Science Demonstration Phase Research Articles

The Faint Optical Counterparts Catalog at 19.4 < r < 22 mag from Herschel-ATLAS SDP Field

We carry out a spectroscopic survey with Hectospec mounted on the 6.5 m Multiple Mirror Telescope to obtain redshifts of optical counterparts of the Herschel-ATLAS sources in the science demonstration phase field. Our work concerns counterparts with a magnitude range at mag. The observation covers two pointings. In total, we newly obtain redshifts of 158 potential counterparts. Among these counterparts, there are 59 high reliability counterparts. For these high reliability counterparts, we do a SED fitting using MAGPHYS with our spectroscopic redshifts and five Herschel band fluxes, and discover 46 new luminous infra-red galaxies and 10 new ultra luminous infra-red galaxies, including two infra-red QSOs.

Estimating photometric redshifts for X-ray sources in the X-ATLAS field using machine-learning techniques

We present photometric redshifts for 1,031 X-ray sources in the X-ATLAS field, using the machine learning technique TPZ (Carrasco Kind & Brunner 2013). X-ATLAS covers 7.1 deg2 observed with the XMM-Newton within the Science Demonstration Phase (SDP) of the H-ATLAS field, making it one of the largest contiguous areas of the sky with both XMMNewton and Herschel coverage. All of the sources have available SDSS photometry while 810 have additionally mid-IR and/or near-IR photometry. A spectroscopic sample of 5,157 sources primarily in the XMM/XXL field, but also from several X-ray surveys and the SDSS DR13 redshift catalogue, is used for the training of the algorithm. Our analysis reveals that the algorithm performs best when the sources are split, based on their optical morphology, into point-like and extended sources. Optical photometry alone is not enough for the estimation of accurate photometric redshifts, but the results greatly improve when, at least, mid-IR photometry is added in the training process. In particular, our measurements show that the estimated photometric redshifts for the X-ray sources of the training sample, have a normalized absolute median deviation, n_mad=0.06, and the percentage of outliers, eta=10-14 percent, depending on whether the sources are extended or point-like. Our final catalogue contains photometric redshifts for 933 out of the 1,031 X-ray sources with a median redshift of 0.9.

TheXMM-Newtonsurvey in the H-ATLAS field

Wide area X-ray and far infrared surveys are a fundamental tool to investigate the link between AGN growth and star formation, especially in the low-redshift universe (z<1). The Herschel Terahertz Large Area survey (H-ATLAS) has covered 550 deg^2 in five far-infrared and sub-mm bands, 16 deg^2 of which have been presented in the Science Demonstration Phase (SDP) catalogue. Here we introduce the XMM-Newton observations in H-ATLAS SDP area, covering 7.1 deg^2 with flux limits of 2e-15, 6e-15 and 9e-15 erg/s/cm^2 in the 0.5--2, 0.5--8 and 2--8 keV bands, respectively. We present the source detection and the catalogue, which includes 1700, 1582 and 814 sources detected by Emldetect in the 0.5--8, 0.5--2 and 2--8 keV bands, respectively; the number of unique sources is 1816. We extract spectra and derive fluxes from power-law fits for 398 sources with more than 40 counts in the 0.5--8 keV band. We compare the best-fit fluxes with the catalogue ones, obtained by assuming a common photon index of Gamma=1.7; we find no bulk difference between the fluxes, and a moderate dispersion of s=0.33 dex. Using wherever possible the fluxes from the spectral fits, we derive the 2--10 keV LogN-LogS, which is consistent with a Euclidean distribution. Finally, we release computer code for the tools developed for this project.

Dust temperature maps of the Galactic plane: TheHerschelspectral energy distribution fitting with Cloudy predictions

Context. Dust grains absorb the interstellar far ultra-violet and visible photons and re-emit them in far-infrared (FIR) wavebands. The dust FIR continuum can be predicted by a grid of models using various values of the interstellar radiation field. Aims. We analyze the dust continuum emission in two Hi-GAL science-demonstration phase (SDP) fields using both the radiative transfer code, Cloudy, and the DustEM dust model, to explore the effect of radiative transfer on dust temperature. The 500 μm submillimeter excess emission and the very small grain (VSG) contribution to the 70 μm intensity are investigated by spectral energy distribution (SED) fitting using the Cloudy model. Methods. By comparing the observation with the model prediction, we derive dust temperature maps of the two SDP fields by fitting the dust SED with 4-band data (SPIRE bands plus PACS 160 μm) using both Cloudy and DustEM models. Considering radiative transfer and grain physics simultaneously, we investigate the existence of a 500 μm excess and estimate the VSG contribution to the 70 μm intensity by fitting the dust SED with 3-band data (160, 250, and 350 μm) and 5-band data (SPIRE and PACS bands), respectively. Results. We confirm that the field with star formation activities have a higher temperature (18.7 ± 0.9 K) than the quiescent region (15.2±0.6 K). We find that the radiative transfer affects the FIR SED of the SDP fields and results in a higher temperature distribution than the dust-only model fit. There is no significant detection of a 500 μm excess in the two SDP fields. The relative contribution from the VSGs to the 70 μm intensity can be up to 50%.

Herschel ★-ATLAS/GAMA: the environmental density of far-infrared bright galaxies at z ≤ 0.5

We compare the environmental and star formation properties of far-infrared detected and non--far-infrared detected galaxies out to $z \sim0.5$. Using optical spectroscopy and photometry from the Galaxy And Mass Assembly (GAMA) and Sloan Digital Sky Survey (SDSS), with far-infrared observations from the {\em Herschel}-ATLAS Science Demonstration Phase (SDP), we apply the technique of Voronoi Tessellations to analyse the environmental densities of individual galaxies. Applying statistical analyses to colour, $r-$band magnitude and redshift-matched samples, we show there to be a significant difference at the 3.5$\sigma$ level between the normalized environmental densities of these two populations. This is such that infrared emission (a tracer of star formation activity) favours underdense regions compared to those inhabited by exclusively optically observed galaxies selected to be of the same $r-$band magnitude, colour and redshift. Thus more highly star-forming galaxies are found to reside in the most underdense environments, confirming previous studies that have proposed such a correlation. However, the degeneracy between redshift and far-infrared luminosity in our flux-density limited sample means that we are unable to make a stronger statement in this respect. We then apply our method to synthetic light cones generated from semi-analytic models, finding that over the whole redshift distribution the same correlations between star-formation rate and environmental density are found.

A Comparison of Star Formation Rate Indicators for Galaxies

With the multi-wavelength data from UV to sub-millimeter in the region of H-ATLAS (Herschel Astrophysical Terahertz Large Area Survey) Science Demonstration Phase (SDP), in combination with the population synthesis model and dust model, the total infrared luminosities of the galaxies were calculated. On this basis, for respectively the strong and weak star-forming galaxies, we studied the differences in the star formation rates calculated by the UV luminosity, infrared luminosity and Hα line, as well as the intrinsic physical origin of such differences. It was found that for the galaxies of strong star-formation activity, the 3 kinds of star formation rate indicators give the basically consistent results with a small dispersion. But at the end of high star formation rate, the star formation rate calculated by the UV luminosity is slightly smaller than that calculated by the Hα-line flux; at the end of low star formation rate, the UV indicator tends to be greater than the Hα indicator; and at both ends, the infrared indicator and Hα indicator have no significant difference. For the weak star-forming galaxies, significant differences exist among the 3 kinds of indicators, and there is a rather large dispersion. The dispersions and systematic difference of the star formation rates calculated by the UV luminosity and Hα line increase with the galactic age and mass. The main cause for the increased systematic difference is that when the extinction of an weak star-forming galaxy is calibrated by its UV continuum spectral slope β, the UV extinction of the galaxy is overestimated, it makes the UV luminosity tends to be large after the extinction correction. In addition, the star formation rates (Hα) of weak starforming galaxies in the MPA/JHU (Max Planck Institute for Astrophysics/Johns Hopkins University) database are generally less than the real values.

A Hi-GAL study of the high-mass star-forming region G29.96–0.02

Context. G29.96-0.02 is a high-mass star-forming cloud observed at 70, 160, 250, 350, and 500 microns as part of the Herschel survey of the Galactic Plane during the Science Demonstration Phase. Aims. We wish to conduct a far-infrared study of the sources associated with this star-forming region by estimating their physical properties and evolutionary stage, and investigating the clump mass function, the star formation efficiency and rate in the cloud. Methods. We have identified the Hi-GAL sources associated with the cloud, searched for possible counterparts at centimeter and infrared wavelengths, fitted their spectral energy distribution and estimated their physical parameters. Results. A total of 198 sources have been detected in all 5 Hi-GAL bands, 117 of which are associated with 24 microns emission and 87 of which are not associated with 24 microns emission. We called the former sources 24 microns-bright and the latter ones 24 microns-dark. The [70-160] color of the 24 microns-dark sources is smaller than that of the 24 microns-bright ones. The 24 microns-dark sources have lower L_bol and L_bol/M_env than the 24 microns-bright ones for similar M_env, which suggests that they are in an earlier evolutionary phase. The G29-SFR cloud is associated with 10 NVSS sources and with extended centimeter continuum emission well correlated with the 70 microns emission. Most of the NVSS sources appear to be early B or late O-type stars. The most massive and luminous Hi-GAL sources in the cloud are located close to the G29-UC region, which suggests that there is a privileged area for massive star formation towards the center of the G29-SFR cloud. Almost all the Hi-GAL sources have masses well above the Jeans mass but only 5% have masses above the virial mass, which indicates that most of the sources are stable against gravitational collapse. The sources with M_env > M_virial and that ...

On the shape of the mass-function of dense clumps in the Hi-GAL fields

Stars form in dense, dusty clumps of molecular clouds, but little is known about their origin and evolution. In particular, the relationship between the mass distribution of these clumps (also known as the "clump mass function", or CMF) and the stellar initial mass function (IMF), is still poorly understood. In order to discern the "true" shape of the CMF and to better understand how the CMF may evolve toward the IMF, large samples of bona-fide pre- and proto-stellar clumps are required. The sensitive observations of the Herschel Space Observatory (HSO) are now allowing us to look at large clump populations in various clouds with different physical conditions. We analyse two fields in the Galactic plane mapped by HSO during its science demonstration phase, as part of the more complete and unbiased Herschel infrared GALactic Plane Survey (Hi-GAL). These fields undergo a source-extraction and flux-estimation pipeline, which allows us to obtain a sample with thousands of clumps. Starless and proto-stellar clumps are separated using both color and positional criteria to find those coincident with MIPS 24 micron sources. We describe the probability density functions of the power-law and lognormal models that are used to fit the CMFs, and we then find their best-fit parameters. For the lognormal model we apply several statistical techniques to the data and compare their results. The CMFs of the two SDP fields show very similar shapes, but very different mass scales. This similarity is confirmed by the values of the best-fit parameters of either the power-law or lognormal model. The power-law model leads to almost identical CMF slopes, whereas the lognormal model shows that the CMFs have similar widths. The similar CMF shape but different mass scale represents an evidence that the overall process of star formation in the two regions is very different.

Far-Infrared and submillimeter properties of SDSS galaxies in the Herschel ATLAS science demonstration phase field

Using the Herschel ATLAS science demonstration phase data cross-identified with SDSS DR7 spectra, we select 297 galaxies with F250μm > 5σ. The sample galaxies are classified into five morphological types, and more than 40% of the galaxies are peculiar/compact galaxies. The peculiar galaxies show higher far-infrared/submillimeter luminosity-to-mass ratios than the other types. We perform and analyze the correlations of far-infrared/submillimeter and Hα luminosities for different morphological types and different spectral types. The Spearman rank coefficient decreases and the scatter increases with the wavelength increasing from 100 μm to 500 μm. We conclude that a single Herschel SPIRE band is not good for tracing star formation activities in galaxies. AGNs contribute less to the far-infrared/submillimeter luminosities and do not show a difference from star-forming galaxies. However, the earlier type galaxies present significant deviations from the best fit of star-forming galaxies.

Herschel-ATLAS/GAMA: spatial clustering of low-redshift submm galaxies

We have measured the clustering properties of low-redshift (z < 0.3) sub-mm galaxies detected at 250 micron in the Herschel-ATLAS Science Demonstration Phase (SDP) field. We selected a sample for which we have high-quality spectroscopic redshifts, obtained from reliably matching the 250-micron sources to a complete (for r < 19.4) sample of galaxies from the GAMA database. Both the angular and spatial clustering strength are measured for all z < 0.3 sources as well as for five redshift slices with thickness delta z=0.05 in the range 0.05 < z < 0.3. Our measured spatial clustering length r_0 is comparable to that of optically-selected, moderately star-forming (blue) galaxies: we find values around 5 Mpc. One of the redshift bins contains an interesting structure, at z = 0.164.

MEASUREMENTS OF CO REDSHIFTS WITH Z-SPEC FOR LENSED SUBMILLIMETER GALAXIES DISCOVERED IN THE H-ATLAS SURVEY

We present new observations from Z-Spec, a broadband 185-305 GHz spectrometer, of five sub-millimeter bright lensed sources selected from the Herschel Astrophysical Terahertz Large Area Survey (H-ATLAS) science demonstration phase (SDP) catalog. We construct a redshift finding algorithm using combinations of the signal-to-noise of all the lines falling in the Z-Spec bandpass to determine redshifts with high confidence, even in cases where the signal-to-noise in individual lines is low. We measure the dust continuum in all sources and secure CO redshifts for four out of five (z~1.5-3). In one source, SDP.17, we tentatively identify two independent redshifts and a water line, confirmed at z=2.308. Our sources have properties characteristic of dusty starburst galaxies, with magnification-corrected star formation rates of 10^(2-3) M_sun/yr. Lower limits for the dust masses (~a few 10^8 M_sun) and spatial extents (~1 kpc equivalent radius) are derived from the continuum spectral energy distributions, corresponding to dust temperatures between 54 and 69K. In the LTE approximation, we derive relatively low CO excitation temperatures (< 100 K) and optical depths (tau<1). Performing a non-LTE excitation analysis using RADEX, we find that the CO lines measured by Z-Spec (from J=4->3 to 10->9, depending on the galaxy) localize the best solutions to either a high-temperature / low-density region, or a low-temperature / high-density region near the LTE solution, with the optical depth varying accordingly. Observations of additional CO lines, CO(1-0) in particular, are needed to constrain the non-LTE models.

Molecular Gas and Star-formation in Selected H-ATLAS SDP Lensed SMGs

AbstractWe present detections of spatially resolved CO(J = 2→1) and CO(J = 3→2) emission, respectively, from the lensed submillimeter (submm) galaxies (SMGs), ID 9 (z = 1.577) and ID 17b (z = 2.308), found in the Science Demonstration Phase (SDP) of the Herschel Astrophysical Terahertz Large Area Survey (H-ATLAS, www.h-atlas.org). The detections were obtained using the Combined Array for Research in Millimeter-wave Astronomy (CARMA, www.mmarray.org) and confirm redshifts of the lensed galaxies. We exploit the CARMA data together with existing high-J observations, to determine, among other physical properties of the lensed SMGs, the CO line luminosities, brightness temperature ratios, gas masses, and spatial sizes.

The rapid evolution of dust content in galaxies over the last five billion years

AbstractThe Herschel-ATLAS (H-ATLAS) will provide an unrivalled sample of galaxies, probing the normal star-forming submillimetre population of galaxies for the first time. Here, we exploit the Science Demonstration Phase (SDP) data to model the evolution of the interstellar content of galaxies in recent history. The most massive H-ATLAS galaxies show a large increase in the dust content five billion years ago compared to the present epoch. These observations are difficult to explain using standard dust models, one possibility could be contributions from a non-stellar source of dust e.g. grain growth in dense clouds; this would imply that less than 10% of dust would be condensed in stellar atmospheres. Alternatively, an initial mass function which becomes top heavy at high star formation rate densities could also explain this discrepancy.

HERSCHEL-ATLAS: TOWARD A SAMPLE OF ∼1000 STRONGLY LENSED GALAXIES

While the selection of strongly lensed galaxies with 500{\mu}m flux density S_500>100 mJy has proven to be rather straightforward (Negrello et al. 2010), for many applications it is important to analyze samples larger than the ones obtained when confining ourselves to such a bright limit. Moreover, only by probing to fainter flux densities is possible to exploit strong lensing to investigate the bulk of the high-z star-forming galaxy population. We describe HALOS (the Herschel-ATLAS Lensed Objects Selection), a method for efficiently selecting fainter candidate strongly lensed galaxies, reaching a surface density of ~1.5-2 deg^-2, i.e. a factor of about 4 to 6 higher than that at the 100 mJy flux limit. HALOS will allow the selection of up to ~1000 candidate strongly lensed galaxies (with amplifications \mu>2) over the full H-ATLAS survey area. Applying HALOS to the H-ATLAS Science Demonstration Phase field (~14.4 deg^2) we find 31 candidate strongly lensed galaxies, whose candidate lenses are identified in the VIKING near-infrared catalog. Using the available information on candidate sources and candidate lenses we tentatively estimate a ~72% purity of the sample. The redshift distribution of the candidate lensed sources is close to that reported for most previous surveys for lensed galaxies, while that of candidate lenses extends to substantially higher redshifts than found in the other surveys. The counts of candidate strongly lensed galaxies are also in good agreement with model predictions (Lapi et al. 2011). Even though a key ingredient of the method is the deep near-infrared VIKING photometry, we show that H-ATLAS data alone allow the selection of a similarly deep sample of candidate strongly lensed galaxies with an efficiency close to 50%; a slightly lower surface density (~1.45 deg^-2) can be reached with a ~70% efficiency.

HERSCHEL-ATLAS GALAXY COUNTS AND HIGH-REDSHIFT LUMINOSITY FUNCTIONS: THE FORMATION OF MASSIVE EARLY-TYPE GALAXIES

Exploiting the Herschel-ATLAS Science Demonstration Phase (SDP) survey data, we have determined the luminosity functions (LFs) at rest-frame wavelengths of 100 and 250 micron and at several redshifts z>1, for bright sub-mm galaxies with star formation rates (SFR) >100 M_sun/yr. We find that the evolution of the comoving LF is strong up to z~2.5, and slows down at higher redshifts. From the LFs and the information on halo masses inferred from clustering analysis, we derived an average relation between SFR and halo mass (and its scatter). We also infer that the timescale of the main episode of dust-enshrouded star formation in massive halos (M_H>3*10^12 M_sun) amounts to ~7*10^8 yr. Given the SFRs, which are in the range 10^2-10^3 M_sun/yr, this timescale implies final stellar masses of order of 10^11-10^12 M_sun. The corresponding stellar mass function matches the observed mass function of passively evolving galaxies at z>1. The comparison of the statistics for sub-mm and UV selected galaxies suggests that the dust-free, UV bright phase, is >10^2 times shorter than the sub-mm bright phase, implying that the dust must form soon after the onset of star formation. Using a single reference Spectral Energy Distribution (SED; the one of the z~2.3 galaxy SMM J2135-0102), our simple physical model is able to reproduce not only the LFs at different redshifts > 1 but also the counts at wavelengths ranging from 250 micron to ~1 mm. Owing to the steepness of the counts and their relatively broad frequency range, this result suggests that the dispersion of sub-mm SEDs of z>1 galaxies around the reference one is rather small.

HerMES: point source catalogues from deep Herschel-SPIRE observations★

We describe the generation of single-band point source catalogues from submillimetre Herschel-SPIRE observations taken as part of the Science Demonstration Phase of the Herschel Multi-tiered Extragalactic Survey (HerMES). Flux densities are found by means of peak-finding and the fitting of a Gaussian point-response function. With highly-confused images, careful checks must be made on the completeness and flux density accuracy of the detected sources. This is done by injecting artificial sources into the images and analysing the resulting catalogues. Measured flux densities at which 50 per cent of injected sources result in good detections at (250, 350, 500) {\mu}m range from (11.6, 13.2, 13.1) mJy to (25.7, 27.1, 35.8) mJy, depending on the depth of the observation (where a `good' detection is taken to be one with positional offset less than one full-width half-maximum of the point-response function, and with the measured flux density within a factor of 2 of the flux density of the injected source). This paper acts as a reference for the 2010 July HerMES public data release.

Herschel★-ATLAS: rapid evolution of dust in galaxies over the last 5 billion years

We present the first direct and unbiased measurement of the evolution of the dust mass function of galaxies over the past 5 billion years of cosmic history using data from the Science Demonstration Phase of the Herschel-ATLAS. The sample consists of galaxies selected at 250{\mu}m which have reliable counterparts from SDSS at z < 0.5, and contains 1867 sources. Dust masses are calculated using both a single temperature grey-body model for the spectral energy distribution and also using a model with multiple temperature components. The dust temperature for either model shows no trend with redshift. Splitting the sample into bins of redshift reveals a strong evolution in the dust properties of the most massive galaxies. At z = 0.4 - 0.5, massive galaxies had dust masses about five times larger than in the local Universe. At the same time, the dust-to-stellar mass ratio was about 3-4 times larger, and the optical depth derived from fitting the UV-sub-mm data with an energy balance model was also higher. This increase in the dust content of massive galaxies at high redshift is difficult to explain using standard dust evolution models and requires a rapid gas consumption timescale together with either a more top-heavy IMF, efficient mantle growth, less dust destruction or combinations of all three. This evolution in dust mass is likely to be associated with a change in overall ISM mass, and points to an enhanced supply of fuel for star formation at earlier cosmic epochs.

Data reduction pipeline for the Hi-GAL survey

We present the data reduction pipeline for the Hi-GAL survey. Hi-GAL is a key project of the Herschel satellite which is mapping the inner part of the Galactic plane (|l| <= 70\cdot and |b| <= 1\cdot), using 2 PACS and 3 SPIRE frequency bands, from 70{\mu}m to 500{\mu}m. Our pipeline relies only partially on the Herschel Interactive Standard Environment (HIPE) and features several newly developed routines to perform data reduction, including accurate data culling, noise estimation and minimum variance map-making, the latter performed with the ROMAGAL algorithm, a deep modification of the ROMA code already tested on cosmological surveys. We discuss in depth the properties of the Hi-GAL Science Demonstration Phase (SDP) data.

MODELING OF THE HERMES SUBMILLIMETER SOURCE LENSED BY A DARK MATTER DOMINATED FOREGROUND GROUP OF GALAXIES

We present the results of a gravitational lensing analysis of the bright $\zs=2.957$ sub-millimeter galaxy (SMG), HERMES J105751.1+573027 found in {\it Herschel}/SPIRE Science Demonstration Phase data from the Herschel Multi-tiered Extragalactic Survey (HerMES) project. The high resolution imaging available in optical and Near-IR channels, along with CO emission obtained with the Plateau de Bure Interferometer, allow us to precisely estimate the intrinsic source extension and hence estimate the total lensing magnification to be $\mu=10.9\pm 0.7$. We measure the half-light radius $R_{\rm eff}$ of the source in the rest-frame Near-UV and $V$ bands that characterize the unobscured light coming from stars and find $R_{\rm eff,*}= [2.0 \pm 0.1]$ kpc, in good agreement with recent studies on the Submillimeter Galaxy population. This lens model is also used to estimate the size of the gas distribution ($R_{\rm eff,gas}= [1.1\pm0.5]$) kpc by mapping back in the source plane the CO (J=5-4) transition line emission. The lens modeling yields a relatively large Einstein radius $R_{\rm Ein}= 4\farcs10 \pm 0\farcs02$, corresponding to a deflector velocity dispersion of [$483\pm 16] \,\kms$. This shows that HERMES J105751.1+573027 is lensed by a {\it galaxy group-size} dark matter halo at redshift $\zl\sim 0.6$. The projected dark matter contribution largely dominates the mass budget within the Einstein radius with $f_{\rm dm}(<R_{\rm Ein})\sim 80%$. This fraction reduces to $f_{\rm dm}(<R_{\rm eff,G1}\simeq 4.5\kpc)\sim 47%$ within the effective radius of the main deflecting galaxy of stellar mass $M_{\rm *,G1}=[8.5\pm 1.6] \times 10^{11}\msun$. At this smaller scale the dark matter fraction is consistent with results already found for massive lensing ellipticals at $z\sim0.2$ from the SLACS survey.

Herschel-ATLAS: counterparts from the ultraviolet-near-infrared in the science demonstration phase catalogue★

We present a technique to identify optical counterparts of 250 um-selected sources from the Herschel-ATLAS survey. Of the 6621 250 um > 32 mJy sources in our science demonstration catalogue we find that ~60 percent have counterparts brighter than r=22.4 mag in the Sloan Digital Sky Survey. Applying a likelihood ratio technique we are able to identify 2423 of the counterparts with a reliability R > 0.8. This is approximately 37 percent of the full 250 micron catalogue. We have estimated photometric redshifts for each of these 2423 reliable counterparts, while 1099 also have spectroscopic redshifts collated from several different sources, including the GAMA survey. We estimate the completeness of identifying counterparts as a function of redshift, and present evidence that 250 um-selected Herschel-ATLAS galaxies have a bimodal redshift distribution. Those with reliable optical identifications have a redshift distribution peaking at z ~ 0.25 +/- 0.05, while sub-mm colours suggest that a significant fraction with no counterpart above the r-band limit have z > 1. We also suggest a method for selecting populations of strongly-lensed high redshift galaxies. Our identifications are matched to UV--NIR photometry from the GAMA survey, and these data are available as part of the Herschel-ATLAS public data release.