of AM-1.8
Empirical effective PSF (ePSF) models have been constructed for Euclid before, most notably by M. Libralato et al. [56], who, within the Euclid Consortium, built ePSFs for the VIS and NISP ERO imaging of MW GC NGC 6397. Our approach is similar in spirit, but aims to produce a general purpose, publicly released library covering the full focal plane in every band, built entirely from public data products.
For DOLPHOT, we construct the model Euclid PSFs empirically from the Q1 calibrated frames themselves. While building empirical models is straightforward in principle, doing so over the full focal plane is more involved in practice. Here, we describe our procedure.
We build the models on the calibrated single-exposure (LE1) frames rather than the stacked MER mosaics, so that the PSFs are sampled at the native pixel scale and in the same distorted frame in which DOLPHOT performs photometry. Each VIS exposure is read and stored per amplifier, such that a single pointing comprises 36 4 = 144 quadrant frames of 2k2k pixels. Each NISP exposure is taken through a single filter and comprises a 4 4 array of 16 detectors of 2k2k pixels. For consistency with our photometry, which treats each VIS amplifier independently (§ 4.2), we make no attempt to stitch the four quadrants into a single image. We therefore derive one PSF library per VIS quadrant and one per NISP detector and filter.
PSF Star Selection
We identify PSF stars by cross-matching each frame against Gaia DR3 (Gaia Collaboration et al. [37]). We adopt Gaia rather than an internal source finder because it provides a clean, externally vetted list of bona fide point sources, which is particularly valuable given the sparseness of the high-latitude Euclid Q1 fields.
For each of the 124 Euclid observations covered by the Q1 release, we compute the sky footprint from the WCS, perform a cone search over that region, and propagate the Gaia proper motions to the Euclid observation epoch before assigning each star pixel coordinates in every frame of that observation (i.e., 4 dithered exposures for NISP and 6 for VIS). We then retain only stars that are (i) isolated, with no Gaia neighbor inside a predefined PSF extraction region; (ii) consistent with single, well-behaved point sources, requiring non_single_star < 0.1, classprob_dsc_combmod_galaxy < 0.1, and ruwe < 1.2; and (iii) located more than half the PSF extraction region from the frame edge. For each surviving star, we extract an image cutout over the PSF extraction region, which we define as a square 55 × 55 pixel window. This is a deliberately generous window (5.5″ for VIS, 16.5″ for NISP) that captures the extended PSF wings while leaving an outer border for local background estimation. Over the same region we also extract the corresponding RMS array and data quality (DQ) extensions, and we associate the cutouts of a given Gaia star across the different dithered exposures using the photutils LinkedEPSFStar data class.
We use the DQ arrays to select the usable pixels within each cutout, retaining only pixels whose first DQ bit is zero (i.e., not flagged as INVALID). We further discard outright any cutout that contains saturated pixels (flagged by bit 3 for VIS and bit 10 for NISP) or in which more than 25% of the pixels are bad. Because we apply this rejection to each cutout individually, a star that saturates in the 561 s VIS exposures can still contribute through its 90 s exposures. In total, these cuts (bad pixels and/or saturation) exclude ∼20% of the VIS PSF star candidates entirely (a median of 918 per VIS quadrant, and 920–1,016 for the three quadrants covering AM-1). Because the Q1 footprint is sparse but vast, this selection still yields a large number of PSF stars: typically ∼3,500 per VIS quadrant (i.e., per ∼2k × 2k single-amplifier readout) and ∼12,000 per NISP detector.
Model Construction and Grid Packaging
We build the empirical PSF for each detector from its full set of cutouts using the iterative ePSF algorithm of J. Anderson & I. R. King [2], as implemented in photutils (L. Bradley et al. [9]). We adopt a mode estimator when evaluating the ePSF model residuals, rather than the median estimator implemented in photutils. This choice is more computationally costly, but more robust against contaminants, such as imaging artifacts, background galaxies, or faint stellar neighbors missing from the Gaia catalog. Because these contaminants only add flux, they skew the residual distribution to one side; the mode stays on its peak, i.e., the clean
stars, while the mean and median are pulled into the tail. Locating that peak reliably requires a large PSF-star sample, which the Q1 footprint provides.
Prior to the fit, we subtract a local sky from each cutout, estimating the background as the mode of the pixels in the outer border of the cutout via a Gaussian kernel-density estimate. We construct each model at an oversampling factor of 5 relative to the native pixels, spanning the full 55 × 55 pixel cutout, and normalize it over that entire region (i.e., a normalization radius of 55 pixels, equal to the cutout size) before iterating to convergence.
Distortion is negligible across a single cutout. Instead, it matters when we place each star’s cutouts from different dithers on a common sky coordinate system, which is mapped into each frame through that frame’s WCS (via LinkedEPSFStar; § 3.1). The model PSFs therefore inherit the accuracy of the pipeline astrometry. Unlike M. Libralato et al. [56], we do not solve jointly for distortion and PSF. Our tests in § 3.3 suggest that any resulting broadening is small, though the larger DR1 datasets are needed to confirm this more thoroughly.
Figure 2 and Figure 3 show the resulting models. The VIS models recover the extended wings and the six diffraction spikes from Euclid’s three-strut secondary mirror support, and a trefoil core with the same three-fold symmetry (Figure 2, inset). An azimuthal Fourier decomposition of the median-combined models over – is dominated by the harmonic, indicating that this structure reflects the telescope optics rather than the model construction.

Figure 2. An example of one of our empirical Euclid VIS PSFs for a representative detector quadrant (DET134E), constructed from the Q1 calibrated frames with the iterative ePSF algorithm (§ 3). We show the central node of the 3 × 3 model grid, oversampled by a factor of five relative to the native VIS pixel scale (0.1″), on a logarithmic stretch. The six diffraction spikes arise from Euclid’s three-strut secondary mirror support. The inset shows the central 0.9″ on a compressed logarithmic stretch, which reveals the three-lobed (trefoil) core.
Figure 3. Illustrative examples of our NISP PSF models and encircled energy (EE) curves. Left: Empirical NISP PSFs for all 16 NISP detectors, each the central node of its model grid, shown on a logarithmic stretch. The detector-to-detector variation of the PSF is modest but clearly present. Right: EE of the empirical VIS and NISP , , PSFs, normalized within the 55-pixel model cutout. The dashed and dotted vertical lines mark the NISP (1.13″) and VIS (1.3″, i.e., the 13-pixel aperture to which the VIS zero point is referenced) photometric apertures. The VIS PSF encloses 99% of its energy within 1.3″, while the NISP PSFs broaden with increasing wavelength.
The left panel of Figure 3 shows the corresponding models for all 16 NISP detectors. The detector-to-detector variation is modest but clearly present, which motivates our choice to derive an independent library per detector rather than adopting a single focal plane average.
We package the models on a spatial grid for each detector, matching the grid format used by the WebbPSF/STPSF libraries in the DOLPHOT JWST modules. At present we include only limited treatment of the PSF spatial variation across each chip. The NISP models are static, i.e., the single detector-averaged ePSF is replicated across all nine grid points. For VIS, we exploit the fact that adjacent quadrants must share the same PSF at their common boundary. At each grid point on a chip edge or corner shared with other quadrants, we define the ePSF as the average of the static ePSFs of all quadrants involved.
This is a coarse approximation of Euclid’s true spatial PSF variation. A fully spatially varying model is a natural next step, but a production-quality version will require the larger PSF star samples expected with DR1. We therefore defer it to future work. Using ePSF grids from Consortium-internal reductions of the NGC 6397 ERO imaging, M. Libralato et al. [56] find that the central pixel of the ePSF varies by a median of 1.2% within a VIS quadrant and by medians of 1.6–2.6% within a NISP detector. Our models capture little of this variation. However, solving for spatially variable residuals in DOLPHOT did not improve our AM-1 photometry (§ 4.1), suggesting that such variation is modest, or at least below what the current data can constrain.
Aperture Corrections and Performance
We determine aperture corrections from the same models, measuring the encircled energy between the PSF fitted magnitude and a standard aperture. For VIS, the zero point provided in the FITS header is calibrated to a 13-pixel circular aperture ([32]), so we measure aperture corrections to a circle of that radius. For the NISP bands, the zero points provided in the FITS headers appear to be calibrated to the Q1 PSFs ([34]), which are normalized to unity in a square box of . We therefore adopt a standard aperture of radius to match the area of that box.
As a quantitative check on the models, we compare their encircled energy (EE) with published Euclid values. The right panel of Figure 3 shows the EE curves, measured as the cumulative azimuthally summed flux about the flux centroid. We find that the VIS PSF encloses 99% of its energy within the (13-pixel) aperture to which the VIS zero point is referenced. For NISP, we measure half-light radii of and in , , , broadening with wavelength as expected. This trend is broadly consistent with the independent ERO analysis of F. Annibali et al. [3], who measure stacked-image FWHMs of in and – in , , and . Our NISP values sit comfortably inside the and specification of the released Q1 NISP PSF model ([34]). Our EE measurements within (–) are consistent with the values reported for both instruments by J.-C. Cuillandre et al. [17]. Our EE is normalized within the 55-pixel model cutout ( for VIS and for NISP), so it does not capture the faint, very extended wings that J.-C. Cuillandre et al. [17] track to . Our values are therefore slight overestimates. Normalizing to the full flux would lower the enclosed fraction at fixed radius by a few percent. This has little effect on and , which are set by the core.
Applying the static VIS library to test frames, we find PSF residuals that are small and comparable to those we obtain from the WebbPSF/STPSF models for JWST ([92]), and well within the range that DOLPHOT compensates for through its on-the-fly PSF adjustments.
The NISP models behave similarly, with one modest exception. Detector DET14 returns a slightly larger aperture correction than the other detectors. The size of this offset tracks the number of PSF stars used, not the presence of a few contaminating objects, suggesting that it reflects the statistics of the current PSF star sample rather than the detector itself. We expect the larger DR1 samples to resolve this issue. Regardless, the offset does not propagate into our photometry, because DOLPHOT’s aperture corrections measure the relationship between PSF fitting and aperture photometry within the standard aperture. Taken together, the EE comparison and the residual tests indicate that the empirical Euclid models are already accurate enough to support a first round of DOLPHOT photometry and validation.
The Euclid DOLPHOT Module
Overview
The overall structure of a DOLPHOT module, including pixel masking, astrometric alignment, star finding, and catalog creation, is common to all supported cameras. D. R. Weisz et al. [92] describe each of these steps in detail for the JWST modules. Here, we focus on what is new in version 2.1 and on the choices specific to Euclid.
All of our photometry uses DOLPHOT version 2.1, a recent upgrade to the widely used version 2.0 (A. Dolphin [22]). Several of the changes in this release were driven directly by Euclid.
The first is in the handling of PSF residuals. In version 2.0 and earlier, the PSF residual was assumed to be small, with the spatial variation of the PSF captured by the model itself. The Q1 data contain too few stars to measure a spatially variable PSF in this way, and version 2.1 therefore introduces supersampled (1x, 3x, or 5x), spatially variable PSF residuals.
Our final photometry does not use spatially variable PSF residuals. The PSF variability appears minimal, and adding extra parameters to the PSF solution (effectively, measuring nine PSF residuals per image instead of one) did not improve the photometry. We expect the spatially variable residuals to be more useful for data with many more well-measured stars per frame, such as the crowded ERO fields, the Galactic bulge imaging released with Q2, and the nearby galaxies expected with DR1, where the residual can be constrained at each grid point and the spatial variation of the Euclid PSF is better sampled. We will explore this issue in greater detail following DR1.
Version 2.1 introduces three new parameters, described fully in the DOLPHOT manual, that set the subpixel sampling, the spatial resolution, and the smoothing of the PSF residual. The first, PSFresSuper, sets the subpixel sampling of the residual image and must be 1, 3, or 5 (3 by default). The second, PSFresSpatial, sets the number of spatial nodes at which the residual is evaluated across an image, with the residual computed at the image corners and at evenly spaced points between them when it exceeds unity (1 by default, i.e., a single residual per image). The third, PSFresKernel, sets the smoothing kernel applied to the supersampled residual, either fit (a quadratic fit), smooth (a boxcar smooth, the default), or blend (a mix of the two). In this analysis, we adopt PSFresKernel=blend and leave PSFresSpatial at its default (Table 3), for the reasons given above.
| Parameter | Value | Parameter | Value |
| UseWCS | 2 | SigFind | 4.5 |
| Align | 2 | SigFinal | 5.0 |
| img_raper | 3 | PSFhotIt | 2 |
| img_rchi | 3 | FitSky | 2 |
| img_rsky | 15 35 | SkipSky | 2 |
| img_rsky2 | 4 7 | NoiseMult | 0.125 |
| img_apsky | 20 35 | PSFresKernel | blend |
| img_rpsf | 15 | CombineChi | 1 |
| MinSep | 1.5 | NISPvega | 0 |
Table 3. DOLPHOT parameters for AM-1. A detailed description of each parameter can be found in the DOLPHOT manual.
The second major change, and the one that drove the version number, is to the star finding. Earlier versions of DOLPHOT were prone to identifying artifacts in the subtracted wings of bright stars as spurious detections on the second and later finding passes. We traced this to the noise model, which did not account for the uncertainty in the PSF model at locations where stars had been subtracted. Such false detections could survive in the photometric solution and, although they were ultimately rejected by the data-quality cuts, could contaminate the PSF adjustments and aperture corrections, corrupting the photometry by up to a few percent. This behavior is now resolved, and we find that SecondPass values greater than one are rarely beneficial.
A third, more visible change is more robust aperture corrections: the first two values printed to the console for each image should now be close to zero when photometry has run nominally.
As with version 2.0, version 2.1 has an OpenMP-parallel sibling that returns identical results (versions 3.0 and 3.1, respectively).
Preprocessing
The DOLPHOT Euclid module provides separate preprocessing utilities for the VIS and NISP images. The two are similar enough that we describe them together here.
As with our PSF models, the module does not reassemble the four quadrants of a VIS detector into a single image, since doing so would force the use of a single average noise for the whole chip rather than the correct noise for each quadrant.
The readout noise can be obtained in two ways: estimated from the RMS extension in comparison with the science image, or read directly from the FITS header. For VIS we find that the two agree almost always, and we therefore adopt the header RDNOISE value by default. For NISP, the header does not carry enough information to recover the effective readout noise once the multiple reads per group and groups per integration have been processed, and neither the documentation nor the help desk could supply this detail. There, we default instead to estimating the readout noise from the good pixels in the RMS and science images. In both cases, the utility can be switched to the alternative method with a user flag.
We then mask bad and saturated pixels using the data quality definitions of Euclid Collaboration et al. [34] for NISP and Euclid Collaboration et al. [32] for VIS, and we use the SATURATE keyword to flag additional saturated pixels in the VIS images.
Finally, we compute and apply pixel area maps from the Projection Value keywords, and we derive third-order SIP coefficients for compatibility with the existing alignment routines in DOLPHOT, which support SIP but not TPV-style distortion. No camera supported by DOLPHOT prior to Euclid distributed its distortion solution in the TPV style. DOLPHOT uses the resulting SIP solution for the initial image registration and then fits residual offsets and a scale term for each image.
Photometry of AM-1
We ran DOLPHOT on all NISP detectors and VIS quadrants overlapping an 80″ × 80″ box centered on the cluster (see Table 2). We adopted the VIS mosaic of the same region as the alignment reference. We ran DOLPHOT with its default parameters except for those listed in Table 3. D. R. Weisz et al. [92] systematically explore how these parameters affect photometry.
Two of these choices follow directly from the changes described in § 4.1. We adopt NoiseMult = 0.125, well above the default of 0.05, because version 2.1 folds the contribution of bright neighbors’ subtraction residuals into the uncertainties through this parameter, which makes it both more important and more sensitive to tune than in earlier versions. We also set PSFhotIt = 2, which iterates the PSF photometry solution, and, in version 2.1, substantially reduces the bias in the recovery of faint stars.
To ensure that only bona fide stars informed the PSF measurements and aperture corrections, we used an initial photometry pass to select high signal-to-noise ( in every band and overall), non-extended sources (). A second DOLPHOT run then used these stars for the PSF and aperture corrections to produce our final photometry.
The DOLPHOT catalog adopts the FITS keyword MAGZEROP for the VIS zero point, and either ZPAB or ZPVEGA for NISP, depending on the NISPvega setting. We set NISPvega to zero (Table 3), so that our photometry is on the AB system.
We applied only signal-to-noise () and sharpness () cuts to construct our final catalog. At the depth of the Q1 data, particularly the 87 s NISP exposures, any additional cuts substantially reduced the completeness on the main sequence. AM-1 lies in the Euclid Deep Field South, which will be observed times over the mission (Euclid Collaboration et al. [29]), compared with the single visit in Q1. These deeper VIS and NISP data should support more stringent cuts (e.g., to better separate stars from compact background galaxies, as in J. T. Warfield et al. [90]). D. R. Weisz et al. [92] discuss the trade-offs among the DOLPHOT quality cuts in more detail. Figure 4 shows the resulting and CMDs.
Figure 4. Euclid CMDs of AM-1 from our DOLPHOT photometry, for all stars within a projected radius of 60″ of the cluster center, which is larger than the ∼ 1′ square field shown in Figure 1. Left: versus . Right: versus . Error bars show the representative photometric uncertainties measured from our artificial star tests (ASTs). The main sequence (MS) is well populated to roughly a magnitude below the MSTO, and the red giant branch (RGB) and red HB are clearly resolved.
To characterize the completeness and photometric uncertainties, we ran ASTs. As always in DOLPHOT, the stars are added one at a time, and the photometry is performed in a small region around each inserted star and compared against the original catalog to determine whether the star is recovered, or whether it is blended with a nearby neighboring star. We apply the same and sharpness cuts to the recovered stars when computing the completeness. We refer the reader to D. R. Weisz et al. [92] for a fuller description of DOLPHOT artificial star tests and their use in characterizing completeness and photometric bias. We show the per filter completeness curves for AM-1 in Figure 5.
Figure 5. Completeness as a function of input magnitude in each of the four Euclid filters, measured from artificial stars inserted into the images and recovered with the same SNR and sharpness cuts applied to our science catalog. The orange dotted line marks 50% completeness. We reach 50% completeness at , roughly 1.5 mag deeper than in the NIR bands (, , ). We do not probe to the completeness limit in , as it is not necessary for the current analysis.
Discussion
Comparison with Euclid Q1 Catalog
As a first test of our photometry, we compared it to the Euclid Q1 photometric catalog of Euclid Collaboration et al. [35]. We accessed the catalog through the IRSA interface,9 selecting point-like objects (point_like_flag = 1 or point_like_prob > 0.7) within a 0.1 0.1 degree box centered on the cluster.
We construct aperture (APHOT) and template-fitting (TPHOT) magnitudes following the MER Photometry Cookbook,10 adopting the 2FWHM aperture (i.e., a circular aperture of diameter twice the FWHM of the worst-resolution band, measured on PSF-matched images). The reference photometric measurement in the MER catalog is FLUX_DETECTION_TOTAL, the Kron flux (R. G. Kron [51]) in the detection mosaic. For the VIS-detected sources considered here (VIS_DET = 1) this is the total flux,
The MER catalog also provides a FLUX_VIS_PSF column, but it is not used in the Cookbook’s prescriptions for total fluxes, nor was its use recommended by the Euclid helpdesk. For the NISP bands , we obtain total fluxes by scaling the detection flux by a color term measured between that band and VIS. For aperture photometry,
and for template-fitting photometry, where the color term is measured against the VIS image smoothed to the resolution of band ,
All MER fluxes are in Jy, which we convert to AB magnitudes .
In Appendix A.1, we show this conversion as a function of DOLPHOT magnitude. As noted by Euclid Collaboration et al. [35], the APHOT shows less scatter at bright magnitudes, while the TPHOT is tighter at the faint end. We find generally good agreement between our photometry and the MER catalog in all bands. In VIS, the offset (DOLPHOT minus MER) is mag, after we remove sources fainter than and reject outliers. In NISP we compared against both flavors of MER photometry. The APHOT offsets are , , and mag. For the TPHOT comparison we did not impose the cut, since PSF fitting delivers reliable photometry to fainter magnitudes, and there we measured , , and mag.
The more consequential difference between the two catalogs is not their zero points but their depth. Figure 6 compares the CMDs built from our photometry and from the MER catalog over the same 60″-radius field. Our catalog contains a factor of more sources and populates the MS to roughly a magnitude below the MSTO, where the MER catalog is essentially empty. AM-1 is relatively uncrowded and we expect even greater contrast in more crowded regions, such as the ERO imaging, the Galactic bulge released as part of Q2, and new regions released with DR1.
Figure 6. A comparison of Euclid CMDs of AM-1 from our DOLPHOT photometry (left) and from the Euclid Q1 MER catalog (right), over the same -radius field and on identical axes. MER photometry is APHOT in all bands as TPHOT is not available for VIS. Our catalog contains a factor of more sources over the same area and extends to roughly a magnitude below the MSTO, whereas the MER catalog is essentially empty in the same region.
Comparison with the VISTA Hemisphere Survey (VHS)
To assess the accuracy of our absolute photometric calibration, we compared our photometry against the VISTA Hemisphere Survey (VHS; R. G. McMahon et al. [63]), which provides an external reference independent of Euclid. The VHS uses the Two Micron All Sky Survey (2MASS; M. F. Skrutskie et al. [82]) as its primary astrometric and baseline photometric calibrator: hundreds of 2MASS point sources act as standard stars in every VISTA Infrared Camera (VIRCAM) exposure, enabling a robust, continuous calibration of the broadband filters. We obtained VHS data within 10′ of AM-1 from the ESO Archive Science Portal; only and are available in this region. We converted the VHS Vega magnitudes to the AB system following C. González-Fernández et al. [40], and then to the Euclid photometric system using the Appendix C conversions for normal stars of Euclid Collaboration et al. [28]. We show this comparison in Appendix § 10.§ 1. We find excellent agreement in both bands: the mean offset (DOLPHOT minus VHS) is mag and mag. This agreement provides an external validation of our NISP photometry.
Analysis of the AM-1 Stellar Population
As a final check on this first application of the DOLPHOT Euclid module, we compared our photometry against results from archival WFPC2 data CMD modeling. Using the WFPC2 imaging described in § 2.2, we ran DOLPHOT exactly as for the Euclid data, except that we
set UseWCS=1 to take advantage of the higher-order distortion corrections built into DOLPHOT.
Our aim here is to validate the Euclid photometry, not to redetermine the properties of AM-1, which are already analyzed in detail by A. Dotter et al. [26]. We therefore derive the stellar-population parameters from the well-characterized WFPC2 CMD and use them to test the Euclid CMDs. For a system newly discovered in Euclid imaging, with no existing HST data, one would instead fit the Euclid CMDs directly to determine the age, metallicity, and distance.
We applied the same goodness-of-fit cuts as for the Euclid data and used the CMD-fitting package MATCH (A. E. Dolphin [24]) to determine the best-fitting distance and stellar population, adopting from the extinction maps of E. F. Schlafly & D. P. Finkbeiner [78]. With the BaSTI (a Bag of Stellar Tracks and Isochrones) evolutionary models (S. L. Hidalgo et al. dalgo et al. 2018),11 we find a best fit for a pseudo-simple stellar population (SSP) with –, to , and a distance modulus . The isochrone plotted in Figure 7 corresponds to the best fit for this pseudo-SSP.
Figure 7. WFPC2 color–magnitude diagram of AM-1, with the best-fitting isochrone overlaid: BaSTI18 in purple (left) and PARSEC in orange (right). We adopt (10.6 Gyr), ( kpc), and , with for BaSTI18 and for PARSEC. Both isochrones provide a good match to the MS, MSTO, and red giant branch.
We also repeated the fit with the PARSEC (PAdova and TRieste Stellar Evolution Code) models (A. Bressan et al. [10]), for which we compute Euclid-system magnitudes using the YBC database of bolometric corrections (Y. Chen et al. [15]), recovering the same age and distance and a slightly lower metallicity, to . We show both model sets throughout. Be-
cause MATCH evaluates the models on a discrete grid, these ranges reflect the grid spacing (0.05 in and 0.1 dex in ) rather than a measured width of the population. Combining the two model sets, we adopt and , where the metallicity uncertainty is the quadrature sum of the 0.05 dex half-resolution and the 0.05 dex systematic offset between the BaSTI18 and PARSEC central values.
Our WFPC2 fit reproduces the main features of AM-1’s stellar population identified by A. Dotter et al. [26]. We recover an age ( Gyr) and a distance (, kpc) in good agreement with their determination, and, like them, we find an old popula-
tion whose CMD is well described by a single isochrone and features a prominent red HB. The one parameter for which our fit differs meaningfully is the metallicity, which we discuss next.
Our best-fit metallicity, , is more metal-rich than the that A. Dotter et al. [26] derive for AM-1 from -enhanced isochrone fits (). An older spectroscopic estimate, from N. Suntzeff et al. [85], is broadly consistent with the A. Dotter et al. [26] value but is based on only two red giants and is not a strong constraint. However, part of the offset from A. Dotter et al. [26] is a matter of composition, as we use scaled-solar models. Converting the A. Dotter et al. [26] metallicity onto the scaled-solar value using the relation from M. Salaris et al. [76] yields , in better agreement with our value of . The remaining 0.4 dex difference is modest and generally within the range of metallicity differences in CMD fitting owing to different adopted physics and bolometric corrections in various stellar libraries (e.g., [38, 25, 91, 81]). Spectroscopy is needed to pin down AM-1’s metallicity and detailed chemical composition.
We then overplot our best-fitting WFPC2 isochrone, in the Euclid photometric system, on the Euclid CMDs in Figure 8 and Figure 9. The isochrones match both the , and , photometry reasonably well. The independent VHS comparison (§ 6.2), not the isochrone match, is our primary test of the , absolute calibration. The small residual between the models and the observed upper RGB is comparable to that found in other model comparisons with optical and near-IR CMDs (e.g., [20, 81, 64]).
Figure 8. Same as Figure 4, but with the BaSTI18 isochrone overlaid. We adopt the parameters given in Figure 7. The isochrone provides a reasonable match to both CMDs. However, on the CMD it is systematically offset by mag from the RGB, red HB, and MSTO. We discuss this offset in § 6.3 and Appendix § 10.§ 2.
Figure 9. Same as Figure 8, but for the PARSEC models. The same offset is present, indicating that it does not depend on the choice of model set.
However, in the CMD, the isochrones are systematically offset from the data, most apparently in the color of the RGB. For and we can appeal to VHS (§ 6.2), but no comparable external reference is available for and . VHS covers only and in this field, and the Q1 MER catalog shares our Euclid zero points. We therefore use the WFPC2 photometry as an internal reference, predicting each star’s and from its WFPC2 color through quadratic transformations fit to the best-fitting isochrone. This is not a general purpose color transformation, but it should be adequate for this application and for this specific population.
Adopting , the quadratic fits to the isochrone colors are
and
The equivalent fits using our best-fitting PARSEC isochrone are
and
For each star we then predict and from its measured F555W and F814W photometry, and compare those predictions with the Euclid magnitudes we measure with DOLPHOT. Because each prediction is referenced to the star’s own measured , the comparison is independent of distance, and it tests the two Euclid bands separately. Although extinction between the bands does not cancel, at the low reddening toward AM-1 () it amounts to no more than a few mmag. As we show in Appendix § 10.§ 2, this comparison isolates the offset to .
Structural Properties
The depth and areal coverage of the Euclid imaging also allow us to characterize the spatial structure of AM-1. R. R. Muñoz et al. [66] previously measured the structure of AM-1, and comparing with their result further validates our photometry.
For the structural fit, we use only our Euclid DOLPHOT catalog, not the deeper WFPC2 imaging, which samples only the central few half-light radii. The Euclid field covers the full radial extent of AM-1 implied by the R. R. Muñoz et al. [66] structural parameters.
We model the on-sky distribution of the member stars as an inhomogeneous Poisson point process and evaluate its unbinned likelihood, following S. E. Koposov et al. [50] and D. R. Weisz et al. [91], to which we refer the reader for full details.
Briefly, the method fits the positions of the individual stars, rather than a binned surface density profile, modeling the cluster and a spatially uniform foreground/background component in a single likelihood. Fitting unbinned positions avoids the loss of spatial information and the sensitivity to bin placement inherent to binned fits, an advantage that is particularly valuable for a sparse system such as AM-1.
We restrict the fit to a clean sample of 306 candidate members, selected in the range and within 100″ of the cluster center (Table 1), the region over which the VIS and NISP coverage overlap completely. Beyond this radius the detector footprints only partially fill the field, so the fit’s assumption of a uniform background over the full circular region would not hold. This selection isolates well-measured MS and RGB stars while limiting potential contamination from the noisy faint end. We fit three radial profiles commonly used for star clusters and low-luminosity systems: an exponential, a Plummer profile (H. C. Plummer [71]), and a single-mass King model (I. R. King [49]). Each model shares a common centroid, half-light radius (), ellipticity, and position angle, together with a constant background surface density. We sample the posterior with the affine-invariant ensemble sampler emcee (D. Foreman-Mackey et al. [36]) under uniform priors, so that the posterior is proportional to the likelihood, and quote posterior medians with 16th–84th percentile intervals.
We find that AM-1 is well-described by all three profiles, with parameters that closely agree where they are directly comparable. The exponential and Plummer fits return nearly identical circularized half-light radii, and respectively ( pc at kpc), while the more extended wings of the King model drive a modestly larger (15.7 pc), as expected for that profile. The King fit returns a low central concentration, and .
We find no evidence that AM-1 is elongated. The ellipticity is consistent with zero ( at 90% confidence). We therefore fix the ellipticity to zero and do not report a position angle.
Our measurements are consistent with the ground-based structural parameters of R. R. Muñoz et al. [66]. They report an effective radius of pc, in good agreement with our King value (15.7 pc) and modestly larger than our exponential and Plummer values ( pc), as expected given the differing assumed profiles. R. R. Muñoz et al. [66] also find a round, regular morphology, consistent with the circular shape we infer.
We combine these measurements with the stellar population fit of § 6.3 to estimate the present-day stellar mass and mass-to-light ratio of AM-1. The MATCH SSP fit returns the total mass formed, which we convert to a standard Kroupa (P. Kroupa [52]) mass range (–) and reduce by the mass lost to stellar evolution over Gyr, retaining of the birth mass as living stars and compact remnants (e.g., S. N. Leitner & A. V. Kravtsov [55]; F. Vincenzo et al. [88]). This yields a present-day mass of . Combined with the luminosity implied by our adopted (; Table 1), we obtain a -band mass-to-light ratio . This is consistent with expectations for an old GC (–).
We have not searched for tidal tails around AM-1, as such a search requires uniform photometry far outside the detectors analyzed here and is beyond the scope of this paper. A tidal-tail search is a natural application of the DOLPHOT Euclid module to the surrounding Q1 imaging and to future releases.
AM-1 in the Context of the Outer-Halo Globular Clusters
AM-1 appears to belong to a small population of distant, diffuse, low-luminosity GCs in the outer MW halo. Its size, luminosity, and distance are comparable to other well-studied clusters of this class such as Pal 3, Pal 4, Pal 14, and Eridanus (e.g., G. L. H. Harris [44]; R. R. Muñoz et al. [66]). Crater is similarly extended ( pc) and even more remote ( kpc; D. R. Weisz et al. [91]). These GCs have ages between and Gyr, except Crater, which is – Gyr old (e.g., P. B. Stetson et al. [84]; A. Dotter et al. [26]; D. R. Weisz et al. [91]).
A. D. Mackey & S. van den Bergh [58] identify these objects as “young halo” clusters and single out AM-1 as the most distant such cluster, a distinction later superseded by the discovery of Crater. AM-1’s dynamics point to an accreted origin. Using Gaia orbits, D. Massari et al. [60] place AM-1 in the high-energy group of clusters that appear to have been accreted from low-mass progenitors that cannot (yet) be individually identified. The chemodynamical analysis of T. M. Callingham et al. [12] classifies AM-1 as accreted but unassociated with any known accretion event, grouping it with Crater, Pal 3, Pal 4, Pal 14, and Eridanus.
More broadly, our study of AM-1 illustrates the promise of Euclid for the study of faint stellar systems throughout the MW halo. Euclid unites two capabilities that have rarely been available in a single dataset: the angular resolution and stable PSF needed to resolve individual stars well below the MSTO, and the wide-area coverage needed to survey large volumes of the halo uniformly. Resolved star studies of systems like AM-1 have traditionally relied on targeted HST or JWST imaging of one object at a time, whereas wide ground-based surveys supply the area but not the angular resolution to cleanly separate members from contaminants.
The Euclid Wide Survey footprint will encompass known Galactic GCs, including distant outer-halo clusters such as Pal 4 and NGC 2419, a few dozen UFDs, and nearly two hundred galaxies within 5 Mpc (e.g., [68]). DR1 should include several southern UFDs (e.g., the Grus and Tucana systems) and dozens of nearby galaxies such as those of the Sculptor group. Deep ground-based imaging is also advancing rapidly. Rubin Observatory commissioning data have already quadrupled the sample of candidate member stars in the ultra-faint satellite Virgo III ([69]), while Euclid adds the space-based angular resolution and near-infrared coverage needed to push further into the crowded, faint regime. We anticipate that the public DOLPHOT Euclid module will support uniform resolved star studies of MW halo satellites and outer halo GCs as the survey grows.
Summary
We have presented a Euclid study of AM-1, the only Galactic GC in the Euclid Q1 release and one of the most distant, diffuse GCs known in the MW. To measure its resolved stars, we developed a Euclid-specific module for DOLPHOT that builds empirical PSF models directly from the public Q1 calibrated frames. Our main results are:
Our DOLPHOT photometry recovers the MS of AM-1 to roughly a magnitude below its MSTO, with point-source depths of , , , and (Table 1), – mag deeper than the public Q1 data set (26.0, 23.8, 24.0, 24.0 in , , , ; [35]).
Our WFPC2 and Euclid CMDs are consistent with an old, intermediate-metallicity stellar population, with an age in agreement with the earlier analysis of A. Dotter et al. [26]. The best-fitting isochrone reproduces both Euclid CMDs reasonably well, but is systematically offset from the observed sequence. We trace this mag offset to the band, although we cannot determine whether it originates in the zero point or in a common error shared by the available bolometric corrections.
Our photometry agrees with the public MER catalog to mag in each available band, where the catalogs overlap. Our NISP photometry agrees with the independently calibrated VHS catalog to
and mag, providing an external check on our absolute calibration.
From our Euclid photometry, we measure –16 pc for AM-1, which places it among the most physically extended GCs known in the MW. The cluster has a low central concentration and shows no evidence for elongation. These parameters are consistent with the previous ground-based measurements of R. R. Muñoz et al. [66]. AM-1’s size, luminosity, and orbit place it among the distant outer halo clusters that dynamical studies identify as accreted ([60, 12]).
We present DOLPHOT v2.1, which includes several Euclid-motivated updates that apply to every camera DOLPHOT supports. These updates include a revised noise model that prevents residuals in the subtracted wings of bright stars from producing spurious detections. Version 2.1 also introduces supersampled, spatially variable PSF residuals, and more robust aperture corrections. OpenMP-parallel builds (v3.0 and v3.1) return results identical to v2.0 and v2.1.
Taken together, these results demonstrate that, even in the absence of a full public release of detailed calibration products, Euclid can deliver precise, deep, resolved star photometry of low-luminosity stellar systems in the MW halo, well beyond what the public catalogs currently provide. As part of our Euclid Guest Investigator Program (24-EGIP24-0005), we are publicly releasing a first version of the DOLPHOT Euclid module so that the broader community can carry out similar analyses using DR1 data.
Acknowledgments
We thank the referee, Luigi Bedin, for a thoughtful and constructive referee report that improved the paper. This material is based upon work supported by the National Aeronautics and Space Administration through the Euclid Guest Investigator Program (ROSES-2024 D.18) under award No. 80NSSC25K7734 (proposal 24-EGIP24-0005; PI A. Savino).
This research has made use of the NASA/IPAC Infrared Science Archive, which is funded by the National Aeronautics and Space Administration and operated by the California Institute of Technology. We thank the IRSA Helpdesk for their assistance with the Euclid Q1 data retrieval and for clarifying the availability of the VIS exposures of AM-1.
Based on observations made with the NASA/ESA Hubble Space Telescope, obtained from the Mikulski Archive for Space Telescopes (MAST) at the Space Telescope Science Institute (STScI). STScI is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-03127.
This research has made use of the SVO Filter Profile Service “Carlos Rodrigo”, funded by MCIN/AEI/10.13039/501100011033/ through grant PID2023-146210NB-I00.
This paper was constructed and written by the authors, but portions were edited with the assistance of Claude Opus v4.8 and v5.0, large language models developed by Anthropic, to improve readability in June 2026. Claude Opus v5.5 was used for a final proofread (i.e., to catch typos, improve clarity, check for redundancy and repetition) in October 2026. All content was reviewed and verified by the authors.
Data Availability
All of the imaging analyzed in this paper is public. The Euclid VIS and NISP observations of AM-1 are part of Euclid Quick Data Release 1 and were retrieved from the NASA/IPAC Infrared Science Archive (Observation ID 3590). The archival HST/WFPC2 observations (GO-6512) were retrieved from the Mikulski Archive for Space Telescopes. The VISTA Hemisphere Survey photometry was obtained from the ESO Archive Science Portal, and the Gaia DR3 astrometry used for PSF star selection from the ESA Gaia Archive.
The DOLPHOT source code, the Euclid module, and the empirical PSF libraries described in § 3 are distributed from the DOLPHOT website (http://americano.dolphinsim.com/dolphot/). The DOLPHOT photometric catalog of AM-1 will be made available upon request.
Facilities: Euclid (VIS, NISP), HST (WFPC2), VISTA (VIRCAM), Gaia
Software: DOLPHOT (A. Dolphin [22]), MATCH (A. E. Dolphin [24]), emcee (D. Foreman-Mackey et al. [36]), photutils (L. Bradley et al. [9]), Astropy (Astropy Collaboration et al. [4, 5, 6]), NumPy (C. R. Harris et al. [43]), SciPy (P. Virtanen et al. [89]), Matplotlib (J. D. Hunter [47]), astroquery (A. Ginsburg et al. [39]).
Photometric Comparisons and Residuals
Catalog Comparison Residuals
Figure 10 and Figure 11 show the photometric comparisons between DOLPHOT and external catalogs discussed in §§ § 6.1 and § 6.2. All offsets quoted, in both the main text and these figures, are uncertainty-weighted means measured after outlier rejection.
Figure 10. Difference between our DOLPHOT photometry and the Euclid Q1 MER catalog as a function of DOLPHOT magnitude, for the MER aperture photometry (APHOT, left) and template-fitting photometry (TPHOT, right). The orange line marks the offset given in § 6.1, listed in the lower right of each panel. We find generally good agreement, with all offsets mag. The MER catalog provides no template-fitting measurement in VIS, and so the corresponding panel is left blank. We caution that the APHOT residuals show a pronounced one-sided tail toward brighter DOLPHOT magnitudes, which grows toward longer wavelengths and is consistent with a fixed aperture absorbing flux from neighboring stars as the PSF broadens; the quoted offsets are therefore measured after rejecting outliers.
Figure 11. Difference between our DOLPHOT photometry and the VHS catalog, converted to the Euclid photometric system, as a function of DOLPHOT magnitude. We restrict the comparison to , where VHS is complete. The orange line marks the mean offset. We find excellent agreement in both bands (, ). Because VHS is calibrated independently of Euclid, this comparison provides an external check on our photometry.
Bolometric Correction Residuals
We provide more detail on which of the two Euclid bands is responsible for the color offset seen in Figure 8 and Figure 9. For each star, we predict its Euclid magnitude from its WFPC2 , photometry through each set of bolometric corrections, and compare with the measured value down to the 90% completeness limit of the Euclid data (, ). Figure 12 shows the resulting residuals as a function of magnitude for each set of bolometric corrections.
Figure 12. Measured Euclid photometry minus that predicted from the WFPC2 , photometry through the bolometric corrections, as a function of magnitude, for BaSTI18 (left) and PARSEC (right). Because cancels from both sides, this comparison is independent of distance and, at the low reddening toward AM-1, effectively unaffected by extinction. Colored lines show the running offset, and the dotted line and label give the offset over the full range, measured to each band’s 90% completeness limit. The measured falls between the two sets of predictions ( and ). In contrast, the measured is brighter than predicted by both sets of corrections ( and ). Taken at face value, this suggests that the zero point, or the system response adopted in the bolometric corrections, is in error.
The measured falls between the two isochrone sets ( with BaSTI18, with PARSEC), and for either set the offset is smaller than the mag disagreement between the model-predicted Euclid magnitudes themselves. The offset is therefore unlikely to be the cause of the observed CMD offset. For , in contrast, both model sets predict magnitudes 0.06–0.10 mag fainter than we measure. The residuals are flat with magnitude and color, i.e., the offset appears to be constant, which is consistent with an error in the zero point or in the system response shared by the Euclid bolometric corrections adopted by the BaSTI and PARSEC models. Filter width alone is unlikely to explain this offset. is the broadest Euclid filter in fractional width (FWHM/), yet it agrees with the models, whereas (FWHM/) is comparable to the JWST/NIRCam wide filters [74, 73, 75]. The publicly available NISP passbands are pre-launch determinations ([28]), so in-flight characterization of the response may help resolve this offset.
References
- [1]Akeson, R., Armus, L., Bachelet, E., et al. 2019, arXiv e-prints, arXiv:1902.05569, doi: 10.48550/arXiv.1902.05569DOI
- [2]Anderson, J., & King, I. R. 2000, PASP, 112, 1360, doi: 10.1086/316632DOI
- [3]Annibali, F., Ferguson, A. M. N., Sanchez-Alarcon, P. M., et al. 2026, A&A, 712, A175, doi: 10.1051/0004-6361/202658966DOI
- [4]Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al. 2013, A&A, 558, A33, doi: 10.1051/0004-6361/201322068
- [5]Astropy Collaboration, Price-Whelan, A. M., Sipőcz, B. M., et al. 2018, AJ, 156, 123, doi: 10.3847/1538-3881/aabc4f
- [6]Astropy Collaboration, Price-Whelan, A. M., Lim, P. L., et al. 2022, ApJ, 935, 167, doi: 10.3847/1538-4357/ac7c74DOI
- [7]Bechtol, K., Drlica-Wagner, A., Balbinot, E., et al. 2015, ApJ, 807, 50, doi: 10.1088/0004-637X/807/1/50DOI
- [8]Belokurov, V., Zucker, D. B., Evans, N. W., et al. 2007, ApJ, 654, 897, doi: 10.1086/509718DOI
- [9]Bradley, L., Sipocz, B., Robitaille, T., et al. 2016, Photutils: Photometry tools,, Astrophysics Source Code Library, record ascl:1609.011 http://ascl.net/1609.011
- [10]Bressan, A., Marigo, P., Girardi, L., et al. 2012, MNRAS, 427, 127, doi: 10.1111/j.1365-2966.2012.21948.xDOI
- [11]Bullock, J. S., & Boylan-Kolchin, M. 2017, ARA&A, 55, 343, doi: 10.1146/annurev-astro-091916-055313DOI
- [12]Callingham, T. M., Cautun, M., Deason, A. J., et al. 2022, MNRAS, 513, 4107, doi: 10.1093/mnras/stac1145DOI
- [13]Cerny, W., Martínez-Vázquez, C. E., Drlica-Wagner, A., et al. 2023, ApJ, 953, 1, doi: 10.3847/1538-4357/acdd78DOI
- [14]Chambers, K. C., Magnier, E. A., Metcalfe, N., et al. 2016, arXiv e-prints, arXiv:1612.05560, doi: 10.48550/arXiv.1612.05560DOI
- [15]Chen, Y., Girardi, L., Fu, X., et al. 2019, A&A, 632, A105, doi: 10.1051/0004-6361/201936612DOI
- [16]CSST Collaboration, Gong, Y., Miao, H., et al. 2026, Science China Physics, Mechanics, and Astronomy, 69, 239501, doi: 10.1007/s11433-025-2809-0
- [17]Cuillandre, J.-C., Bertin, E., Bolzonella, M., et al. 2025, A&A, 697, A6, doi: 10.1051/0004-6361/202450803DOI
- [18]Dalcanton, J. J., Williams, B. F., Seth, A. C., et al. 2009, ApJS, 183, 67, doi: 10.1088/0067-0049/183/1/67DOI
- [19]Dalcanton, J. J., Williams, B. F., Lang, D., et al. 2012a, ApJS, 200, 18, doi: 10.1088/0067-0049/200/2/18DOI
- [20]Dalcanton, J. J., Williams, B. F., Melbourne, J. L., et al. 2012b, ApJS, 198, 6, doi: 10.1088/0067-0049/198/1/6DOI
- [21]Dark Energy Survey Collaboration, Abbott, T., Abdalla, F. B., et al. 2016, MNRAS, 460, 1270, doi: 10.1093/mnras/stw641DOI
- [22]Dolphin, A. 2016, DOLPHOT: Stellar photometry,, Astrophysics Source Code Library, record ascl:1608.013 http://ascl.net/1608.013
- [23]Dolphin, A. E. 2000, PASP, 112, 1383, doi: 10.1086/316630DOI
- [24]Dolphin, A. E. 2002, MNRAS, 332, 91, doi: 10.1046/j.1365-8711.2002.05271.xDOI
- [25]Dolphin, A. E. 2012, ApJ, 751, 60, doi: 10.1088/0004-637X/751/1/60DOI
- [26]Dotter, A., Sarajedini, A., & Yang, S.-C. 2008, AJ, 136, 1407, doi: 10.1088/0004-6256/136/4/1407DOI
- [27]Drlica-Wagner, A., Bechtol, K., Rykoff, E. S., et al. 2015, ApJ, 813, 109, doi: 10.1088/0004-637X/813/2/109
- [28]Euclid Collaboration, Schirmer, M., Jahnke, K., et al. 2022, A&A, 662, A92, doi: 10.1051/0004-6361/202142897
- [29]Euclid Collaboration, Mellier, Y., Abdurro’uf, et al. 2025a, A&A, 697, A1, doi: 10.1051/0004-6361/202450810
- [30]Euclid Collaboration, Cropper, M. S., Al-Bahlawan, A., et al. 2025b, A&A, 697, A2, doi: 10.1051/0004-6361/202450996
- [31]Euclid Collaboration, Jahnke, K., Gillard, W., et al. 2025c, A&A, 697, A3, doi: 10.1051/0004-6361/202450786
- [32]Euclid Collaboration, McCracken, H. J., Benson, K., et al. 2026a, A&A, 711, A2, doi: 10.1051/0004-6361/202554594
- [33]Euclid Collaboration, Aussel, H., Tereno, I., et al. 2026b, A&A, 711, A1, doi: 10.1051/0004-6361/202554610
- [34]Euclid Collaboration, Polenta, G., Frailis, M., et al. 2026c, A&A, 711, A3, doi: 10.1051/0004-6361/202554657
- [35]Euclid Collaboration, Romelli, E., Kümmel, M., et al. 2026d, A&A, 711, A4, doi: 10.1051/0004-6361/202554586
- [36]Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125, 306, doi: 10.1086/670067
- [37]Gaia Collaboration, Vallenari, A., Brown, A. G. A., et al. 2023, A&A, 674, A1, doi: 10.1051/0004-6361/202243940
- [38]Gallart, C., Zoccali, M., & Aparicio, A. 2005, ARA&A, 43, 387, doi: 10.1146/annurev.astro.43.072103.150608DOI
- [39]Ginsburg, A., Sipőcz, B. M., Brasseur, C. E., et al. 2019, AJ, 157, 98, doi: 10.3847/1538-3881/aafc33DOI
- [40]González-Fernández, C., Hodgkin, S. T., Irwin, M. J., et al. 2018, MNRAS, 474, 5459, doi: 10.1093/mnras/stx3073
- [41]Griggio, M., Libralato, M., Gerasimov, R., et al. 2026, A&A, 711, A88, doi: 10.1051/0004-6361/202660441DOI
- [42]Gwyn, S., McConnachie, A. W., Cuillandre, J.-C., et al. 2025, AJ, 170, 324, doi: 10.3847/1538-3881/ae03abDOI
- [43]Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357, doi: 10.1038/s41586-020-2649-2DOI
- [44]Harris, G. L. H. 2010, PASA, 27, 475, doi: 10.1071/AS09063DOI
- [46]Hunt, L. K., Annibali, F., Cuillandre, J. C., et al. 2025, A&A, 697, A9, doi: 10.1051/0004-6361/202450781DOI
- [47]Hunter, J. D. 2007, Computing in Science and Engineering, 9, 90, doi: 10.1109/MCSE.2007.55DOI
- [48]Ivezić, Ž., Kahn, S. M., Tyson, J. A., et al. 2019, ApJ, 873, 111, doi: 10.3847/1538-4357/ab042cDOI
- [49]King, I. R. 1966, AJ, 71, 64, doi: 10.1086/109857DOI
- [50]Koposov, S. E., Belokurov, V., Torrealba, G., & Evans, N. W. 2015, ApJ, 805, 130, doi: 10.1088/0004-637X/805/2/130DOI
- [51]Kron, R. G. 1980, ApJS, 43, 305, doi: 10.1086/190669DOI
- [52]Kroupa, P. 2001, MNRAS, 322, 231, doi: 10.1046/j.1365-8711.2001.04022.xDOI
- [53]Laevens, B. P. M., Martin, N. F., Ibata, R. A., et al. 2015, ApJL, 802, L18, doi: 10.1088/2041-8205/802/2/L18DOI
- [54]Laureijs, R., Amiaux, J., Arduini, S., et al. 2011, arXiv e-prints, arXiv:1110.3193, doi: 10.48550/arXiv.1110.3193DOI
- [55]Leitner, S. N., & Kravtsov, A. V. 2011, ApJ, 734, 48, doi: 10.1088/0004-637X/734/1/48DOI
- [56]Libralato, M., Bedin, L. R., Griggio, M., et al. 2024, A&A, 692, A96, doi: 10.1051/0004-6361/202452295DOI
- [57]Libralato, M., Griggio, M., Gerasimov, R., et al. 2026, A&A, 712, A174, doi: 10.1051/0004-6361/202661247DOI
- [58]Mackey, A. D., & van den Bergh, S. 2005, MNRAS, 360, 631, doi: 10.1111/j.1365-2966.2005.09080.xDOI
- [59]Madore, B. F., & Arp, H. C. 1979, ApJL, 227, L103, doi: 10.1086/182876DOI
- [60]Massari, D., Koppelman, H. H., & Helmi, A. 2019, A&A, 630, L4, doi: 10.1051/0004-6361/201936135DOI
- [61]Massari, D., Dalessandro, E., Erkal, D., et al. 2025, A&A, 697, A8, doi: 10.1051/0004-6361/202449696DOI
- [62]Mau, S., Cerny, W., Pace, A. B., et al. 2020, ApJ, 890, 136, doi: 10.3847/1538-4357/abcc67
- [63]McMahon, R. G., Banerji, M., Gonzalez, E., et al. 2013, The Messenger, 154, 35
- [64]McQuinn, K. B. W., B. Newman, M. J., Savino, A., et al. 2024, ApJ, 961, 16, doi: 10.3847/1538-4357/ad1105DOI
- [65]Milone, A. P., Cordoni, G., Marino, A. F., et al. 2025, A&A, 696, A221, doi: 10.1051/0004-6361/202553746DOI
- [66]Muñoz, R. R., Côté, P., Santana, F. A., et al. 2018, ApJ, 860, 66, doi: 10.3847/1538-4357/aac16bDOI
- [67]Mutlu-Pakdil, B., Sand, D. J., Crnojević, D., et al. 2021, ApJ, 918, 88, doi: 10.3847/1538-4357/ac0db8
- [68]Pace, A. B. 2025, The Open Journal of Astrophysics, 8, 142, doi: 10.33232/001c.144859
- [69]Pai, A., Drlica-Wagner, A., Ferguson, P. S., et al. 2026, arXiv e-prints, arXiv:2607.25044, doi: 10.48550/arXiv.2607.25044DOI
- [70]Perrin, M. D., Sivaramakrishnan, A., Lajoie, C.-P., et al. 2014, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9143, Space Telescopes and Instrumentation 2014: Optical, Infrared, and Millimeter Wave, ed. J. M. Oschmann, Jr., M. Clampin, G. G. Fazio, & H. A. MacEwen, 91433X, doi: 10.1117/12.2056689DOI
- [71]Plummer, H. C. 1911, MNRAS, 71, 460, doi: 10.1093/mnras/71.5.460DOI
- [72]Radburn-Smith, D. J., de Jong, R. S., Seth, A. C., et al. 2011, ApJS, 195, 18, doi: 10.1088/0067-0049/195/2/18
- [73]Rodrigo, C., & Solano, E. 2020, in XIV.0 Scientific Meeting (virtual) of the Spanish Astronomical Society, 182
- [74]Rodrigo, C., Solano, E., & Bayo, A. 2012, SVO Filter Profile Service Version 1.0, Tech. rep., IVOA, doi: 10.5479/ADS/bib/2021ivoa.rept.1015R
- [75]Rodrigo, C., Cruz, P., Aguilar, J. F., et al. 2024, A&A, 689, A93, doi: 10.1051/0004-6361/202449998DOI
- [76]Salaris, M., Chieffi, A., & Straniero, O. 1993, ApJ, 414, 580, doi: 10.1086/173105DOI
- [77]Savino, A., Weisz, D. R., Dolphin, A. E., et al. 2025, ApJ, 979, 205, doi: 10.3847/1538-4357/ada24fDOI
- [78]Schlafly, E. F., & Finkbeiner, D. P. 2011, ApJ, 737, 103, doi: 10.1088/0004-637X/737/2/103DOI
- [79]Shread, E. E., Piotrowska, J. M., Weisz, D. R., Boylan-Kolchin, M., & Harrison, F. A. 2026, ApJ, 1003, 127, doi: 10.3847/1538-4357/ae644cDOI
- [80]Simon, J. D. 2019, ARA&A, 57, 375, doi: 10.1146/annurev-astro-091918-104453DOI
- [81]Skillman, E. D., Monelli, M., Weisz, D. R., et al. 2017, ApJ, 837, 102, doi: 10.3847/1538-4357/aa60c5DOI
- [82]Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163, doi: 10.1086/498708DOI
- [83]Smith, S. E. T., Cerny, W., Hayes, C. R., et al. 2024, ApJ, 961, 92, doi: 10.3847/1538-4357/ad0d9fDOI
- [84]Stetson, P. B., Bolte, M., Harris, W. E., et al. 1999, AJ, 117, 247, doi: 10.1086/300670DOI
- [85]Suntzeff, N., Olszewski, E., & Stetson, P. B. 1985, AJ, 90, 1481, doi: 10.1086/113858DOI
- [86]Tan, C. Y., Drlica-Wagner, A., Pace, A. B., et al. 2026, ApJ, 1000, 87, doi: 10.3847/1538-4357/ae4479DOI
- [87]Tsiane, K., Mau, S., Drlica-Wagner, A., et al. 2025, The Open Journal of Astrophysics, 8, 89, doi: 10.33232/001c.142072
- [88]Vincenzo, F., Matteucci, F., de Boer, T. J. L., Cignoni, M., & Tosi, M. 2016, MNRAS, 460, 2238, doi: 10.1093/mnras/stw1145DOI
- [89]Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, Nature Methods, 17, 261, doi: 10.1038/s41592-019-0686-2
- [90]Warfield, J. T., Richstein, H., Kallivayalil, N., et al. 2023, Research Notes of the American Astronomical Society, 7, 23, doi: 10.3847/2515-5172/acbb72DOI
- [91]Weisz, D. R., Koposov, S. E., Dolphin, A. E., et al. 2016, ApJ, 822, 32, doi: 10.3847/0004-637X/822/1/32DOI
- [92]Weisz, D. R., Dolphin, A. E., Savino, A., et al. 2024, ApJS, 271, 47, doi: 10.3847/1538-4357/ad2600
- [93]Williams, B. F., Lang, D., Dalcanton, J. J., et al. 2014, ApJS, 215, 9, doi: 10.1088/0067-0049/215/1/9DOI
- [94]Willman, B., Blanton, M. R., West, A. A., et al. 2005a, AJ, 129, 2692, doi: 10.1086/430214DOI
- [96]Zucker, D. B., Belokurov, V., Evans, N. W., et al. 2006a, ApJL, 650, L41, doi: 10.1086/508628DOI










