Search NASA⌕ Search

Engineering topics

Lee, Janice C.

Publications and source records attributed to Lee, Janice 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↗

Spectral Energy Distributions for 258 Local Volume Galaxies

We present model spectral energy distribution (SED) fits to ultraviolet/optical/infrared observations for the 258 nearby galaxies in the Local Volume Legacy survey, a sample dominated by lower-luminosity dwarf irregular systems. The data for each galaxy include up to 26 spatially-integrated broadband and narrowband fluxes from the Galaxy Evolution Explorer, Spitzer Space Telescope, and Infrared Astronomical Satellite space-based platforms, and from the Sloan Digital Sky Survey, Two Micron All Sky Survey, and other ground-based efforts. The CIGALE SED fitting package is employed using a delayed star-formation history with an optional late burst or quenching episode to constrain 11 different free parameters that characterize the properties of each galaxy's stellar and dust emission, with the overriding constraint that the ultraviolet/optical emission absorbed by interstellar dust grains is emitted in equal energy portions at infrared wavelengths. The main results are: (i) 94% of the SED fits yield reduced χ 2 values less than 3; (ii) the modeled stellar masses agree with those derived from 3.6 μm-based measures with a scatter of 0.07 dex; (iii) for a typical galaxy in the sample the SED-derived star-formation rate averaged over the past 100 Myr is about 88% of the value derived from standard hybrid indicators on similar timescales; and (iv) there is a statistically significant inverse relation between the stellar mass fraction appearing in the late burst and the total stellar mass. These results build upon prior SED modeling efforts in the local volume and lay the groundwork for future studies of more distant low-metallicity galaxies with JWST.

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↗

GALEX–SDSS–WISE Legacy Catalog (GSWLC): Star formation rates, stellar masses, and dust attenuations of 700,000 low-redshift galaxies

In this paper, we present the GALEX–SDSS–WISE Legacy Catalog (GSWLC), a catalog of physical properties (stellar masses, dust attenuations, and star formation rates [SFRs]) for ∼700,000 galaxies with Sloan Digital Sky Survey (SDSS) redshifts below 0.3. GSWLC contains galaxies within the Galaxy Evolution Explorer footprint, regardless of a UV detection, covering 90% of SDSS. The physical properties were obtained from UV/optical spectral energy distribution (SED) fitting following Bayesian methodology of Salim et al., with improvements such as blending corrections for low-resolution UV photometry, flexible dust attenuation laws, and emission-line corrections. GSWLC also includes mid-IR SFRs derived from IR templates based on 22 μm Wide-field Infrared Survey Explorer observations. These estimates are independent of UV/optical SED fitting, in order to separate possible systematics. The paper argues that the comparison of specific SFRs (sSFRs) is more informative and physically motivated than the comparison of SFRs. The sSFRs resulting from the UV/optical SED fitting are compared to the mid-IR sSFRs and to sSFRs from three published catalogs. For “main-sequence” galaxies with no active galactic nucleus (AGN) all sSFRs are in very good agreement (within 0.1 dex on average). In particular, the widely used aperture-corrected SFRs from the MPA/JHU catalog show no systematic offsets, in contrast to some integral field spectroscopy results. For galaxies below the main sequence (log sSFR <−11), mid-IR (s)SFRs based on fixed luminosity–SFR conversion are severely biased (up to 2 dex) because the dust is primarily heated by old stars. Furthermore, mid-IR (s)SFRs are overestimated by up to 0.6 dex for galaxies with AGNs, presumably due to nonstellar dust heating. UV/optical (s)SFRs are thus preferred to IR-based (s)SFRs for quenched galaxies and those that host AGNs.

79 ASTRONOMY AND ASTROPHYSICS↗

DIRECT DETECTION OF LYMAN CONTINUUM ESCAPE FROM LOCAL STARBURST GALAXIES WITH THE COSMIC ORIGINS SPECTROGRAPH

We report on the detection of Lyman continuum radiation in two nearby starburst galaxies. Tol 0440-381, Tol 1247-232, and Mrk 54 were observed with the Cosmic Origins Spectrograph on board the Hubble Space Telescope . The three galaxies have radial velocities of ∼13,000 km s{sup −1}, permitting a ∼35 Å window on the restframe Lyman continuum shortward of the Milky Way Lyman edge at 912 Å. The chosen instrument configuration using the G140L grating covers the spectral range from 912 to 2000 Å. We developed a dedicated background subtraction method to account for the temporal and spatial background variations of the detector, which is crucial at the low flux levels around 912 Å. This modified pipeline allowed us to significantly improve the statistical and systematic detector noise and will be made available to the community. We detect Lyman continuum in all three galaxies. However, we conservatively interpret the emission in Tol 0440-381 as an upper limit due to possible contamination by geocoronal Lyman series lines. We determined the current star formation properties from the far-ultraviolet continuum and spectral lines and used synthesis models to predict the Lyman continuum radiation emitted by the current population of hot stars. We discuss various model uncertainties such as, among others, atmospheres and evolution models. Lyman continuum escape fractions were derived from a comparison between the observed and predicted Lyman continuum fluxes. Tol 1247-232, Mrk 54, and Tol 0440-381 have absolute escape fractions of (4.5 ± 1.2)%, (2.5 ± 0.72)%, and <(7.1 ± 1.1)%, respectively.

79 ASTRONOMY AND ASTROPHYSICS↗

A DEEPER LOOK AT FAINT Hα EMISSION IN NEARBY DWARF GALAXIES

We present deep Hα imaging of three nearby dwarf galaxies, carefully selected to optimize observations with the Maryland-Magellan Tunable Filter (MMTF) on the Magellan 6.5 m telescope. An effective bandpass of ∼13 Å is used, and the images reach 3σ flux limits of ∼8 × 10{sup −18} erg s{sup −1} cm{sup −2}, which is about an order of magnitude lower than standard narrowband observations obtained by the most recent generation of local Hα galaxy surveys. The observations were originally motivated by the finding that the Hα/FUV flux ratio of galaxies systematically declines as global galactic properties such as the star formation rate (SFR) and stellar mass decrease. The three dwarf galaxies selected for study have SFRs that, when calculated from their Hα luminosities using standard conversion recipes, are ∼50% of those based on the FUV. Follow-up studies of many of the potential causes for the trends in the Hα/FUV flux ratio have been performed, but the possibility that previous observations have missed a non-negligible fraction of faint ionized emission in dwarf galaxies has not been investigated. The MMTF observations reveal both diffuse and structured Hα emission (filaments, shells, possible single-star H ii regions) spanning extents up to 2.5 times larger relative to previous observations. However, only up to an additional ∼5% of Hα flux is captured, which does not account for the trends in the Hα/FUV ratio. Beyond investigation of the Hα/FUV ratio, the impact of the newly detected extended flux on our understanding of star formation, the properties of H ii regions, and the propagation of ionizing photons warrant further investigation.

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↗

On the mass–metallicity–star formation rate relation for galaxies at z∼2

Recent studies have shown that the local mass–metallicity (M{sub *}–Z) relation depends on the specific star formation rate (sSFR). Whether such a dependence exists at higher redshifts, and whether the resulting M{sub *}–Z–SFR relation is redshift invariant, is debated. We re-examine these issues by applying the non-parametric techniques of Salim et al. to ∼130 z∼2.3 galaxies with N2 and O3 measurements from Keck Baryonic Structure Survey (KBSS). We find that the KBSS M{sub *}–Z relation depends on sSFR at intermediate masses where such dependence exists locally. KBSS and SDSS galaxies of the same mass and sSFR (“local analogs”) are similarly offset in the BPT diagram relative to the bulk of local star-forming galaxies, and thus we posit that metallicities can be compared self-consistently at different redshifts as long as the masses and sSFRs of the galaxies are similar. We find that the M{sub *}–Z–SFR relation of z∼2 galaxies is consistent with the local one at logM{sub ∗}<10, but is offset up to −0.25 dex at higher masses, so it is altogether not redshift invariant. This high-mass offset could arise from a bias that [O iii]-based, high-redshift spectroscopic surveys have against high-metallicity galaxies, but additional evidence disfavors this possibility. We identify three causes for the reported discrepancy between N2 and O3N2 metallicities at z∼2: (1) a smaller offset that is also present for SDSS galaxies, which we remove with new N2 calibration, (2) a genuine offset due to differing ISM condition, which is also present in local analogs, and (3) an additional offset due to unrecognized active galactic nucleus contamination.

79 ASTRONOMY AND ASTROPHYSICS↗

LEGUS discovery of a light echo around supernova 2012aw

We have discovered a luminous light echo around the normal Type II-Plateau Supernova (SN) 2012aw in Messier 95 (M95; NGC 3351), detected in images obtained approximately two years after explosion with the Wide Field Channel 3 on board the Hubble Space Telescope by the Legacy ExtraGalactic Ultraviolet Survey. The multi-band observations span from the near-ultraviolet through the optical (F275W, F336W, F438W, F555W, and F814W). The apparent brightness of the echo at the time was ∼21–22 mag in all of these bands. The echo appears circular, although less obviously as a ring, with an inhomogeneous surface brightness, in particular, a prominent enhanced brightness to the southeast. The SN itself was still detectable, particularly in the redder bands. We are able to model the light echo as the time-integrated SN light scattered off of diffuse interstellar dust in the SN environment. We have assumed that this dust is analogous to that in the Milky Way with R{sub V}=3.1. The SN light curves that we consider also include models of the unobserved early burst of light from the SN shock breakout. Our analysis of the echo suggests that the distance from the SN to the scattering dust elements along the echo is ≈45 pc. The implied visual extinction for the echo-producing dust is consistent with estimates made previously from the SN itself. Finally, our estimate of the SN brightness in F814W is fainter than that measured for the red supergiant star at the precise SN location in pre-SN images, possibly indicating that the star has vanished and confirming it as the likely SN progenitor.

79 ASTRONOMY AND ASTROPHYSICS↗

The relationship between stellar mass, gas metallicity, and star formation rate for Hα-selected galaxies at z ≈ 0.8 from the NewHα survey

Using a sample of 299 Hα-selected galaxies at z≈0.8, we study the relationship between galaxy stellar mass, gas-phase metallicity, and star formation rate (SFR), and compare to previous results. We use deep optical spectra obtained with the IMACS spectrograph at the Magellan telescope to measure strong oxygen lines. We combine these spectra and metallicities with (1) rest-frame UV-to-optical imaging, which allows us to determine stellar masses and dust attenuation corrections, and (2) Hα narrowband imaging, which provides a robust measurement of the instantaneous SFR. Our sample spans stellar masses of ∼10{sup 9}–6 × 10{sup 11} M{sub ⊙}, SFRs of 0.4–270 M{sub ⊙} yr{sup −1}, and metal abundances of 12+log(O/H)≈8.3–9.1 (≈0.4–2.6 Z{sub ⊙}). The correlations that we find between the Hα-based SFR and stellar mass (i.e., the star-forming “main sequence”) and between the stellar mass and metallicity are both consistent with previous z∼1 studies of star-forming galaxies. We then study the relationship between the three properties using various plane-fitting techniques and a curve-fitting projection. In all cases, we exclude strong dependence of the M{sub ⋆}–Z relation on SFR, but are unable to distinguish between moderate and no dependence. Our results are consistent with previous mass–metallicity–SFR studies. We check whether data set limitations may obscure a strong dependence on the SFR by using mock samples drawn from the Sloan Digital Sky Survey. These experiments reveal that the adopted signal-to-noise ratio cuts may have a significant effect on the measured dependence. Further work is needed to investigate these results, and to test whether a “fundamental metallicity relation” or a “fundamental plane” describes star-forming galaxies across cosmic time.

79 ASTRONOMY AND ASTROPHYSICS↗

A CRITICAL LOOK AT THE MASS-METALLICITY-STAR FORMATION RATE RELATION IN THE LOCAL UNIVERSE. I. AN IMPROVED ANALYSIS FRAMEWORK AND CONFOUNDING SYSTEMATICS

It has been proposed that the (stellar) mass-(gas) metallicity relation of galaxies exhibits a secondary dependence on star formation rate (SFR), and that the resulting M {sub *}-Z-SFR relation may be redshift-invariant, i.e., ''fundamental''. However, conflicting results on the character of the SFR dependence, and whether it exists, have been reported. To gain insight into the origins of the conflicting results, we (1) devise a non-parametric, astrophysically motivated analysis framework based on the offset from the star-forming ({sup m}ain{sup )} sequence at a given M {sub *} (relative specific SFR); (2) apply this methodology and perform a comprehensive re-analysis of the local M {sub *}-Z-SFR relation, based on SDSS, GALEX, and WISE data; and (3) study the impact of sample selection and of using different metallicity and SFR indicators. We show that metallicity is anti-correlated with specific SFR regardless of the indicators used. We do not find that the relation is spurious due to correlations arising from biased metallicity measurements or fiber aperture effects. We emphasize that the dependence is weak/absent for massive galaxies (log M {sub *} > 10.5), and that the overall scatter in the M {sub *}-Z-SFR relation does not greatly decrease from the M {sub *}-Z relation. We find that the dependence is stronger for the highest SSFR galaxies above the star-forming sequence. This two-mode behavior can be described with a broken linear fit in 12+log(O/H) versus log (SFR/M {sub *}), at a given M {sub *}. Previous parameterizations used for comparative analysis with higher redshift samples that do not account for the more detailed behavior of the local M {sub *}-Z-SFR relation may incorrectly lead to the conclusion that those samples follow a different relationship.

79 ASTRONOMY AND ASTROPHYSICS↗

THE NATURE OF THE SECOND PARAMETER IN THE IRX-{beta} RELATION FOR LOCAL GALAXIES

We present an analysis of 98 galaxies of low-dust content, selected from the Spitzer Local Volume Legacy survey, aimed at examining the relation between the ultraviolet (UV) color and dust attenuation in normal star-forming galaxies. The IRX-{beta} diagram relates the total dust attenuation in a galaxy, traced by the far-IR (FIR) to UV ratio, to the observed UV color, indicated by {beta}. Previous research has indicated that while starburst galaxies exhibit a relatively tight IRX-{beta} relation, normal star-forming galaxies do not, and have a much larger spread in the total-IR to far-UV (FUV) luminosity for a fixed UV color. We examine the role that the age of the stellar population plays as the ''second parameter'' responsible for the observed deviation and spread of star-forming galaxies from the starburst relation. We model the FUV to FIR spectral energy distribution of each galaxy according to two broad bins of star formation history (SFH): constant and instantaneous burst. We find clear trends between stellar population mean age estimators (extinction-corrected FUV/NIR, U - B, and EW(H{alpha})) and the UV color {beta}; the trends are mostly driven by the galaxies best-described by instantaneous burst populations. We also find a significant correlation between {beta} and the mean age directly determined from the best-fit instantaneous models. As already indicated by other authors, the UV attenuation in star-forming galaxies may not be recovered with the UV color alone and is highly influenced by the stellar population's mean age and SFH. Overall, the scatter in the IRX-{beta} diagram is better correlated with {beta} than with the perpendicular distance, d{sub p}.

79 ASTRONOMY AND ASTROPHYSICS↗

MEASURING GALAXY STAR FORMATION RATES FROM INTEGRATED PHOTOMETRY: INSIGHTS FROM COLOR-MAGNITUDE DIAGRAMS OF RESOLVED STARS

We use empirical star formation histories (SFHs), measured from Hubble-Space-Telescope-based resolved star color-magnitude diagrams, as input into population synthesis codes to model the broadband spectral energy distributions (SEDs) of 50 nearby dwarf galaxies (6.5 < log M{sub *}/M{sub Sun} < 8.5, with metallicities {approx}10% solar). In the presence of realistic SFHs, we compare the modeled and observed SEDs from the ultraviolet (UV) through near-infrared and assess the reliability of widely used UV-based star formation rate (SFR) indicators. In the FUV through i bands, we find that the observed and modeled SEDs are in excellent agreement. In the Spitzer 3.6 {mu}m and 4.5 {mu}m bands, we find that modeled SEDs systematically overpredict observed luminosities by up to {approx}0.2 dex, depending on treatment of the TP-AGB stars in the synthesis models. We assess the reliability of UV luminosity as a SFR indicator, in light of independently constrained SFHs. We find that fluctuations in the SFHs alone can cause factor of {approx}2 variations in the UV luminosities relative to the assumption of a constant SFH over the past 100 Myr. These variations are not strongly correlated with UV-optical colors, implying that correcting UV-based SFRs for the effects of realistic SFHs is difficult using only the broadband SED. Additionally, for this diverse sample of galaxies, we find that stars older than 100 Myr can contribute from <5%-100% of the present day UV luminosity, highlighting the challenges in defining a characteristic star formation timescale associated with UV emission. We do find a relationship between UV emission timescale and broadband UV-optical color, though it is different than predictions based on exponentially declining SFH models. Our findings have significant implications for the comparison of UV-based SFRs across low-metallicity populations with diverse SFHs.

79 ASTRONOMY AND ASTROPHYSICS↗

NEBULAR ATTENUATION IN H{alpha}-SELECTED STAR-FORMING GALAXIES AT z = 0.8 FROM THE NewH{alpha} SURVEY

We present measurements of the dust attenuation of H{alpha}-selected emission-line galaxies at z = 0.8 from the NewH{alpha} narrowband survey. The analysis is based on deep follow-up spectroscopy with Magellan/IMACS, which captures the strong rest-frame optical emission lines from [O II] {lambda}3727 to [O III] {lambda}5007. The spectroscopic sample used in this analysis consists of 341 confirmed H{alpha} emitters. We place constraints on the active galactic nucleus (AGN) fraction using diagnostics that can be applied at intermediate redshift. We find that at least 5% of the objects in our spectroscopic sample can be classified as AGNs and 2% are composite, i.e., powered by a combination of star formation and AGN activity. We measure the dust attenuation for individual objects from the ratios of the higher order Balmer lines. The H{beta} and H{gamma} pair of lines is detected with S/N > 5 in 55 individual objects and the H{beta} and H{delta} pair is detected in 50 individual objects. We also create stacked spectra to probe the attenuation in objects without individual detections. The median attenuation at H{alpha} based on the objects with individually detected lines is A(H{alpha}) = 0.9 {+-} 1.0 mag, in good agreement with the attenuation found in local samples of star-forming galaxies. We find that the z = 0.8 galaxies occupy a similar locus of attenuation as a function of magnitude, mass, and star formation rate (SFR) as a comparison sample drawn from the SDSS DR4. Both the results from the individual z = 0.8 galaxies and from the stacked spectra show consistency with the mass-attenuation and SFR-attenuation relations found in the local universe, indicating that these relations are also applicable at intermediate redshift.

79 ASTRONOMY AND ASTROPHYSICS↗