The Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey (CANDELS) is designed to document the first third of galactic evolution, over the approximate redshift (z) range 8-1.5. It will image >250,000 distant galaxies using three separate cameras on the Hubble Space Telescope, from the mid-ultraviolet to the near-infrared, and will find and measure Type Ia supernovae at z > 1.5 to test their accuracy as standardizable candles for cosmology. Five premier multi-wavelength sky regions are selected, each with extensive ancillary data. The use of five widely separated fields mitigates cosmic variance and yields statistically robust and complete samples of galaxies down to a stellar mass of 10 9 M to z ≈ 2, reaching the knee of the ultraviolet luminosity function of galaxies to z ≈ 8. The survey covers approximately 800 arcmin 2 and is divided into two parts. The CANDELS/Deep survey (5σ point-source limit H = 27.7 mag) covers ∼125 arcmin 2 within Great Observatories Origins Deep Survey (GOODS)-N and GOODS-S. The CANDELS/Wide survey includes GOODS and three additional fields (Extended Groth Strip, COSMOS, and Ultra-deep Survey) and covers the full area to a 5σ pointsource limit of H 27.0 mag. Together with the Hubble Ultra Deep Fields, the strategy creates a three-tiered "wedding-cake" approach that has proven efficient for extragalactic surveys. Data from the survey are nonproprietary and are useful for a wide variety of science investigations. In this paper, we describe the basic motivations for the survey, the CANDELS team science goals and the resulting observational requirements, the field selection and geometry, and the observing design. The Hubble data processing and products are described in a companion paper.

This paper describes the Hubble Space Telescope imaging data products and data reduction procedures for the Cosmic Assembly Near-IR Deep Extragalactic Legacy Survey (CANDELS). This survey is designed to document the evolution of galaxies and black holes at z ∼ 1.5 − 8, and to study Type Ia SNe beyond z > 1.5. Five premier multi-wavelength sky regions are selected, each with extensive multiwavelength observations. The primary CANDELS data consist of imaging obtained in the Wide Field Camera 3 / infrared channel (WFC3/IR) and UVIS channel, along with the Advanced Camera for Surveys (ACS). The CANDELS/Deep survey covers ∼ 125 square arcminutes within GOODS-N and GOODS-S, while the remainder consists of the CANDELS/Wide survey, achieving a total of ∼ 800 square arcminutes across GOODS and three additional fields (EGS, COSMOS, and UDS). We summarize the observational aspects of the survey as motivated by the scientific goals and present a detailed description of the data reduction procedures and products from the survey. Our data reduction methods utilize the most up to date calibration files and image combination procedures. We have paid special attention to correcting a range of instrumental effects, including CTE degradation for ACS, removal of electronic bias-striping present in ACS data after SM4, and persistence effects and other artifacts in WFC3/IR. For each field, we release mosaics for individual epochs and eventual mosaics containing data from all epochs combined, to facilitate photometric variability studies and the deepest possible photometry. A more detailed overview of the science goals and observational design of the survey are presented in a companion paper.

Spectroscopic+photometric redshifts, stellar mass estimates, and rest-frame colors from the 3D-HST survey are combined with structural parameter measurements from CANDELS imaging to determine the galaxy size-mass distribution over the redshift range 0 < z < 3. Separating early-and late-type galaxies on the basis of star-formation activity, we confirm that early-type galaxies are on average smaller than late-type galaxies at all redshifts, and we find a significantly different rate of average size evolution at fixed galaxy mass, with fast evolution for the early-type population, R eff ∝ (1 + z) −1.48 , and moderate evolution for the late-type population, R eff ∝ (1 + z) −0.75 . The large sample size and dynamic range in both galaxy mass and redshift, in combination with the high fidelity of our measurements due to the extensive use of spectroscopic data, not only fortify previous results, but also enable us to probe beyond simple average galaxy size measurements. At all redshifts the slope of the size-mass relation is shallow, R eff ∝ M 0.22 * , for late-type galaxies with stellar mass > 3 × 10 9 M , and steep, R eff ∝ M 0.75 * , for early-type galaxies with stellar mass > 2 × 10 10 M . The intrinsic scatter is 0.2 dex for all galaxy types and redshifts. For late-type galaxies, the logarithmic size distribution is not symmetric but is skewed toward small sizes: at all redshifts and masses a tail of small late-type galaxies exists that overlaps in size with the early-type galaxy population. The number density of massive (∼ 10 11 M ), compact (R eff < 2kpc) early-type galaxies increases from z = 3 to z = 1.5 − 2 and then strongly decreases at later cosmic times.

We present the evolution of the luminosity-size and stellar massYsize relations of luminous (L V k 3:4 ; 10 10 h À2 70 L) and massive (M Ã k 3 ; 10 10 h À2 70 M) galaxies in the last $11 Gyr. We use very deep near-infrared images of the Hubble Deep FieldYSouth and the MS 1054-03 field in the J s , H, and K s bands from FIRES to retrieve the sizes in the optical rest frame for galaxies with z > 1. We combine our results with those from GEMS at 0:2 < z < 1 and SDSS at z $ 0:1 to achieve a comprehensive picture of the optical rest-frame size evolution from z ¼ 0 to 3. Galaxies are differentiated according to their light concentration using the Sérsic index n. For less concentrated objects, the galaxies at a given luminosity were typically $3 AE 0:5 (AE2) times smaller at z $ 2:5 than those we see today. The stellar massYsize relation has evolved less: the mean size at a given stellar mass was $2 AE 0:5 times smaller at z $ 2:5, evolving proportionally to (1 þ z) À0:40AE0:06. Simple scaling relations between dark matter halos and baryons in a hierarchical cosmogony predict a stronger (although consistent within the error bars) than observed evolution of the stellar massYsize relation. The observed luminosity-size evolution out to z $ 2:5 matches well recent infall model predictions for Milky WayYtype objects. For low-n galaxies, the evolution of the stellar massYsize relation would follow naturally if the individual galaxies grow inside out. For highly concentrated objects, the situation is as follows: at a given luminosity, these galaxies were $2:7 AE 1:1 times smaller at z $ 2:5 (or, put differently, were typically $2:2 AE 0:7 mag brighter at a given size than they are today), and at a given stellar mass the size has evolved proportionally to (1 þ z) À0:45AE0:10 .