spacer link to MAST page spacer logo image spacer

The Hubble Ultra Deep Field 2012 (HUDF12) Program

Observational Summary

The Hubble Ultra Deep Field 2012 program (HUDF12: Ellis et al. 2013; Koekemoer et al. 2013) is a large HST program awarded 128 orbits (HST Program ID 12498; PI: R. Ellis), designed to probe the contribution of early galaxies to cosmic reionization. The main project website is accessible at This program significantly extends previous data on the main Ultra Deep Field (UDF) with new deep observations using the HST WFC3/IR camera in a range of filters that are crucial to addressing these questions.

Specifically, our new observations quadruple the exposure time of F105W on the UDF main field by adding 72 new orbits to data obtained previously in program ID 11563 (PI: G. Illingworth; see Bouwens et al. 2011), as well as providing 30 orbits in the completely new F140W filter, and providing an additional 26 orbits of depth in the F160W filter. The new observations are obtained with observing strategies as consistent as possible with previous observations, and are combined with these data along with all other existing data on this field, as was committed to in the proposal, to deliver a single, unified set of mosaics of the UDF providing a legacy dataset on this field of lasting scientific value.

UDF12 Observing Summary (HST Program ID 12498)
Field Camera / Filter Orbits Exposures Exposure time (s)
UDF-main: WFC3/IR F105W 72 144 198423
WFC3/IR F140W 30 60 82676
WFC3/IR F160W 26 52 71652
UDF-par2*:ACS/WFC F814W 128 256 322944
*Note: for the ACS UDF-par2 mosaics, please use the updated v1.1 mosaics, which have corrected header keywords (the pixels values in the v1.1 mosaics are identical to v1.0, just the header keywords are corrected).

The WFC3/IR observations for the HUDF12 program 12498 were all obtained during the period 4 August 2012 to 16 September 2012, at a spacecraft orient of 264 degrees for all observations. This enabled all the parallel ACS exposures to be placed on the UDF-par2 field, exactly overlapping the previous ACS parallels on this field from program 11563. The full 128 orbits of ACS observations on the UDF-par2 field were obtained in the F814W filter, complementing the existing ACS coverage of this field to offer the most efficient, deepest possible dropout selection of galaxies at z>6.5 (justified further in McLure et al. 2011). Each orbit was divided into two WFC3/IR prime exposures, together with two corresponding ACS parallel exposures. The WFC3/IR exposures were all obtained using the SPARS100 readout pattern with either 15 or 16 samples per exposure, depending on the orbital visibility (thus yielding exposure times of either 1300s or 1400s, after accounting for the additional two samples at the start of each exposure).

The dither patterns followed a strategy that was matched to the previous observations of the UDF, in particular including small sub-pixel dithers to provide good sampling of the sub-pixel phase space, along with larger dithers on scales of ~1-3 arcseconds to move sources around the detectors, mitigating the impact of the ACS chip gap in the parallel observations as well as moving around detector artifacts and areas of persistence in the WFC3/IR observations.

Data Processing and Products

All the WFC3 and ACS data have been processed using the Mosaicdrizzle pipeline (see Koekemoer et al. 2002, 2011 for further details about this pipeline in general, and Koekemoer et al. 2013, for details about the HUDF12 processing in particular), which includes calibration, astrometric alignment, bad pixel rejection, and final drizzle combination. Initial calibrated versions of all the WFC3 and ACS exposures were produced as a first-pass calibration, using the standard Pyraf/STSDAS calacs and calwf3 routines. These enabled initial quality inspection of the data, prior to subsequent refinemment. The WFC3/IR data were then recalibrated using superdarks that we constructed from all the on-orbit darks observed in this readout mode (providing a higher signal-to-noise dark file than the standard reference files), and all pixels affected by persistence were masked out. In addition, a fraction of the WFC3/IR exposures were affected by changing sky backgrounds, for which further processing was necessary in addition to the standard calwf3 up-the-ramp sampling, in order to ensure that the pixel statistics in the final images were correct. The ACS images underwent additional processing for correcting CTE losses and bias destriping, all of which are incorporated into the calacs code distributed under Pyraf/STSDAS. Further details of the processing are provided in Koekemoer et al. (2013).

Astrometric alignment of the all exposures was carried out within Mosaicdrizzle using the published catalogs for the UDF main field (Beckwith et al. 2006) as well as catalogs generated from the UDF-par2 data (Bouwens et al. 2011), using in an iterative process that involves solving for shifts and rotations using a combination of catalog-matching and cross-correlation. Once the images were aligned, bad pixels and residual cosmic rays could be rejected by comparing each image to a deep stack of the full set of images. Satellite trails were also identified and masked. In addition, low-level residual background structure in the WFC3/IR images was removed using median filtering of the empty regions of all the images, restricting the spatial scale of any subtracted structure to 5" or larger to ensure that small-scale features were retained in the images. The final combination using drizzle (Fruchter & Hook 2002) produces distortion-corrected mosaics consisting of the weighted sum of all the input exposures (*_drz.fits), in units of electrons/second, along with the correponding inverse variance weight files (*_wht.fits). Correlated noise can be accounted for using the prescriptions provided in Casertano et al. (2000). In addition, these mosaics combine the new observations from the HUDF12 program (HST program ID 12498, PI: R. Ellis) as well as the previous data from the HUDF09 program (HST program ID 11563, PI: G. Illingworth), together with exposures from other programs (in particular CANDELS: HST program IDs 12060, 12061, 12062, PI. S. Faber and H. Ferguson) although the latter only add a relatively small number of additional orbits. This full-depth combination was committed to in the original HUDF12 proposal, and provides a single, unified set of mosaics of the UDF, constituting a lasting legacy dataset on this field.

Final Combined Full-Depth UDF-main WFC3/IR mosaics (orbits)
Camera / Filter HUDF09 HUDF12 Other Final*
WFC3/IR F105W 24 72 4 100
WFC3/IR F125W 34 - 5 39
WFC3/IR F140W - 30 - 30
WFC3/IR F160W 53 26 5 84
* Final combined WFC3/IR depth from all the following programs on the UDF-main field: HUDF09: HST Program ID 11563 (PI: G. Illingworth); HUDF12: HST Prorgram ID 12498 (PI: R. Ellis); Other: HST Program ID 12099 (PI: A. Riess) and CANDELS, HST Program ID 12060,12061,12062 (PI: S. Faber, H. Ferguson)

Data Download Options

Explore the data via an interactive display.

The combined mosaics are being made available as part of the the STScI High-Level Science Products, via web browser and ftp.

  • Via web browser: is the starting navigation point to all the HUDF12 data.
  • FTP:
    ftp and login as anonymous:
    cd /pub/hlsp/hudf12
    From there the data of interest can be directly downloaded.
  • WGET
    The entire directory can be downloaded using wget commands such as:
    wget -q -nH --cut-dirs=3 -r -l0 -c -N -np -R 'index*' -erobots=off


Beckwith, S. V. W. 2006, AJ, 132,1729
Bouwens, R. et al. 2011, ApJ, 737, 90
Casertano, S. et al. 2000, AJ, 120, 2747
Ellis, R. S. et al. 2013, ApJ, 763, L7
Fruchter, A. S and Hook, R. N. 2002, PASP, 114, 144
Grogin, N. A. et al. 2011, ApJS. 197, 35
Koekemoer, A. M. et al. 2002, HST Calibration Workshop, 337
Koekemoer, A. M. et al. 2011, ApJS, 197, 36
Koekemoer, A. M. et al. 2013, ApJS, 209, 3
McLure, R. J. et al. 2011, MNRAS 218, 2074