Hubble Deep Field Research Articles

ESO imaging survey: infrared observations of CDF-S and HDF-S

This paper presents infrared data obtained from observations carried out at the ESO 3.5m New Technology Telescope (NTT) of the Hubble Deep Field South (HDF-S) and the Chandra Deep Field South (CDF-S). These data were taken as part of the ESO Imaging Survey (EIS) program, a public survey conducted by ESO to promote follow-up observations with the VLT. In the HDF-S field the infrared observations cover an area of ~53 square arcmin, encompassing the HST WFPC2 and STIS fields, in the JHKs passbands. The seeing measured in the final stacked images ranges from 0.79 to 1.22 and the median limiting magnitudes (AB system, 2 aperture, 5sigma detection limit) are J_AB~23.0, H_AB~22.8 and K_AB~23.0 mag. Less complete data are also available in JKs for the adjacent HST NICMOS field. For CDF-S, the infrared observations cover a total area of \~100 square arcmin, reaching median limiting magnitudes (as defined above) of J_AB~23.6 and K_AB~22.7 mag. For one CDF-S field H-band data are also available. This paper describes the observations and presents the results of new reductions carried out entirely through the un-supervised, high-throughput EIS Data Reduction System and its associated EIS/MVM C++-based image processing library developed, over the past 5 years, by the EIS project and now publicly available. The paper also presents source catalogs extracted from the final co-added images which are used to evaluate the scientific quality of the survey products, and hence the performance of the software. This is done comparing the results obtained in the present work with those obtained by other authors from independent data and/or reductions carried out with different software packages and techniques. The final science-grade catalogs and co-added images are available at CDS.

A cosmic relation between extinction and star formation

Abstract We study the relation between the star formation intensity of galaxies and the extinction by dust of their emitted light. We employ a detailed statistical analysis of Hubble Deep Field North data to show a clear positive correlation between the extinction and star formation intensity at all epochs from redshift 0.4 to 6.5. The extinction evidently increases with time for a given star formation intensity, consistent with the expected increase in the metallicity with time. Our observational results are well fitted at all epochs by a double-power-law model with a fixed shape that simply shifts with redshift. The correlation between the extinction and the star formation intensity can be interpreted by combining two other trends: the correlation between the star formation rate and the gas content of galaxies, and the evolution of the dust-to-gas ratio in galaxies. If we assume that Kennicutt's observed relation for the former is valid at each redshift, then our findings imply an interesting variation in the dust-to-gas ratio in galaxies within each epoch and with time, and suggest new ways to investigate the cosmic evolution of this quantity.

Surface photometry and structure of high redshift disk galaxies in the HDF-S NICMOS field

Aims. We report on a photometric study of a sample of 22 disk galaxies in the Hubble Deep Field South NICMOS parallel field. The redshift range of the galaxies is $z=0.5{-}2.6$. Methods. We use deep NICMOS J and H band and STIS open mode images, taken as part of the HDF-S project, to construct rest-frame B -profiles and $(U-V)$ color profiles of the galaxies. Before fitting isophotes, images are deconvolved with PSF. Derived surface brightness profiles are approximated by Sersic luminosity distribution. Results. Significantly large population of disks cannot be represented by an exponential disk, but this can be well done by Sersic law, if $n n does not vary significantly with redshift. Galactic sizes decrease with redshift as $r_{\rm e}(z)/r_{\rm e}(0) = 1-0.26z$. The rest frame $(U-V)$ color shows a clear decrease at $z\approx2$, concordantly with the understanding of more intense star formation at earlier epochs. Color gradients $\Delta(U-V)/\Delta r$ are small and roughly constant at $z 2$, dominantly positive gradients appear, possibly indicating centrally concentrated star-formation. On the basis of $(U-V)$ color and chemical evolution models, the disks observed at $z\sim 2.5$ have formed between $z=3.5{-}7$. Scale radii r e of the galaxies correlate with the scale surface brightnesses $\mu_{\rm e}$ for the sample. None of the studied parameters shows clear dependence on absolute B luminosity for the galaxies.

Anglo-Australian Telescope Imaging and Microslit Spectroscopy in the Southern Hubble Deep Field

We present a deep photometric (B- and R-band) catalog and an associated spectroscopic redshift survey conducted in the vicinity of the Hubble Deep Field South. The spectroscopy yields 53 extragalactic redshifts in the range 0<z<1.4 substantially increasing the body of spectroscopic work in this field to over 200 objects. The targets are selected from deep AAT prime focus images complete to R<24 and spectroscopy is 50% complete at R=23. There is now strong evidence for a rich cluster at z\simeq 0.58 flanking the WFPC2 field which is consistent with a known absorber of the bright QSO in this field. We find that photometric redshifts of z<1 galaxies in this field based on HST data are accurate to \sigma_z/(1+z)=0.03 (albeit with small number statistics). The observations were carried out as a community service for Hubble Deep Field science, to demonstrate the first use of the `nod & shuffle' technique with a classical multi-object spectrograph and to test the use of `microslits' for ultra-high multiplex observations along with a new VPH grism and deep-depletion CCD. The reduction of this new type of data is also described.

The Multiwavelength Survey by Yale‐Chile (MUSYC): Survey Design and Deep Public UBVRI z ′ Images and Catalogs of the Extended Hubble Deep Field–South

We present UBVRIz' optical images taken with MOSAIC on the CTIO 4 m telescope of the 0.32 deg2 Extended Hubble Deep Field-South. This is one of four fields comprising the MUSYC survey, which is optimized for the study of galaxies at z = 3, active galactic nucleus (AGN) demographics, and Galactic structure. Our methods used for astrometric calibration, weighted image combination, and photometric calibration in AB magnitudes are described. We calculate corrected aperture photometry and its uncertainties and find through tests that these provide a significant improvement upon standard techniques. Our photometric catalog of 62,968 objects is complete to a total magnitude of RAB = 25, with R-band counts consistent with results from the literature. We select z 3 Lyman break galaxy (LBG) candidates from their UVR colors and find a sky surface density of 1.4 arcmin-2 and an angular correlation function w(θ) = (2.3 ± 1.0)θ-0.8, consistent with previous findings that high-redshift Lyman break galaxies reside in massive dark matter halos. Our images and catalogs are available online.

16 μm Imaging around the Hubble Deep Field–North with theSpitzerIRS

We present a pilot study of 16 μm imaging within the GOODS northern field. Observations were obtained using the "peak-up" imaging capability of the Spitzer IRS. We survey 35 arcmin2 to an average 3 σ depth of 0.075 mJy and detect 149 sources. The survey partially overlaps the area imaged at 15 μm by the ISO, and we demonstrate that our photometry and galaxy number counts are consistent with its measurements. We infer the total infrared luminosity of 16 μm detections using a comparison to local templates and find a wide range of LIR, from ~109 to 1012 L☉. Approximately of the detected sources have counterparts in the Chandra 2 Ms catalog, and we show that the hard-band (2-8 keV) detected sources are likely to have strong active galactic nucleus (AGN) contributions to the X-ray flux. The ultradeep sensitivity of Chandra implies that some X-ray detections may be purely starbursting objects. We examine the 16 to 24 μm flux ratio and conclude that it shows evidence for the detection of redshifted polycyclic aromatic hydrocarbon (PAH) emission at z ~ 0.5 and z > 0.8.

Faint Blue Objects in the Hubble Deep Field–South Revealed: White Dwarfs, Subdwarfs, and Quasars

We explore the nature of the faint blue objects in the Hubble Deep Field-South. We have derived proper motions for the point sources in the Hubble Deep Field-South using a 3 yr baseline. Combining our proper-motion measurements with spectral energy distribution fitting enabled us to identify four quasars and 42 stars, including three white dwarf candidates. Two of these white dwarf candidates, HDF-S 1444 and 895, are found to display significant proper motion, 21.1 ± 7.9 and 34.9 ± 8.0 mas yr-1, and are consistent with being thick-disk or halo white dwarfs located at ~2 kpc. The other faint blue objects analyzed by Mendez & Minniti do not show any significant proper motion and are inconsistent with being halo white dwarfs; they do not contribute to the Galactic dark matter. The observed population of stars and white dwarfs is consistent with standard Galactic models.

The Lensed Arc Production Efficiency of Galaxy Clusters: A Comparison of Matched Observed and Simulated Samples

We compare the statistical properties of giant gravitationally lensed arcs produced in matched simulated and observed cluster samples. The observed sample consists of 10 X-ray selected clusters at redshifts z ~ 0.2 imaged with HST by Smith et al. The simulated dataset is produced by lensing the Hubble Deep Field, which serves as a background source image, with 150 realizations (different projections and shifts) of five simulated z = 0.2 clusters from a LambdaCDM N-body simulation. The real and simulated clusters have similar masses, the real photometric redshift is used for each background source, and all the observational effects influencing arc detection in the real dataset, including light from cluster galaxies, are simulated in the artificial dataset. We develop, and apply to both datasets, an objective automatic arc-finding algorithm. We find consistent arc statistics in the real and in the simulated sample, with an average of ~ 1 detected giant (length to width ratio >= 10) arc per cluster and ~ 0.2 giant luminous (R<22.3 mag) arc per cluster. Thus, taking into account a realistic source population and observational effects, the clusters predicted by LambdaCDM have the same arc-production efficiency as the observed clusters. If, as suggested by other studies, there is a discrepancy between the predicted and the observed total number of arcs on the sky, it must be the result of differences between the redshift dependent cluster mass functions, and not due to differences in the lensing efficiency of the most massive clusters.

Gamma‐Ray Burst–Selected High‐Redshift Galaxies: Comparison to Field Galaxy Populations toz∼ 3

We study the internal structural properties of 37 long duration gamma-ray burst (GRB) host galaxies imaged with the Hubble Space Telescope. Fitting exponential disk and r^1/4 models to the surface brightness profiles of eight z < 1.2 bright host galaxies, we find that the disk model is slightly preferred for most hosts, although two galaxies are fit best with an r^1/4 profile. We furthermore measure the central concentrations and asymmetries of all 37 host galaxies using the CAS system, and compare with field galaxies in the Hubble Deep Field. Our first main conclusion is that GRB hosts exhibit a surprisingly high diversity of galaxy types. A significant fraction (68%) of host galaxies are situated in a region of the concentration-asymmetry diagram occupied by spirals or peculiar/merging galaxies. Twelve hosts (32%) are situated in the region occupied by elliptical galaxies, having high concentration indices. We also find some evidence for evolution in GRB host galaxy morphology, such that hosts at z > 1 have a relatively high light concentration, indicating that these systems are perhaps progenitors of massive galaxies, or are compact blue star forming galaxies. We find that GRB hosts at z > 1 are different from the general field population at z > 1 in terms of light concentration at >99.5% confidence, yet have sizes similar to the general z > 1 population. This is the opposite of the effect at z < 1 where GRB hosts are smaller than average. We argue that GRB hosts trace the starburst population at high redshift, as similarly concentrated galaxies at z > 1 are undergoing a disproportionate amount of star formation for their luminosities. Our results show that GRBs are perhaps ideal tracers of typical galaxies undergoing star formation at any epoch.

Radio Observations of the Hubble Deep Field-South Region. I. Survey Description and Initial Results

This paper is the first of a series describing the results of the Australia Telescope Hubble Deep Field–South radio survey. The survey was conducted at four wavelengths, 20, 11, 6, and 3 cm, over a 4 year period and achieves an rms sensitivity of about 10 μJy at each wavelength. We describe the observations and data reduction processes and present data on radio sources close to the center of the Hubble Deep Field–South. We discuss in detail the properties of a subset of these sources. The sources include both starburst galaxies and galaxies powered by an active galactic nucleus and range in redshift from 0.1 to 2.2. Some of them are characterized by unusually high radio-to-optical luminosities, presumably caused by dust extinction.

Radio Observations of the Hubble Deep Field-South Region. II. The 1.4 GHz Catalog and Source Counts

This paper is part of a series describing the results from the Australia Telescope Hubble Deep Field–South (ATHDFS) survey obtained with the Australia Telescope Compact Array. This survey consists of observations at 1.4, 2.5, 5.2, and 8.7 GHz, all centered on the Hubble Deep Field–South. Here we present the first results from the extended observing campaign at 1.4 GHz. A total of 466 sources have been cataloged to a local sensitivity of 5 σ (11 μJy rms). A source extraction technique is developed that (1) successfully excludes spurious sources from the final source catalogs and (2) accounts for the nonuniform noise in our image. A source catalog is presented, and the general properties of the 1.4 GHz image are discussed. We also present source counts derived from our ATHDFS 1.4 GHz catalog. Particular attention is given to ensuring that the counts are corrected for survey incompleteness and systematic effects. Our counts are consistent with other surveys (e.g., the Australia Telescope ESO Slice Project, VIRMOS, and the Phoenix Deep Field), and we find, in common with these surveys, that the HDF-N counts are systematically lower.

Looking for Life in the IR and Neighboring Spectral Regions

The paper focuses on the issues for detection of life by remote observations. The problem is divided into the detection of terrestrial planets and the detection of life on them. The problem of detection of planets is compared with the observations of the Hubble Deep Field. The difficulty of detection will likely result in many interesting discoveries before detection techniques are adequate. It is proposed that life is the activity of mutually assisting persistence processes. A persistence process is one that uses the Gibbs Free Energy of the environment for repair and development. It is pointed out that because life produces approximations to equilibrium it is intrinsically difficult to detect by remote observations. Chemical signs of life will be distinctive if the abiotic processes that might produce them are implausible, and preferably absent. The ease of remote detection of terrestrial life is contrasted with problems likely to arise for other planets. It is recommended that initial searches be focused on giant planets. The spectral region of the discovery observations will be set by technical issues which are not yet resolved for any technique. For determining whether life is present, as many spectral regions as possible should be observed.

The Assembly of Diversity in the Morphologies and Stellar Populations of High‐Redshift Galaxies

We have used deep images from the Hubble Space Telescope to study the evolution in the morphologies, sizes, stellar masses, colors, and internal color dispersion of galaxies in the Hubble Deep Field-North. We selected two galaxy samples at 0.7 ? z ? 1.4 and 1.9 ? z ? 3.0 from a near-infrared, flux-limited catalog with complete photometric and spectroscopic redshift information. At z ~ 1 the majority of galaxies with M(B) ? -20.5 have rest-frame optical morphologies of early- to mid-type Hubble-sequence galaxies, and many galaxies show strong transformations between their rest-frame UV and optical morphologies. In stark contrast, galaxies at z ~ 2.3 all have compact and irregular rest-frame optical morphologies with little difference between their rest-frame UV and optical morphologies, and with no clearly evident Hubble-sequence candidates. The mean galaxy size increases from z ~ 2.3 to z ~ 1 by roughly 40%, and the number density of galaxies larger than 3 kpc increases by a factor of ?7. The size-luminosity distribution of z ~ 1 galaxies is broadly consistent with that in the local universe, allowing for passive evolution. However, we argue that galaxies at z ~ 2.3 are not the fully formed progenitors of present-day galaxies, and they must continue to grow in both their sizes and stellar masses. We have quantified the differences in morphology by measuring the galaxies' internal UV-optical color dispersion, which constrains the amount of current star formation relative to older stellar populations. The mean and scatter in the galaxies' UV-optical total colors and internal color dispersion increase from z ~ 2.3 to z ~ 1. At z ~ 1 many galaxies with large internal color dispersion are spirals, with a few irregular and interacting systems. Few z ~ 2.3 galaxies have high internal color dispersion, and we infer that those that do are also actively merging. Using simple models, we interpret the change in the total color and internal color dispersion as evidence for the presence of older and more diverse stellar populations at z ~ 1 that are not generally present at z 2. The z ~ 2.3 galaxies do not increase their stellar diversity as rapidly as they could given basic timescale arguments and simple models. We conclude that the star formation histories of galaxies at z 2 are dominated by discrete, recurrent starbursts, which quickly homogenize the galaxies' stellar content and are possibly associated with mergers. The increase in the diversification of stellar populations by z ~ 1 implies that merger-induced starbursts occur less frequently than at higher redshifts, and more quiescent modes of star formation become the dominant mechanism. This transition in the mode of star formation coincides with the emergence of Hubble-sequence galaxies, which seems to occur around z ~ 1.4.

Spectroscopy and stellar populations of star-forming galaxies atz~ 3 in the Hubble Deep Field – South

We present results of VLT/FORS2 spectroscopy of galaxies at z~3 in the Hubble Deep Field-South (HDF-S). A sample of galaxies was drawn from the photo-z catalogue based on the HST/WFPC2 optical images and the deep near-infrared images obtained with VLT/ISAAC as a part of the FIRES project. We selected galaxies with photometric redshift between 2.5 and 4. Most of the selected galaxies are bright in rest-frame UV wavelengths and satisfy color selection criteria of Lyman break galaxies (LBGs) at z~3. The number of target galaxies with I(AB) 2 discovered by FIRES. DRGs have larger stellar masses, larger dust attenuation than our UV-luminous LBG sample, and their star formation rates are often comparable to LBGs. These trends suggest that majority of DRGs are indeed the most massive systems at the redshift and are still in the active star-forming phase. Unless the number density of DRGs is much smaller than LBGs, estimates based on UV selected sample could miss substantial part of stellar mass density at z~3.

Hubble ultra deep field object surface brightness variation

The technique of adjacent pixel brightness variation in log scale was applied to ultra deep field objects captured by the Hubble telescope. The local surface brightness fluctuation has a Gaussian–like distribution implying a random accretion of material, with collisions, as a random-walk motion. The larger fluctuation of log brightness in the mid to outer regions of some galaxies is consistent with the presence of dark matter. The absence of such fluctuation enhancement in two i -drop candidates at z ∼ 3 may mean that some dark matter is the result of nucleosynthesis in the evolution process. Large fluctuation in the B+V bands in contrast to the i +z band near the center in an i -drop candidate at z ∼ 3 was interpreted to be an AGN center. The distribution average shifts towards zero for more spiral galaxies, signifying the use of local surface brightness fluctuation distribution as a morphology parameter.

The space density of moderate-luminosity active galaxies at z = 3

Abstract We present an estimate of the space density of active galactic nuclei (AGN) at z = 3. Combining deep X-ray data with Lyman break galaxy (LBG) colour selection in the rest-frame ultraviolet makes for highly efficient identification of AGN in a narrow redshift range (z ∼ 2.5–3.5). Using Chandra data from the Groth–Westphal Strip (GWS) and the Hubble Deep Field North (HDF-N), we find a total of 15 X-ray-detected LBGs at z ∼ 3, the majority of which are unlikely to have been identified in blanket follow-up surveys of X-ray-detected objects. We find the comoving space density of moderate-luminosity AGN (MLAGN; LX = 1043–44.5 erg s−1) at z = 3 to be a factor of ∼10 higher than that of the most powerful objects. The available data are consistent with a roughly constant space density of MLAGN from z = 0.5 to 3, and they are also consistent with a mild decline in the space density above z = 1 as predicted by the luminosity-dependent density evolution models of Ueda et al. This strong AGN activity at z = 3 argues against previous suggestions that the majority of black hole accretion occurs at low redshift. A further implication of our investigation is that, as far as can be determined from current data, the majority of the AGN population at z ∼ 3 is selected by the LBG dropout technique. Although this is sensitive to AGN with an ultraviolet excess arising from an accretion disc, it predominantly identifies star-forming galaxies. A significant fraction of X-ray sources seem likely to be hosted by these more typical LBGs, further strengthening the starburst–AGN connection at high redshift.

VLT narrow-band photometry in the Lyman continuum of two galaxies atz$\mathsf{\sim3}$

We have performed narrow-band imaging observations with the Very Large Telescope, aimed at detecting the Lyman continuum (LC) flux escaping from galaxies at z~3. We do not find any significant LC flux from our sample of two galaxies in the Hubble Deep Field South, at z=3.170 and 3.275. The corresponding lower limits on the F_{1400}/F_{900} flux density (per Hz) ratio are 15.6 and 10.2 (3-sigma confidence level). After correction for the intergalactic hydrogen absorption, the resulting limits on the relative escape fraction of the LC are compared with those obtained by different approaches, at similar or lower redshifts. One of our two objects has a relative escape fraction lower than the detection reported by Steidel et al. in a composite spectrum of z~3 galaxies. A larger number of objects is required to reach a significant conclusion. Our comparison shows the potential of narrow-band imaging for obtaining the best limit on the relative escape fraction at z~3. Stacking a significant number of galaxies observed through a narrow-band filter would provide constraint on the galactic contribution to the cosmic reionization.

Hubble Space TelescopeACS Weak‐Lensing Analysis of the Galaxy Cluster RDCS 1252.9−2927 atz= 1.24

We present a weak-lensing analysis of one of the most distant massive galaxy clusters known, RDCS 1252.9-2927, at z = 1.24, using deep images from the Advanced Camera for Surveys (ACS) on board the Hubble Space Telescope (HST). By taking advantage of the depth and angular resolution of the ACS images, we detect for the first time at z > 1 a clear weak-lensing signal in both the i (F775W) and z (F850LP) filters. We measure a 5 σ signal in the i band and a 3 σ signal in the shallower z-band image. The two radial mass profiles are found to be in very good agreement with each other, and they provide a measurement of the total mass of the cluster inside a 1 Mpc radius of M(< 1 Mpc) = (8.0 ± 1.3) × 1014 M☉ in the current cosmological concordance model, h = 0.70, Ωm = 0.3, and ΩΛ = 0.7, assuming a redshift distribution of background galaxies as inferred from the Hubble Deep Field surveys. A weak-lensing signal is detected out to the boundary of our field (3' radius, corresponding to 1.5 Mpc at the cluster redshift). We detect a small offset between the centroid of the weak-lensing mass map and the brightest cluster galaxy, and we discuss the possible origin of this discrepancy. The cumulative weak-lensing radial mass profile is found to be in good agreement with the X-ray mass estimate based on Chandra and XMM-Newton observations, at least out to R500 ≃ 0.5 Mpc.

Constraints on z ≈ 10 Galaxies from the Deepest Hubble Sp a ce Telescope NICMOS Fields

We use all available fields with deep NICMOS imaging to search for J110-dropouts (H160,AB 28) at z ≈ 10. Our primary data set for this search is the two J110 + H160 NICMOS fields taken in parallel with the Advanced Camera for Surveys (ACS) Hubble Ultra Deep Field (UDF). The 5 σ limiting magnitudes were ~28.6 in J110 and ~28.5 in H160 (06 apertures). Several shallower fields were also used: J110 + H160 NICMOS frames available over the Hubble Deep Field (HDF) North, the HDF-South NICMOS parallel, and the ACS UDF (with 5 σ limiting magnitudes in J110 and H160 ranging from 27.0 to 28.2). The primary selection criterion was (J110-H160)AB > 1.8. Eleven such sources were found in all search fields using this criterion. Eight of these are clearly ruled out as credible z ≈ 10 sources, either as a result of detections (>2 σ) blueward of J110 or their colors redward of the break (H160-K ~ 1.5) (redder than 98% of lower redshift dropouts). The nature of the three remaining sources could not be determined from the data. This number appears consistent with the expected contamination from low-redshift interlopers. Analysis of the stacked images for the three candidates also suggests some contamination. Regardless of their true redshifts, the actual number of z ≈ 10 sources must be three or fewer. To assess the significance of these results, two lower redshift samples (a z ~ 3.8 B-dropout and z ~ 6 i-dropout sample) were projected to z ~ 7-13 using a (1 + z)-1 size scaling (for fixed luminosity). They were added to the image frames and the selection was repeated, giving 15.6 and 4.8 J110-dropouts, respectively. This suggests that to the limit of this probe (≈0.3L), there has been evolution from z ~ 3.8 and possibly from z ~ 6. This is consistent with the strong evolution already noted at z ~ 6 and z ~ 7.5 relative to z ~ 3-4. Even assuming that three sources from this probe are at z ≈ 10, the rest-frame continuum UV (~1500 A) luminosity density at z ~ 10 (integrated down to 0.3L) is just 0.19 times that at z ~ 3.8 (or 0.19 times, including the small effect from cosmic variance). However, if none of our sources are at z ≈ 10, this ratio has a 1 σ upper limit of 0.07.

High-resolution studies of radio sources in theHubble DeepandFlanking Fields

18 days of MERLIN data and 42 h of A-array VLA data at 1.4 GHz have been combined to image a 10-arcmin field centred on the Hubble Deep Field (HDF). This area also includes the Hubble Flanking Fields (HFF). A complete sample of 92 radio sources with S 1.4 > 40 μJy was detected using the VLA data alone and then imaged with the MERLIN+VLA combination. The combined images offer (i) higher angular resolution (synthesized beams of diameter 0.2-0.5 arcsec), (ii) improved astrometric accuracy, and (iii) improved sensitivity compared with VLA-only data. The images are amongst the most sensitive yet made at 1.4GHz, with rms noise levels of 3.3 μJy beam -1 in the 0.2-arcsec images. Virtually all the sources are resolved, with angular sizes in the range 0.2 to 3 arcsec. The central 3-arcmin square was imaged separately to search for sources down to 27 μJy. No additional sources were detected, indicating that sources fainter than 40 μJy are heavily resolved with MERLIN and must have typical angular sizes > 0.5 arcsec. Radio sources associated with compact galaxies have been used to align the HDF, the HFF and a larger CFHT optical field to the radio-based International Celestial Reference Frame. The HST optical fields have been registered to 70 per cent of the sources are starburst-type systems associated with major disc galaxies in the redshift range 0.3-1.3. Chandra detections are associated with 55 of the 92 radio sources, but their X-ray flux densities do not appear to be correlated with the radio flux densities or morphologies. The most recent submillimetre results on the HDF and HFF do not provide any unambiguous identifications with these latest radio data, except for HDF8550.1, but suggest at least three strong candidates.