Photo-z Codes Research Articles

Inferring More from Less: Prospector as a Photometric Redshift Engine in the Era of JWST

The advent of the James Webb Space Telescope (JWST) signals a new era in exploring galaxies in the high-z universe. Current and upcoming JWST imaging will potentially detect galaxies at z ∼ 20, creating a new urgency in the quest to infer accurate photometric redshifts (photo-z) for individual galaxies from their spectral energy distributions, as well as masses, ages, and star formation rates. Here we illustrate the utility of informed priors encoding previous observations of galaxies across cosmic time in achieving these goals. We construct three joint priors encoding empirical constraints of redshifts, masses, and star formation histories in the galaxy population within the Prospector Bayesian inference framework. In contrast with uniform priors, our model breaks an age–mass–redshift degeneracy, and thus reduces the mean bias error in masses from 0.3 to 0.1 dex, and in ages from 0.6 to 0.2 dex in tests done on mock JWST observations. Notably, our model recovers redshifts at least as accurately as the state-of-the-art photo-z code EAzY in deep JWST fields, but with two advantages: tailoring a model based on a particular survey is rendered mostly unnecessary given well-motivated priors; obtaining joint posteriors describing stellar, active galactic nuclei, gas, and dust contributions becomes possible. We can now confidently use the joint distribution to propagate full non-Gaussian redshift uncertainties into inferred properties of the galaxy population. This model, “Prospector-β,” is intended for fitting galaxy photometry where the redshift is unknown, and will be instrumental in ensuring the maximum science return from forthcoming photometric surveys with JWST. The code is made publicly available online as a part of Prospector 9 9 The version used in this work corresponds to the state of the Git repository at commit https://github.com/bd-j/prospector/commit/820ad72363a1f9c22cf03610bfe6e361213385cd..

The PAU survey: estimating galaxy photometry with deep learning

ABSTRACT With the dramatic rise in high-quality galaxy data expected from Euclid and Vera C. Rubin Observatory, there will be increasing demand for fast high-precision methods for measuring galaxy fluxes. These will be essential for inferring the redshifts of the galaxies. In this paper, we introduce Lumos, a deep learning method to measure photometry from galaxy images. Lumos builds on BKGnet, an algorithm to predict the background and its associated error, and predicts the background-subtracted flux probability density function. We have developed Lumos for data from the Physics of the Accelerating Universe Survey (PAUS), an imaging survey using a 40 narrow-band filter camera (PAUCam). PAUCam images are affected by scattered light, displaying a background noise pattern that can be predicted and corrected for. On average, Lumos increases the SNR of the observations by a factor of 2 compared to an aperture photometry algorithm. It also incorporates other advantages like robustness towards distorting artefacts, e.g. cosmic rays or scattered light, the ability of deblending and less sensitivity to uncertainties in the galaxy profile parameters used to infer the photometry. Indeed, the number of flagged photometry outlier observations is reduced from 10 to 2 per cent, comparing to aperture photometry. Furthermore, with Lumos photometry, the photo-z scatter is reduced by ≈10 per cent with the Deepz machine-learning photo-z code and the photo-z outlier rate by 20 per cent. The photo-z improvement is lower than expected from the SNR increment, however, currently the photometric calibration and outliers in the photometry seem to be its limiting factor.

Spectroscopic and Photometric Redshift Estimation by Neural Networks for the China Space Station Optical Survey (CSS-OS)

Abstract The estimation of spectroscopic and photometric redshifts (spec-z and photo-z) is crucial for future cosmological surveys. It can directly affect several powerful measurements of the universe, such as weak lensing and galaxy clustering. In this work, we explore the accuracies of spec-z and photo-z that can be obtained by the China Space Station Optical Surveys, which is a next-generation space survey, using a neural network. The one-dimensional Convolutional Neural Networks and Multi-Layer Perceptron (MLP, the simplest form of an artificial neural network) are employed to derive spec-z and photo-z, respectively. The mock spectral and photometric data used for training and testing the networks are generated based on the COSMOS catalog. The networks have been trained with noisy data by creating Gaussian random realizations to reduce the statistical effects, resulting in a similar redshift accuracy for data with both high and low signal-to-noise ratios. The probability distribution functions of the predicted redshifts are also derived via Gaussian random realizations of the testing data, and then the best-fit redshifts and 1σ errors also can be obtained. We find that our networks can provide excellent redshift estimates with accuracies of ∼0.001 and 0.01 on spec-z and photo-z, respectively. Compared to existing photo-z codes, our MLP has a similar accuracy but is more efficient in the training process. The fractions of catastrophic redshifts or outliers can be dramatically suppressed compared to the ordinary template-fitting method. This indicates that the neural network method is feasible and powerful for spec-z and photo-z estimations in future cosmological surveys.

MAGPHYS+photo-z: Constraining the Physical Properties of Galaxies with Unknown Redshifts

Abstract We present an enhanced version of the multiwavelength spectral modeling code MAGPHYS that allows the estimation of galaxy photometric redshift and physical properties (e.g., stellar mass, star formation rate, dust attenuation) simultaneously, together with robust characterization of their uncertainties. The self-consistent modeling over ultraviolet to radio wavelengths in MAGPHYS+photo-z is unique compared to standard photometric redshift codes. The broader wavelength consideration is particularly useful for breaking certain degeneracies in color versus redshift for dusty galaxies with limited observer-frame ultraviolet and optical data (or upper limits). We demonstrate the success of the code in estimating redshifts and physical properties for over 4000 infrared-detected galaxies at 0.4 < z < 6.0 in the COSMOS field with robust spectroscopic redshifts. We achieve high photo-z precision ( ), high accuracy (i.e., minimal offset biases; median(Δz/(1 + z spec)) ≲ 0.02), and low catastrophic failure rates (η ≃ 4%) over all redshifts. Interestingly, we find that a weak 2175 Å absorption feature in the attenuation curve models is required to remove a subtle systematic z phot offset ( ) that occurs when this feature is not included. As expected, the accuracy of derived physical properties in MAGPHYS+photo-z decreases strongly as redshift uncertainty increases. The all-in-one treatment of uncertainties afforded with this code is beneficial for accurately interpreting physical properties of galaxies in large photometric data sets. Finally, we emphasize that MAGPHYS+photo-z is not intended to replace existing photo-z codes, but rather offers flexibility to robustly interpret physical properties when spectroscopic redshifts are unavailable. The MAGPHYS+photo-z code is publicly available online.

Testing photometric redshift measurements with filter definition of the Chinese Space Station Optical Survey (CSS-OS)

The Chinese Space Station Optical Survey (CSS-OS) is a major science project of the Space Application System of the China Manned Space Program. This survey is planned to perform both photometric imaging and slitless spectroscopic observations, and it will focus on different cosmological and astronomical goals. Most of these goals are tightly dependent on the accuracy of photometric redshift (photo-z) measurement, especially for the weak gravitational lensing survey as a main science driver. In this work, we assess if the current filter definition can provide accurate photo-z measurement to meet the science requirement. We use the COSMOS galaxy catalog to create a mock catalog for the CSS-OS. We compare different photo-z codes and fitting methods that using the spectral energy distribution (SED) template-fitting technique, and choose to use a modified LePhare code in photo-z fitting process. Then we investigate the CSS-OS photo-z accuracy in certain ranges of filter parameters, such as band position, width, and slope. We find that the current CSS-OS filter definition can achieve reasonably good photo-z results with sigma_z~0.02 and outlier fraction ~3%.

Quasar Photometric Redshifts and Candidate Selection: A New Algorithm Based on Optical and Mid-infrared Photometric Data

Abstract We present a new algorithm to estimate quasar photometric redshifts (photo-zs), by considering the asymmetries in the relative flux distributions of quasars. The relative flux models are built with multivariate Skew-t distributions in the multidimensional space of relative fluxes as a function of redshift and magnitude. For 151,392 quasars in the SDSS, we achieve a photo-z accuracy, defined as the fraction of quasars with the difference between the photo-z z p and the spectroscopic redshift z s , within 0.1, of 74%. Combining the WISE W1 and W2 infrared data with the SDSS data, the photo-z accuracy is enhanced to 87%. Using the Pan-STARRS1 or DECaLS photometry with WISE W1 and W2 data, the photo-z accuracies are 79% and 72%, respectively. The prior probabilities as a function of magnitude for quasars, stars, and galaxies are calculated, respectively, based on (1) the quasar luminosity function, (2) the Milky Way synthetic simulation with the Besançon model, and (3) the Bayesian Galaxy Photometric Redshift estimation. The relative fluxes of stars are obtained with the Padova isochrones, and the relative fluxes of galaxies are modeled through galaxy templates. We test our classification method to select quasars using the DECaLS g, r, z, and WISE W1 and W2 photometry. The quasar selection completeness is higher than 70% for a wide redshift range , and a wide magnitude range mag. Our photo-z regression and classification method has the potential to extend to future surveys. The photo-z code will be publicly available.

Photometric redshift analysis in the Dark Energy Survey Science Verification data

We present results from a study of the photometric redshift performance of the Dark Energy Survey (DES), using the early data from a Science Verification period of observations in late 2012 and early 2013 that provided science-quality images for almost 200 sq. deg. at the nominal depth of the survey. We assess the photometric redshift (photo-z) performance using about 15 000 galaxies with spectroscopic redshifts available from other surveys. These galaxies are used, in different configurations, as a calibration sample, and photo-z's are obtained and studied using most of the existing photo-z codes. A weighting method in a multidimensional colour–magnitude space is applied to the spectroscopic sample in order to evaluate the photo-z performance with sets that mimic the full DES photometric sample, which is on average significantly deeper than the calibration sample due to the limited depth of spectroscopic surveys. Empirical photo-z methods using, for instance, artificial neural networks or random forests, yield the best performance in the tests, achieving core photo-z resolutions σ68 ∼ 0.08. Moreover, the results from most of the codes, including template-fitting methods, comfortably meet the DES requirements on photo-z performance, therefore, providing an excellent precedent for future DES data sets.

First simultaneous optical/near-infrared imaging of an X-ray selected, high-redshift cluster of galaxies with GROND

The XMM-Newton Distant Cluster Project is a serendipitous survey for clusters of galaxies at redshifts z>=0.8 based on deep archival XMM-Newton observations. ... Low-significance candidate high-z clusters are followed up with the seven-channel imager GROND (Gamma-Ray Burst Optical and Near-Infrared Detector) that is mounted at a 2m-class telescope. ... The test case is XMMU J0338.7+0030, suggested to be at z~1.45+/-0.15 from the analysis of the z-H vs H colour-magnitude diagram obtained from the follow-up imaging. Later VLT-FORS2 spectroscopy enabled us to identify four members, which set this cluster at z=1.097+/-0.002. To reach a better knowledge of its galaxy population, we observed XMMU J0338.7+0030 with GROND for about 6 hr. The publicly available photo-z code le Phare was used. The Ks-band number counts of the non-stellar sources out of the 832 detected down to z'~26 AB-mag in the 3.9x4.3 square arcmin region of XMMU J0338.7+0030 imaged at all GROND bands clearly exceed those computed in deep fields/survey areas at ~20.5 - 22.5 AB-mag. The photo-z's of the three imaged spectroscopic members yield z=1.12+/-0.09. The spatial distribution and the properties of the GROND sources with a photo-z in the range 1.01 - 1.23 confirm the correspondence of the X-ray source with a galaxy over-density at a significance of at least 4.3 sigma. Candidate members that are spectro-photometrically classified as elliptical galaxies define a red locus in the i'-z' vs z' colour-magnitude diagram that is consistent with the red sequence of the cluster RDCS J0910+5422 at z=1.106. XMMU J0338.7+0030 hosts also a population of bluer late-type spirals and irregulars. The starbursts among the photometric members populate both loci, consistently with previous results. The analysis of the available data set indicates that XMMU J0338.7+0030 is a low-mass cluster (M_200 ~ 1E14 M_sun) at z=1.1. (Abridged)

Photo-zperformance for precision cosmology - II. Empirical verification1★

The success of future large scale weak lensing surveys will critically depend on the accurate estimation of photometric redshifts of very large samples of galaxies. This in turn depends on both the quality of the photometric data and the photo-z estimators. In a previous study, (Bordoloi et al. 2010) we focussed primarily on the impact of photometric quality on photo-z estimates and on the development of novel techniques to construct the N(z) of tomographic bins at the high level of precision required for precision cosmology, as well as the correction of issues such as imprecise corrections for Galactic reddening. We used the same set of templates to generate the simulated photometry as were then used in the photo-z code, thereby removing any effects of "template error". In this work we now include the effects of "template error" by generating simulated photometric data set from actual COSMOS photometry. We use the trick of simulating redder photometry of galaxies at higher redshifts by using a bluer set of passbands on low z galaxies with known redshifts. We find that "template error" is a rather small factor in photo-z performance, at the photometric precision and filter complement expected for all-sky surveys. With only a small sub-set of training galaxies with spectroscopic redshifts, it is in principle possible to construct tomographic redshift bins whose mean redshift is known, from photo-z alone, to the required accuracy of 0.002(1+z).

Photometric redshift requirements for lens galaxies in galaxy-galaxy lensing analyses

Weak gravitational lensing is a valuable probe of galaxy formation and cosmology. Here we quantify the effects of using photometric redshifts (photo-z) in galaxy-galaxy lensing, for both sources and lenses, both for the immediate goal of using galaxies with photo-z as lenses in the Sloan Digital Sky Survey (SDSS) and as a demonstration of methodology for large, upcoming weak lensing surveys that will by necessity be dominated by lens samples with photo-z. We calculate the bias in the lensing mass calibration as well as consequences for absolute magnitude (i.e., k-corrections) and stellar mass estimates, for a large sample of SDSS Data Release 8 (DR8) galaxies. The redshifts are obtained with the template based photo-z code ZEBRA on the SDSS DR8 ugriz photometry. We assemble and characterise the calibration samples (~9k spectroscopic redshifts from four surveys) to obtain photometric redshift errors and lensing biases corresponding to our full SDSS DR8 lens and source catalogues. Our tests of the calibration sample also highlight the impact of observing conditions in the imaging survey when the spectroscopic calibration covers a small fraction of its footprint; atypical imaging conditions in calibration fields can lead to incorrect conclusions regarding the photo-z of the full survey. For the SDSS DR8 catalogue, we find sigma_z/(1+z)=0.096 and 0.113 for the lens and source catalogues, with flux limits of r=21 and r=21.8, respectively. We also explore the systematic uncertainty in the lensing signal calibration when using source photo-z, and both lens and source photo-z; given the size of existing training samples, we can constrain the lensing signal calibration (and therefore the normalization of the surface mass density) to within 2 and 4 per cent, respectively. [ABRIDGED]

Unsupervised self-organized mapping: a versatile empirical tool for object selection, classification and redshift estimation in large surveys

We present an application of unsupervised machine learning – the self-organized map (SOM) – as a tool for visualizing, exploring and mining the catalogues of large astronomical surveys. Self-organization culminates in a low-resolution representation of the ‘topology’ of a parameter volume, and this can be exploited in various ways pertinent to astronomy. Using data from the Cosmological Evolution Survey (COSMOS), we demonstrate two key astronomical applications of the SOM: (i) object classification and selection, using galaxies with active galactic nuclei as an example, and (ii) photometric redshift estimation, illustrating how SOMs can be used as totally empirical predictive tools. With a training set of ∼3800 galaxies with zspec≤ 1, we achieve photometric redshift accuracies competitive with other (mainly template fitting) techniques that use a similar number of photometric bands [σ(Δz) = 0.03 with a ∼2 per cent outlier rate when using u* band to 8 m photometry]. We also test the SOM as a photo-z tool using the PHoto-z Accuracy Testing (PHAT) synthetic catalogue of Hildebrandt et al., which compares several different photo-z codes using a common input/training set. We find that the SOM can deliver accuracies that are competitive with many of the established template fitting and empirical methods. This technique is not without clear limitations, which are discussed, but we suggest it could be a powerful tool in the era of extremely large –‘petabyte’– data bases where efficient data mining is a paramount concern.

Implementation of PhotoZ under Astro-WISE

We describe the implementation of the PhotoZ code in the framework of the Astro-WISE package and as part of the Photometric Classification Server of the PanSTARRS pipeline. Both systems allow the automatic measurement of photometric redshifts for the millions of objects being observed in the PanSTARRS project or expected to be observed by future surveys like KIDS, DES or EUCLID.

A comparison of six photometric redshift methods applied to 1.5 million luminous red galaxies

We present an updated version of MegaZ-LRG (Collister et al.,(2007)) with photometric redshifts derived with the neural network method, ANNz as well as five other publicly available photo-z codes (HyperZ, SDSS, Le PHARE, BPZ and ZEBRA) for ~1.5 million Luminous Red Galaxies (LRGs) in SDSS DR6. This allows us to identify how reliable codes are relative to each other if used as described in their public release. We compare and contrast the relative merits of each code using ~13000 spectroscopic redshifts from the 2SLAQ sample. We find that the performance of each code depends on the figure of merit used to assess it. As expected, the availability of a complete training set means that the training method performs best in the intermediate redshift bins where there are plenty of training objects. Codes such as Le PHARE, which use new observed templates perform best in the lower redshift bins. All codes produce reasonable photometric redshifts, the 1-sigma scatters ranging from 0.057 to 0.097 if averaged over the entire redshift range. We also perform tests to check whether a training set from a small region of the sky such as 2SLAQ produces biases if used to train over a larger area of the sky. We conclude that this is not likely to be a problem for future wide-field surveys. The complete photometric redshift catalogue including redshift estimates and errors on these from all six methods can be found at www.star.ucl.ac.uk/~mbanerji/MegaZLRGDR6/megaz.html

PHAT: PHoto-zAccuracy Testing

Here we introduce PHAT, the PHoto-z Accuracy Testing programme, an international initiative to test and compare different methods of photo-z estimation. Two different test environments are set up, one (PHAT0) based on simulations to test the basic functionality of the different photo-z codes, and another one (PHAT1) based on data from the GOODS survey. The accuracy of the different methods is expressed and ranked by the global photo-z bias, scatter, and outlier rates. Most methods agree well on PHAT0 but produce photo-z scatters that can differ by up to a factor of two even in this idealised case. A larger spread in accuracy is found for PHAT1. Few methods benefit from the addition of mid-IR photometry. Remaining biases and systematic effects can be explained by shortcomings in the different template sets and the use of priors on the one hand and an insufficient training set on the other hand. Scatters of 4-8% in Delta_z/(1+z) were obtained, consistent with other studies. However, somewhat larger outlier rates (>7.5% with Delta_z/(1+z)>0.15; >4.5% after cleaning) are found for all codes. There is a general trend that empirical codes produce smaller biases than template-based codes. The systematic, quantitative comparison of different photo-z codes presented here is a snapshot of the current state-of-the-art of photo-z estimation and sets a standard for the assessment of photo-z accuracy in the future. The rather large outlier rates reported here for PHAT1 on real data should be investigated further since they are most probably also present (and possibly hidden) in many other studies. The test data sets are publicly available and can be used to compare new methods to established ones and help in guiding future photo-z method development. (abridged)

Mock galaxy redshift catalogues from simulations: implications for Pan-STARRS1

We describe a method for constructing mock galaxy catalogues which are well suited for use in conjunction with large photometric surveys. We use the semi-analytic galaxy formation model of Bower et al. implemented in the Millennium simulation. We apply our method to the specific case of the surveys soon to commence with PS1, the first of 4 telescopes planned for the Pan-STARRS system. PS1 has 5 photometric bands (grizy), and will carry out an all-sky 3pi survey and a medium deep survey (MDS) over 84 sq.deg. We calculate the expected magnitude limits for extended sources in the two surveys. We find that, after 3 years, the 3pi survey will have detected over 10^8 galaxies in all 5 bands, 10 million of which will lie at redshift z>0.9, while the MDS will have detected over 10^7 galaxies with 0.5 million lying at z>2. These numbers at least double if detection in the shallowest band, y is not required. We then evaluate the accuracy of photometric redshifts estimated using an off-the-shelf photo-z code. With the grizy bands alone it is possible to achieve an accuracy in the 3pi survey of Delta z/(1+z)~0.06 for 0.25<z<0.8, which could be reduced by about 15% using near infrared photometry from the UKIDDS survey, but would increase by about 25% for the deeper sample without the y band photometry. For the MDS an accuracy of Delta z/(1+z)~0.05 is achievable for 0.02<z<1.5 using grizy. A dramatic improvement in accuracy is possible by selecting only red galaxies. In this case, Delta z/(1+z)~0.02-0.04 is achievable for ~100 million galaxies at 0.4<z<1.1 in the 3pi survey and for 30 million galaxies in the MDS at 0.4<z<2. We investigate the effect of using photo-z in the estimate of the baryonic acoustic oscillation scale. We find that PS1 will achieve a similar accuracy in this estimate as a spectroscopic survey of 20 million galaxies.

A blind test of photometric redshifts on ground-based data

Aims. We analyse the relative performance of different photo-z codes in blind applications to ground-based data. Methods. We tested the codes on imaging datasets with different depths and filter coverages and compared the results to large spectroscopic catalogues. The photo-z error behaviour was analysed to select cleaner subsamples with more secure photo-z estimates. We consider Hyperz, BPZ, and the code used in the CADIS, COMBO-17, and HIROCS surveys. Results. The photo-z error estimates of the three codes do not correlate tightly with the accuracy of the photo-z's. While very large errors sometimes indicate a true catastrophic photo-z failure, smaller errors are usually not meaningful. For any given dataset, we find significant differences in redshift accuracy and outlier rates between the different codes when compared to spectroscopic redshifts. However, different codes excel in different regimes. The agreement between different sets of photo-z's is better for the subsample with secure spectroscopic redshifts than for the whole catalogue. Conclusions. Running today's photo-z codes on well-calibrated ground-based data can lead to reasonable accuracy. The actual performance on a given dataset is largely dependent on the template choice and on realistic instrumental response curves. It would be desirable to improve the photo-z error estimation for future applications so as to get a better handle on rejecting objects with grossly inaccurate photo-z's. The secure spectroscopic subsamples commonly used for assessments of photo-z accuracy may be biased toward objects for which the photo-z's are easier to estimate than for a complete flux-limited sample, resulting in very optimistic estimates. (abridged)