Search NASA⌕ Search

Engineering topics

Whitmore, Bradley C.

Publications and source records attributed to Whitmore, Bradley C..

Star cluster classification using deep transfer learning with PHANGS- HST

Currently available star cluster catalogues from the Hubble Space Telescope (HST) imaging of nearby galaxies heavily rely on visual inspection and classification of candidate clusters. Here, the time-consuming nature of this process has limited the production of reliable catalogues and thus also post-observation analysis. To address this problem, deep transfer learning has recently been used to create neural network models that accurately classify star cluster morphologies at production scale for nearby spiral galaxies (D ≲ 20 Mpc). Here, we use HST ultraviolet (UV)–optical imaging of over 20 000 sources in 23 galaxies from the Physics at High Angular resolution in Nearby GalaxieS (PHANGS) survey to train and evaluate two new sets of models: (i) distance-dependent models, based on cluster candidates binned by galaxy distance (9–12, 14–18, and 18–24 Mpc), and (ii) distance-independent models, based on the combined sample of candidates from all galaxies. We find that the overall accuracy of both sets of models is comparable to previous automated star cluster classification studies (~60–80 per cent) and shows improvement by a factor of 2 in classifying asymmetric and multipeaked clusters from PHANGS-HST. Somewhat surprisingly, while we observe a weak negative correlation between model accuracy and galactic distance, we find that training separate models for the three distance bins does not significantly improve classification accuracy. We also evaluate model accuracy as a function of cluster properties such as brightness, colour, and spectral energy distribution (SED)-fit age. Based on the success of these experiments, our models will provide classifications for the full set of PHANGS-HST candidate clusters (N ~ 200 000) for public release.

79 ASTRONOMY AND ASTROPHYSICS↗

Deep transfer learning for star cluster classification: I. application to the PHANGS– HST survey

ABSTRACT We present the results of a proof-of-concept experiment that demonstrates that deep learning can successfully be used for production-scale classification of compact star clusters detected in Hubble Space Telescope(HST) ultraviolet-optical imaging of nearby spiral galaxies ($D\lesssim 20\, \textrm{Mpc}$) in the Physics at High Angular Resolution in Nearby GalaxieS (PHANGS)–HST survey. Given the relatively small nature of existing, human-labelled star cluster samples, we transfer the knowledge of state-of-the-art neural network models for real-object recognition to classify star clusters candidates into four morphological classes. We perform a series of experiments to determine the dependence of classification performance on neural network architecture (ResNet18 and VGG19-BN), training data sets curated by either a single expert or three astronomers, and the size of the images used for training. We find that the overall classification accuracies are not significantly affected by these choices. The networks are used to classify star cluster candidates in the PHANGS–HST galaxy NGC 1559, which was not included in the training samples. The resulting prediction accuracies are 70 per cent, 40 per cent, 40–50 per cent, and 50–70 per cent for class 1, 2, 3 star clusters, and class 4 non-clusters, respectively. This performance is competitive with consistency achieved in previously published human and automated quantitative classification of star cluster candidate samples (70–80 per cent, 40–50 per cent, 40–50 per cent, and 60–70 per cent). The methods introduced herein lay the foundations to automate classification for star clusters at scale, and exhibit the need to prepare a standardized data set of human-labelled star cluster classifications, agreed upon by a full range of experts in the field, to further improve the performance of the networks introduced in this study.

Wei, Wei↗

The age, mass, and size distributions of star clusters in M51

We present a new catalog of 3816 compact star clusters in the grand design spiral galaxy M51 based on observations taken with the Hubble Space Telescope. The age distribution of the clusters declines starting at very young ages, and can be represented by a power law, dN/dτ∝τ{sup γ}, with γ=−0.65±0.15. No significant changes in the shape of the age distribution at different masses is observed. The mass function of the clusters younger than τ ≈ 400 Myr can also be described by a power law, dN/dM∝M{sup β}, with β ≈ −2.1±0.2. We compare these distributions with the predictions from various cluster disruption models, and find that they are consistent with models where clusters disrupt approximately independent of their initial mass, but not with models where lower mass clusters are disrupted earlier than their higher mass counterparts. We find that the half-light radii of clusters more massive than M ≈ 3×10{sup 4} M{sub ⊙} and with ages between 100 and 400 Myr are larger by a factor of ≈3–4 than their counterparts that are younger than 10{sup 7} years old, suggesting that the clusters physically expand during their early life.

79 ASTRONOMY AND ASTROPHYSICS↗

VERSION 1 OF THE HUBBLE SOURCE CATALOG

The Hubble Source Catalog is designed to help optimize science from the Hubble Space Telescope (HST) by combining the tens of thousands of visit-based source lists in the Hubble Legacy Archive (HLA) into a single master catalog. Version 1 of the Hubble Source Catalog includes WFPC2, ACS/WFC, WFC3/UVIS, and WFC3/IR photometric data generated using SExtractor software to produce the individual source lists. The catalog includes roughly 80 million detections of 30 million objects involving 112 different detector/filter combinations, and about 160,000 HST exposures. Source lists from Data Release 8 of the HLA are matched using an algorithm developed by Budavári and Lubow. The mean photometric accuracy for the catalog as a whole is better than 0.10 mag, with relative accuracy as good as 0.02 mag in certain circumstances (e.g., bright isolated stars). The relative astrometric residuals are typically within 10 mas, with a value for the mode (i.e., most common value) of 2.3 mas. The absolute astrometric accuracy is better than 0{sup ″} .1 for most sources, but can be much larger for a fraction of fields that could not be matched to the PanSTARRS, SDSS, or 2MASS reference systems. In this paper we describe the database design with emphasis on those aspects that enable the users to fully exploit the catalog while avoiding common misunderstandings and potential pitfalls. We provide usage examples to illustrate some of the science capabilities and data quality characteristics, and briefly discuss plans for future improvements to the Hubble Source Catalog.

79 ASTRONOMY AND ASTROPHYSICS↗

STAR CLUSTER PROPERTIES IN TWO LEGUS GALAXIES COMPUTED WITH STOCHASTIC STELLAR POPULATION SYNTHESIS MODELS

We investigate a novel Bayesian analysis method, based on the Stochastically Lighting Up Galaxies (slug) code, to derive the masses, ages, and extinctions of star clusters from integrated light photometry. Unlike many analysis methods, slug correctly accounts for incomplete initial mass function (IMF) sampling, and returns full posterior probability distributions rather than simply probability maxima. We apply our technique to 621 visually confirmed clusters in two nearby galaxies, NGC 628 and NGC 7793, that are part of the Legacy Extragalactic UV Survey (LEGUS). LEGUS provides Hubble Space Telescope photometry in the NUV, U, B, V, and I bands. We analyze the sensitivity of the derived cluster properties to choices of prior probability distribution, evolutionary tracks, IMF, metallicity, treatment of nebular emission, and extinction curve. We find that slug's results for individual clusters are insensitive to most of these choices, but that the posterior probability distributions we derive are often quite broad, and sometimes multi-peaked and quite sensitive to the choice of priors. In contrast, the properties of the cluster population as a whole are relatively robust against all of these choices. We also compare our results from slug to those derived with a conventional non-stochastic fitting code, Yggdrasil. We show that slug's stochastic models are generally a better fit to the observations than the deterministic ones used by Yggdrasil. However, the overall properties of the cluster populations recovered by both codes are qualitatively similar.

79 ASTRONOMY AND ASTROPHYSICS↗

The link between the formation rates of clusters and stars in galaxies

The goal of this paper is to test whether the formation rate of star clusters is proportional to the star formation rate (SFR) in galaxies. As a first step, we present the mass functions of compact clusters younger than 10 Myr in seven star-forming galaxies of diverse masses, sizes, and morphologies: the Large and Small Magellanic Clouds, NGC 4214, NGC 4449, M83, M51, and the Antennae. These cluster mass functions (CMFs) are well represented by power laws, dN/dM∝M{sup β}, with similar exponents β=−1.92±0.27, but with amplitudes that differ by factors up to ∼10{sup 3}, corresponding to vast differences in the sizes of the cluster populations in these galaxies. We then normalize these CMFs by the SFRs in the galaxies, derived from dust-corrected Hα luminosities, and find that the spread in the amplitudes collapses, with a remaining rms deviation of only σ(logA)=0.2. This is close to the expected dispersion from random uncertainties in the CMFs and SFRs. Thus, the data presented here are consistent with exact proportionality between the formation rates of stars and clusters. However, the data also permit weak deviations from proportionality, at the factor of two level, within the statistical uncertainties. We find the same spread in amplitudes when we normalize the mass functions of much older clusters, with ages in the range 100–400 Myr, by the current SFR. This is another indication of the general similarity among the cluster populations of different galaxies.

79 ASTRONOMY AND ASTROPHYSICS↗

ALMA observations of the Antennae galaxies. I. A new window on a prototypical merger

We present the highest spatial resolution (≈0.''5) CO (3-2) observations to date of the 'overlap' region in the merging Antennae galaxies (NGC 4038/39), taken with the Atacama Large Millimeter/submillimeter Array. We report on the discovery of a long (3 kpc), thin (aspect ratio 30/1), filament of CO gas that breaks up into roughly 10 individual knots. Each individual knot has a low internal velocity dispersion (≈10 km s{sup –1}); the dispersion of the ensemble of knots in the filament is also low (≈10 km s{sup –1}). At the other extreme, we find that the individual clouds in the supergiant molecular cloud 2 region discussed by Wilson and collaborators have a large range of internal velocity dispersions (10 to 80 km s{sup –1}), and a large dispersion among the ensemble (≈80 km s{sup –1}). Other large-scale features observed in CO emission, and their correspondence with historical counterparts using observations in other wavelengths, are also discussed. We compare the locations of small-scale CO features with a variety of multi-wavelength observations, in particular broad- (BVIJH) and narrow-band data (H{sub α} and Pa{sub β}) taken with the Hubble Space Telescope, and radio (3.6 cm) continuum observations taken with the Karl G. Jansky Very Large Array. This comparison leads to the development of an evolutionary classification system that provides a framework for studying the sequence of star cluster formation and evolution—from diffuse supergiant molecular clouds (SGMCs) to proto, embedded, emerging, young, intermediate/old clusters. The relative timescales have been assessed by determining the fractional population of sources at each evolutionary stage. The main uncertainty in this estimate is the identification of four regions as candidate protoclusters (i.e., strong compact CO emission but no clearly associated radio emission). Using the evolutionary framework, we estimate that the maximum age range of clusters in a single GMC is ≈10 Myr, which suggests that the molecular gas is removed over this timescale, resulting in the cessation of star formation and the destruction of the GMC within a radius of about 200 pc.

79 ASTRONOMY AND ASTROPHYSICS↗

An expanded HST/WFC3 survey of M83: Project overview and targeted supernova remnant search

We present an optical/NIR imaging survey of the face-on spiral galaxy M83, using data from the Hubble Space Telescope Wide Field Camera 3 (WFC3). Seven fields are used to cover a large fraction of the inner disk, with observations in nine broadband and narrowband filters. In conjunction with a deep Chandra survey and other new radio and optical ground-based work, these data enable a broad range of science projects to be pursued. We provide an overview of the WFC3 data and processing and then delve into one topic, the population of young supernova remnants (SNRs). We used a search method targeted toward soft X-ray sources to identify 26 new SNRs. Many compact emission nebulae detected in [Fe II] 1.644 μm align with known remnants and this diagnostic has also been used to identify many new remnants, some of which are hard to find with optical images. We include 37 previously identified SNRs that the data reveal to be <0.''5 in angular size and thus are difficult to characterize from ground-based data. The emission line ratios seen in most of these objects are consistent with shocks in dense interstellar material rather than showing evidence of ejecta. We suggest that the overall high elemental abundances in combination with high interstellar medium pressures in M83 are responsible for this result. Future papers will expand on different aspects of the these data including a more comprehensive analysis of the overall SNR population.

79 ASTRONOMY AND ASTROPHYSICS↗

Star-cluster mass and age distributions of two fields in M83 based on HST/WFC3 observations

We study star clusters in two fields in the nearby spiral galaxy M83 using broadband and narrowband optical imaging taken with the Wide Field Camera 3 onboard the Hubble Space Telescope. We present results on the basis of several different catalogs of star clusters in inner and outer fields, and we conclude that different methods of selection do not strongly affect the results, particularly for clusters older than ≈10 Myr. The age distributions can be described by a power law, dN/dτ∝τ{sup γ}, with γ ≈ –0.84 ± 0.12 in the inner field, and γ ≈ –0.48 ± 0.12 in the outer field for τ ≳ 10 Myr. We bracket the difference, Δγ, between the two fields to be in the 0.18 to 0.36 range, based on estimates of the relative star-formation histories. The mass functions can also be described by a power law, dN/dM∝M {sup β}, with β ≈ –1.98 ± 0.14 and β ≈ –2.34 ± 0.26 in the inner and outer fields, respectively. We conclude that the shapes of the mass and age distributions of the clusters in the two fields are similar, as predicted by the quasi-universal model. Any differences between the two fields are at the ≈2σ-3σ (≈1σ-2σ) level for the age (mass) distributions. Therefore, any dependence of these distributions on the local environment is probably weak. We compare the shapes of the distributions with those predicted by two popular cluster disruption models. We find that both show evidence that the clusters are disrupted at a rate that is approximately independent of their mass. We compare the shapes of the distributions with those predicted by two popular cluster disruption models, and find that both show evidence that the clusters are disrupted at a rate that is approximately independent of their mass, and that observational results do not support the earlier disruption of lower-mass clusters relative to their higher-mass counterparts.

79 ASTRONOMY AND ASTROPHYSICS↗

The luminosity function of star clusters in 20 star-forming galaxies based on Hubble legacy archive photometry

Luminosity functions (LFs) have been determined for star cluster populations in 20 nearby (4-30 Mpc), star-forming galaxies based on Advanced Camera for Surveys source lists generated by the Hubble Legacy Archive (HLA). These cluster catalogs provide one of the largest sets of uniform, automatically generated cluster candidates available in the literature at present. Comparisons are made with other recently generated cluster catalogs demonstrating that the HLA-generated catalogs are of similar quality, but in general do not go as deep. A typical cluster LF can be approximated by a power law, dN/dL∝L {sup α}, with an average value for α of –2.37 and rms scatter = 0.18 when using the F814W ('I') band. A comparison of fitting results based on methods that use binned and unbinned data shows good agreement, although there may be a systematic tendency for the unbinned (maximum likelihood) method to give slightly more negative values of α for galaxies with steeper LFs. We find that galaxies with high rates of star formation (or equivalently, with the brightest or largest numbers of clusters) have a slight tendency to have shallower values of α. In particular, the Antennae galaxy (NGC 4038/39), a merging system with a relatively high star formation rate (SFR), has the second flattest LF in the sample. A tentative correlation may also be present between Hubble type and values of α, in the sense that later type galaxies (i.e., Sd and Sm) appear to have flatter LFs. Hence, while there do appear to be some weak correlations, the relative similarity in the values of α for a large number of star-forming galaxies suggests that, to first order, the LFs are fairly universal. We examine the bright end of the LFs and find evidence for a downturn, although it only pertains to about 1% of the clusters. Our uniform database results in a small scatter (≈0.4 to 0.5 mag) in the correlation between the magnitude of the brightest cluster (M {sub brightest}) and log of the number of clusters brighter than M{sub I} = –9 (log N). We also examine the magnitude of the brightest cluster versus log SFR for a sample including both dwarf galaxies and ULIRGs. This shows that the correlation extends over roughly six orders of magnitude but with scatter that is larger than for our spiral sample, probably because of the high levels of extinction in many of the LIRGs.

79 ASTRONOMY AND ASTROPHYSICS↗

The Resolved Stellar Population in 50 Regions of M83 from HST/WFC3 Early Release Science Observations

We present a multi-wavelength photometric study of approximately 15,000 resolved stars in the nearby spiral galaxy M83 (NGC 5236, D = 4.61 Mpc) based on Hubble Space Telescope Wide Field Camera 3 observations using four filters: F336W, F438W, F555W, and F814W. We select 50 regions (an average size of 260 pc by 280 pc) in the spiral arm and inter-arm areas of M83 and determine the age distribution of the luminous stellar populations in each region. This is accomplished by correcting for extinction toward each individual star by comparing its colors with predictions from stellar isochrones.We compare the resulting luminosity-weighted mean ages of the luminous stars in the 50 regions with those determined from several independent methods, including the number ratio of red-to-blue supergiants, morphological appearance of the regions, surface brightness fluctuations, and the ages of clusters in the regions. We find reasonably good agreement between these methods. We also find that young stars are much more likely to be found in concentrated aggregates along spiral arms, while older stars are more dispersed. These results are consistent with the scenario that star formation is associated with the spiral arms, and stars form primarily in star clusters and then disperse on short timescales to form the field population. The locations ofWolf-Rayet stars are found to correlate with the positions of many of the youngest regions, providing additional support for our ability to accurately estimate ages. We address the effects of spatial resolution on the measured colors, magnitudes, and age estimates. While individual stars can occasionally show measurable differences in the colors and magnitudes, the age estimates for entire regions are only slightly affected.

Hubble Space Telescope↗

Hubble space telescope observations of young star clusters in NGC-4038/4039, 'the antennae' galaxies

New, high-resolution images of the disks of NGC 4038/4039 obtained with the Wide Field Camera of the Hubble Space Telescope (HST) are presented. NGC 4038/4039, nicknamed The Antennae, is a prototypical example of a pair of colliding galaxies believed to be at an early stage of a merger. Down to the limiting magnitude of V approximately 23 mag, the HST images reveal a population of over 700 blue pointlike objects within the disks. The mean absolute magnitude of these objects is M(sub V) = -11 mag, with the brightest objects reaching M(sub V) approximately -15. Their mean apparent color indices ar U - V = -0.7 mag and V - 1 = 0.8 mag on the Johnson UVI passband system, while their mean indices corrected for internal reddening are (u - v)(sub 0) = -1.0 mag and (V - I(sub 0) = 0.5. Their mean effective radius, determined from slightly resolved images, is 18 pc (for H(sub 0) = 50 km/s /Mpc). Based on their luminosities and resolution, most of these objects cannot be individual stars, but are likely young compact star clusters. The brighter ones are similar to the objects found in NGC 1275 and NGC 7252, which appear to be young globular clusters formed during recent galazy mergers. Based on their U - V and V - I colors, the brightest, bluest clusters of NGC 4038/4039 appear to be less than 10 Myr old. Most of these bright clusters are relatively tightly clustered themselves, with typically a dozen individual clusters belonging to a complex identified as a giant H II region from ground-based observations. The cluster luminosity function (LF) is approximately a power law, phi(L)dL proportional to L(exp -1.78+/-0.05)dL, with no hint of a turnover at fainter magnitudes. This power-law shape agrees with the LF of Magellanic Cloud clusters and Galactic open clusters, but differs from the LF of old globular cluster systems that is typically Gaussian with a Full Width at Half Maximum (FWHM) of approximately 3 mag. Besides the blue clusters, we also find about a dozen extremely red objects with V - I greater than 3.0. The highest number density of these red objects is found in the SE quadrant, where star formation appears to be most recent. We propose that these objects may be very young star clusters still embedded in their placental dust cocoons.

Whitmore, Bradley C.↗

Hubble Space Telescope discovery of candidate young globular clusters in the merger remnant NGC 7252

New, high-resolution images of the central region of NGC 7252 obtained with the Planetary Camera of the HST are presented. NGC 7252 is a prototypical example of a remnant of two merged disk galaxies. Our most striking result is the discovery of a population of about 40 blue pointlike objects in this galaxy. The mean absolute magnitude of these objects is Mv = -13 mag; the mean color is V-I = 0.7 mag; and the mean effective radius is 10 pc. The luminosities, colors, projected spatial distribution, and sizes are all compatible with the hypothesis that these objects formed within the last 1 Gyr following the collision of two spiral galaxies, and that they are young globular clusters. It therefore appears that the number of globular clusters may increase during the merger of gas-rich galaxies. This weakens van den Bergh's objection against ellipticals being formed through disk mergers, based mainly on the fact that disk galaxies have fewer globular clusters per unit luminosity than ellipticals do. NGC 7252 shows a single, semistellar nucleus; relatively bright spiral structure is seen within 1.6 kpc of the center, presumably formed through the continued infall of gas into a disk around the center of the galaxy.

Whitmore, Bradley C.↗

What determines the morphological fractions in clusters of galaxies?

A reexamination of Dressler's (1980) sample of nearly 6000 galaxies in 55 clusters shows that the morphology-clustercentric radius relation is more fundamental than the morphology-local density relation. This conclusion is supported by improved correlations when the projected clustercentric radius is used as the independent parameter, and by a comparison of galaxies with the same normalized clustercentric radii by different values of the projected local density.

Whitmore, Bradley C.↗

Environmental influences on galaxy evolution

We investigate the role of mergers and interactions in the evolution of galaxies by studying galaxies in compact groups. Compact groups of galaxies have high spatial densities and low velocity dispersions making these regions ideal laboratories in which to study the effect of interactions and mergers. Based on a detailed spectroscopic and multi-color imaging study, we find that both the isophotal shapes and the stellar kinematics indicate that many of the elliptical galaxies in compact groups have been affected by tidal interactions. At the same time, however, we find that only a few elliptical galaxies in compact groups have evidence for the young stellar populations that would be expected if they are the result of recent merger of two spiral galaxies. Therefore, we conclude that tidal interactions affect galaxy properties at the current epoch, but the bulk of basic galaxy formation and transformation must have occurred at much higher redshift.

Zepf, Stephen E.↗

Blue ellipticals in compact groups

By studying galaxies in compact groups, the authors examine the hypothesis that mergers of spiral galaxies make elliptical galaxies. The authors combine dynamical models of the merger-rich compact group environment with stellar evolution models and predict that roughly 15 percent of compact group ellipticals should be 0.15 mag bluer in B - R color than normal ellipticals. The published colors of these galaxies suggest the existence of this predicted blue population, but a normal distribution with large random errors can not be ruled out based on these data alone. However, the authors have new ultraviolet blue visual data which confirm the blue color of the two ellipticals with blue B - R colors for which they have their own colors. This confirmation of a population of blue ellipticals indicates that interactions are occurring in compact groups, but a blue color in one index alone does not require that these ellipticals are recent products of the merger of two spirals. The authors demonstrate how optical spectroscopy in the blue may distinguish between a true spiral + spiral merger and the swallowing of a gas-rich system by an already formed elliptical. The authors also show that the sum of the luminosity of the galaxies in each group is consistent with the hypothesis that the final stage in the evolution of compact group is an elliptical galaxy.

Zepf, Stephen E.↗

New observations and a photographic atlas of polar-ring galaxies

A photographic atlas of polar-ring galaxies and related objects is presented. The atlas includes kinematically confirmed polar-ring galaxies (category A), good candidates based on their morphological appearance (category B), possible candidates (category C), and possibly related objects (category D). New photometric and kinematic observations are reported for several galaxies in the catalog, including observations that show that UGC 7576 and UGC 9796 ( = II ZW 73) are S0 galaxies with polar rings. Roughly 0.5 percent of all nearby S0 galaxies appear to have polar rings. When corrected for various selection effects (e.g., nonoptimal viewing orientation, possible dimming, or limited lifetime of the ring) the percentage increases to about 5 percent of S0 galaxies which have, or have had a polar ring.

Whitmore, Bradley C.↗