The WFC3 Infrared Spectroscopic Parallel Survey uses the Hubble Space Telescope (HST) infrared grism capabilities to obtain slitless spectra of thousands of galaxies over a wide redshift range including the peak of star formation history of the universe. We select a population of very strong emission-line galaxies with rest-frame equivalent widths (EWs) higher than 200 Å. A total of 176 objects are found over the redshift range 0.35 < z < 2.3 in the 180 arcmin 2 area that we have analyzed so far. This population consists of young and low-mass starbursts with high specific star formation rates (sSFR). After spectroscopic follow-up of one of these galaxies with Keck/Low Resolution Imaging Spectrometer, we report the detection at z = 0.7 of an extremely metal-poor galaxy with 12 + log(O/H) = 7.47 ± 0.11. After estimating the active galactic nucleus fraction in the sample, we show that the high-EW galaxies have higher sSFR than normal star-forming galaxies at any redshift. We find that the nebular emission lines can substantially affect the total broadband flux density with a median brightening of 0.3 mag, with some examples of line contamination producing brightening of up to 1 mag. We show that the presence of strong emission lines in low-z galaxies can mimic the color-selection criteria used in the z ∼ 8 dropout surveys. In order to effectively remove low-redshift interlopers, deep optical imaging is needed, at least 1 mag deeper than the bands in which the objects are detected. Without deep optical data, most of the interlopers cannot be ruled out in the wide shallow HST imaging surveys. Finally, we empirically demonstrate that strong nebular lines can lead to an overestimation of the mass and the age of galaxies derived from fitting of their spectral energy distribution (SED). Without removing emission lines, the age and the stellar mass estimates are overestimated by a factor of 2 on average and up to a factor of 10 for the high-EW galaxies. Therefore, the contribution of emission lines should be systematically taken into account in SED fitting of star-forming galaxies at all redshifts.

We present evidence for strong Hα emission in galaxies with spectroscopic redshifts in the range of 3.8 < z < 5.0 over the Great Observatories Origins Deep Survey (GOODS) fields. Among 74 galaxies detected in the Spitzer IRAC 3.6 and 4.5 µm bands, more than 70 % of the galaxies show clear excess at 3.6 µm compared to the expected flux density from stellar continuum only. We provide evidence that this 3.6 µm excess is due to Hα emission redshifted into the 3.6 µm band, and classify these 3.6 µm excess galaxies to be Hα emitter (HAE) candidates. The selection of HAE candidates using an excess in broad-band filters is sensitive to objects whose rest-frame Hα equivalent width is larger than 350Å. The Hα inferred SFRs of the HAEs range between 20 and 500 M ⊙ yr −1 and is a a factor of ∼6 larger than SFRs inferred from the UV continuum. The ratio between the Hα luminosity and UV luminosity of HAEs is also on average larger than that of local starbursts. Possible reasons for such strong Hα emission in these galaxies include different dust extinction properties, young stellar population ages, extended star formation histories, low metallicity, and a top-heavy stellar initial mass function. Although the correlation between UV slope β and L Hα /L U V raises the possibility that HAEs prefer a dust extinction curve which is steeper in the UV, the most dominant factor that results in strong Hα emission appears to be star formation history. The Hα equivalent widths of HAEs are large despite their relatively old stellar population ages constrained by SED fitting, suggesting that at least 60 % of HAEs produce stars at a constant rate. Under the assumption that the gas supply is sustained, HAEs are able to produce 50 % of the stellar mass density that is encompassed in massive (M * > 10 11 M ⊙ ) galaxies at z ∼ 3. This 'strong Hα phase' of star formation plays a dominant role in galaxy growth at z ∼ 4, and they are likely progenitors of massive red galaxies at lower redshifts.

We present the WFC3 Infrared Spectroscopic Parallel (WISP) Survey. WISP is obtaining slitless, near-infrared grism spectroscopy of ∼90 independent, high-latitude fields by observing in the pure-parallel mode with the Wide Field Camera Three on the Hubble Space Telescope for a total of ∼250 orbits. Spectra are obtained with the G 102 (λ = 0.8-1.17 μm, R ∼ 210) and G 141 grisms (λ = 1.11-1.67 μm, R ∼130), together with direct imaging in the J and H bands (F110W and F140W, respectively). In the present paper, we present the first results from 19 WISP fields, covering approximately 63 arcmin 2 . For typical exposure times (∼6400 s in G 102 and ∼2700 s in G 141 ), we reach 5σ detection limits for emission lines of f ∼ 5 × 10 −17 erg s −1 cm −2 for compact objects. Typical direct imaging 5σ limits are 26.3 and 26.1 mag. (AB) in F110W and F140W, respectively. Restricting ourselves to the lines measured with the highest confidence, we present a list of 328 emission lines, in 229 objects, in a redshift range 0.3 < z < 3. The single-line emitters are likely to be a mix of Hα and [O iii]5007,4959 Å, with Hα predominating. The overall surface density of high-confidence emission-line objects in our sample is approximately 4 per arcmin 2 . These first fields show high equivalent width sources, active galactic nucleus, and post-starburst galaxies. The median observed star formation rate (SFR) of our Hα-selected sample is 4 M yr −1 . At intermediate redshifts, we detect emission lines in galaxies as faint as H 140 ∼ 25, or M R < −19, and are sensitive to SFRs down to less than 1 M yr −1 . The slitless grisms on WFC3 provide a unique opportunity to study the spectral properties of galaxies much fainter than L * at the peak of the galaxy assembly epoch.

Aims. We present an 8-band (u * , g , r , i , z , Y, J, K s ) optical to near-infrared deep photometric catalog based on the observations made with MegaCam and WIRCam at the CFHT, and compute photometric redshifts, z p in the north ecliptic pole (NEP) region. AKARI infrared satellite carried out a deep survey in the NEP region at near-to mid-infrared wavelengths. Our optical to nearinfrared catalog allows us to identify the counterparts and z p for the AKARI sources. Methods. We obtained seven-band (g , r , i , z , Y, J, K s ) imaging data, and we crossmatched them with existing u * -band data (limiting magnitude = 24.6 mag [5σ; AB]) to design the band-merged catalog. We included all z -band sources with counterparts in at least one of the other bands in the catalog. We used a template-fitting methods to compute z p for all the cataloged sources. Results. The estimated 4σ detection limits within a 1 arcsec aperture radius are 26.7, 25.9, 25.1, and 24.1 mag [AB] for the optical g , r , i , and z -bands and 23.4, 23.0, and 22.7 mag for the near-infrared Y, J, and K s -bands, respectively. There are a total of 85 797 sources in the band-merged catalog. An astrometric accuracy of this catalog determined by examining coordinate offsets with regard to 2MASS is 0.013 arcsec with a root mean square offset of 0.32 arcsec. We distinguish 5441 secure stars from extended sources using the u * − J versus g − K s colours, combined with the SExtractor stellarity index of the images. Comparing with galaxy spectroscopic redshifts, we find a photometric redshift dispersion, σ Δz/(1+z) , of 0.032 and catastrophic failure rate, Δz 1+z > 0.15, of 5.8% at z < 1, while a dispersion of 0.117 and a catastrophic failure rate of 16.6% at z > 1. We extend the estimate of the z p uncertainty over the full magnitude/redshift space with a redshift probability distribution function and find that our redshifts are highly accurate with z < 22 at z p < 2.5 and for fainter sources with z < 24 at z p < 1. From the investigation of photometric properties of AKARI infrared sources (23 354 sources) using the g z K s diagram, < 5% of AKARI sources with optical counterparts are classified as high-z (1.4 < z < 2.5) star-forming galaxies. Among the high-z star-forming galaxies, AKARI mid-infrared detected sources seem to be affected by stronger dust extinction compared with sources with non-detections in the AKARI mid-infrared bands.