FUNCIÓN PRINCIPAL DE LOS TRIBUNALES.

resultant φ∗ _{parameter would not be comparable to its equivalent in Equation 1.7.}

1.7

Overview

Until recently, any measurement that required NIR observations over a large area of sky was forced to rely on shallow 2MASS imaging. The UKIRT Infrared Deep Sky Survey (UKIDSS, Lawrence et al. 2007) is a programme of five NIR surveys, and greatly super- sedes 2MASS. This thesis uses data from the shallowest (but most extensive) of these surveys: the Large Area Survey (UKIDSS-LAS). The survey is detailed in Section 2.3. The LAS is over 10x deeper than the 2MASS survey inK, and should therefore be much less susceptible to the SB effects discussed in the previous sections. Smith et al. (2009) pro- duced a K-band SDSS-UKIDSS LAS luminosity function that appeared to agree closely with the results of Cole et al. (2001) and Jones et al. (2006). At first glance, this would suggest that 2MASS and UKIDSS photometry are consistent, the reported NIR LFs ro- bust and the NIR/optical offset a physical phenomenon. However, due to an unresolved issue with the UKIDSS extraction software discussed in Chapter 3, Smith et al. question the validity of their own results. They also restricted their analysis toK-band only where, for the purpose of recovering the pre- and post attenuated CSED, it is more desirable to recover measurements in all available filters (i.e. Y JHK).

Cosmic variance may also produce significant uncertainty. NIR surveys typically are based on lower redshift samples, e.g., 6dfGS (Jones et al., 2006) with_hz_{i ≈}0.05, with the optical data accessing deeper populations; e.g., SDSS Blanton et al. (2003) with_hz_{i ≈}0.1). In an ideal situation a derivation of the CSED would be produced from within a single survey, where the impact of cosmic variance would affect all filter measurements consistently. The purpose of this thesis is to generate such a dataset. In Chapter 3, by combining data from the Millennium Galaxy Catalogue (MGC), SDSS and UKIDSS LAS surveys I derive

ugrizY JHK luminosity distributions and pre- and post attenuated cosmic SEDs from

within a single well understood volume of_∼71,000h3Mpc3, using a sample of_∼10 thou- sand galaxies, selected using a deep BM GC = 20 mag apparent magnitude cut. In later

chapters, I create a larger sample of _∼ 250 thousand galaxies using the UKIDSS, SDSS and GAMA datasets, produce a second ugrizY JHK CSED, and compare the results. The creation of the GAMA sample is detailed in Chapter 4, and the results derived are

2

Surveys

In order to generate the total luminosity density parameters required to model the CSED, data from a number of instruments must be combined with a spectroscopic survey. In this chapter I describe the surveys that provided these datasets. I detail their area coverage, limiting depth and the photometric and astrometric accuracy of the data they produce.

2.1

The Millennium Galaxy Catalogue (MGC)

The Millennium Galaxy Catalogue (Liske et al., 2003) is a deep (Blim = 24 mag, µlim = 26 mag arcsec−2_{),B band galaxy survey created using the Wide Field Camera on the 2.5m} Isaac Newton Telescope (INT), with observations taken between 1999 and 2001. Survey- ing a long (75◦_{), thin (0.5}◦_{), equatorial strip amounting to 37 deg}2 _{of sky (30.88 deg}2_after cleaning and cropping), the MGC contains data on over a million objects, with 10,095 galaxies brighter than B = 20 mag (this resolved sub-catalogue is referred to as MGC- Bright, and the integrity of every object within it has been verified by eye and fixed where

necessary). Photometry was performed during a single night, where 20 fields across the survey strip were observed, interspersed with observations of Landolt standard stars from standard areas SA98, SA101, SA104 and SA107.

Using the extensive overlap regions, for objects in the range 17 _≤ BMGC ≤ 21 mag, the astrometric rms uncertainty has been shown to be_±0.08 arcsec, and the photometric un- certainty_±0.023 mag. The seeing ranged from 0.9 to 2 arcsec, with a median of 1.3 arcsec. Following observation, the data was sent to the Cambridge Astronomy Survey unit for primary data analysis. The zero point for each field was calculated by comparing it with the aforementioned Landolt standard star fields.

Object extraction was achieved using theSExtractorprogram (Bertin & Arnouts, 1996), with a constant surface brightness threshold, µlim, of 26 mag arcsec−2. Photometry was

extinction corrected using dust maps created by Schlegel et al. (1998).

Background subtraction in SExtractoris carried out by estimating the local background inside a number of meshes within a grid that covers the entire image. The background values are calculated using 3σ clipped median values within each mesh. A background map is then calculated by interpolating between the values from each mesh. The survey team selected the largest possible mesh size available inSExtractor, in order to carry out the background subtraction. This criteria limits the smoothing of low surface brightness extended objects. An artificial neural network was used to give every object in the survey a stellaricity parameter (1=star, 0=galaxy). Driver et al. (2005), have obtained redshifts for 96 per cent of MGC-Bright galaxies using the SDSS spectroscopic survey and 2dFGRS data releases, combined with a dedicated MGCz 2dF survey. All data is publicly available from http://www.eso.org/_∼jliske/mgc/.