2. Resultados ambientales, financieros y económicos
2.2 Impacto de los ingresos por tasas retributivas sobre los presupuestos de las
2.2.2 Inversiones ambientales con el recaudo de las tasas retributivas: los fondos
There are various different surveys of star forming galaxies. Most surveys fall into two main classes; emission line selected and magnitude selected. The narrowband filter, objective prisms and Fabry–Perot interferometers based surveys are generally emission–line selected, while large sky surveys tend to be broadband selected. The design of a survey plays an important role in shaping galaxy samples used to probe the star formation in galaxies. A vast majority of the published SFR density measurements are based on galaxy samples either drawn from narrowband filter surveys or broadband surveys. The differing sensitivities, depth, and image/spectral qualities between these types of surveys are possible causes for some of the discrepancy seen in published measurements of local (z. 1) SFR densities.
1.3.1 Narrowband filter vs Broadband spectroscopic surveys
First attempts at determining the global star formation history based on emission–line selected surveys involve the observations undertaken using objective prisms on Schmidt telescopes (e.g. Universidad Complutense de Madrid survey, Gallego et al. 1995). While the use of a Schmidt telescope, which has a large field of view, provided the necessary sky coverage needed to obtain a representative sample of star forming galaxies, the sensitivity of the photographic plates in the instrumental set up restricted the observations to the low redshift Universe (z < 0.04, Zamorano et al. 1994).
Narrowband surveys at optical wavelengths are able to provide deep imaging over a narrow redshift slice, effec- tively yielding a relatively large volume limited samples of galaxies. In comparison to broadband surveys, there are several advantages to selecting objects through a narrowband filter. The target selection is achieved through emission–lines, so they are most effective at detecting weak emission–line sources, producing a galaxy sample complete down to a pre–determined flux limit. Also, the galaxies are selected using a quantity they aim to mea- sure, which scales directly with SFR (Jones & Bland-Hawthorn 2001; Westra & Jones 2008). The sky background, one of the frustrating limitations of broadband surveys, is greatly reduced in narrowband images (Thompson et al. 1995). The night sky spectrum at wavelengths longer than∼ 7000 ˚A is dominated by the emission from atmo- spheric hydroxyls (OH) lines (i.e. the OH forest). In the case of broadband surveys, the measurements of emission
14 INTRODUCTION
features that are redshifted in to this region carry large uncertainties. In contrast, by placing narrowband filters in the regions where the OH emission is weak, the night sky contamination can be greatly reduced to recover the light from faint objects.
There are, however, a number of drawbacks to narrowband surveys as well. The main disadvantages are the need to assume common corrections for stellar Balmer absorption, dust obscuration and contamination by AGNs for the sample as a whole rather than for each galaxy, and the insensitivity to low equivalent widths due to the lack of spectroscopy (Westra et al. 2010). These assumptions introduce large uncertainties and can lead to a systematic underestimate of the final SFR density (Spector et al. 2012; Massarotti et al. 2001; James et al. 2004). Moreover, due to the relatively small volumes probed by emission–line selected surveys, the sample variance (also called cosmic variance) related issues resulting from the large scale structure of the galaxy distribution can be significant for this type of survey.
Broadband spectroscopic surveys on the other hand do not suffer from the same disadvantages. Spectroscopy allows individual corrections for dust obscuration to be applied through Balmer decrements, and AGNs to be excluded through emission–line ratios based on common AGN/star forming diagnostics (e.g. Kewley & Dopita 2002; Kauffmann et al. 2003; Cid Fernandes et al. 2011a). Additionally, spectroscopy allows several different ways to account for the underlying stellar absorption. One of the simplest methods of accounting for the effects of stellar absorption is by applying a constant correction to the Balmer line equivalent widths (Hopkins et al. 2003, 2013). Chapter 2 describes how a constant correction propagates to the Balmer line fluxes. This assumption can introduce some uncertainty to the line flux and luminosity measurements (Chapter 2), and must therefore be restricted to examining the gross characteristics of large samples of galaxies (Hopkins et al. 2003). Other methods involve comparing a galaxy spectrum with a library of single stellar population models generated using population synthesis models (e.g. Bruzual & Charlot 2003) to fit the continuum shape of that spectrum. This accounts for weak features, and Balmer stellar absorption. Once the best–fit stellar population synthesis model to the continuum is subtracted and any remaining residuals are removed, Gaussian profiles are fitted simultaneously to all the emission lines, requiring that all the lines belonging to the Balmer and forbidden–line series have the same width, and velocity offset. This requirement on line widths and velocity offsets allows stronger/multiple lines to be used to constrain the weaker lines. This process is described in detail in Brinchmann et al. (2004) and Tremonti et al. (2004). Also, Hopkins et al. (2013) describe the process employed by GANDALF v1.5, which simultaneously fits both Gaussian emission line and stellar population templates to the data, taking into account the derived stellar kinematics, while also correcting for the diffuse obscuration. The spectroscopic surveys have an additional advantage that they have a greater sky coverage than narrowband surveys, such that sample variance issues can be overcome.
The main unfavourable quality of a broadband survey related to calculating SFRs is the broadband selection, which can bias a galaxy sample towards a particular passband, which may or may not be as sensitive to the underlying star forming population as narrowband filters. We explore how this selection can affect the derived SFR densities in Chapter 2. The data used for the analyses presented in this thesis are taken from the SDSS and GAMA surveys, the former is a shallow survey and the later covers the largest sky area for its depth. The subsequent sections provide brief descriptions for both surveys.
1.3.2 Sloan Digital Sky Survey
The Sloan Digital Sky Survey (SDSS, York et al. 2000) has imaged ∼ 10 000 deg2 in five optical broad–band filters (u, g, r, i, z) at 3551, 4686, 6165, 7481 and 8931 ˚A, using a wide–field imager with a mosaic CCD camera on a2.5 m telescope located at Apache Point Observatory, Sunspot, New Mexico. It surveyed the sky in a drift– scan mode and has limiting (AB) magnitudesu < 22.0, g < 22.2, r < 22.2, i < 21.3, z < 20.5 with point spread function width of 1.4 arcsec (Gunn et al. 1998). Photometric catalogues are then used to identify the spectroscopic targets on the same telescope, using a 640–fibre–fed pair of multiobject double spectrographs. The wavelength coverage is fromλλ 3800–9200 ˚A with a spectral resolution ofλ/∆λ ≈ 2000 (FWHM ∼ 2.4 ˚A at λ5000) (Abazajian et al. 2009). The limiting magnitude for the spectra is r < 17.77, which is substantially brighter than that for the imaging such that the spectroscopic completeness of the survey is∼ 100%. The spectroscopic
target selection is achieved through a targeting algorithm, and a tiling algorithm (Blanton et al. 2003c) then assigns the spectroscopic fibres to the targets. The main source of incompleteness in SDSS arises from the55” separation between fibres. As a result of this incompleteness a small fraction of galaxies are missed, largely biased towards regions with a high surface density of galaxies.
The SDSS–DR7 (Data Release 7) galaxies are used for the analysis presented in Chapter 2, which includes the spectra for∼ 106 objects over a total sky area of 9380 deg2. The SDSS galaxies with spectra are classified into
three main samples; the main galaxy sample (MGC, Strauss et al. 2002), the luminous red galaxy sample (LRG, Eisenstein et al. 2001), and the quasar sample (Richards et al. 2002).
The main galaxy sample, used in this thesis, is complete to a Petrosian r–band magnitude limit of 17.77 with further cuts in half–light surface brightness, µR50 < 24.5 mag arcsec−2, sky brightness, fibre magnitude and
various other flags, see Strauss et al. (2002) for details. The LRG sample, described in detail in Eisenstein et al. (2001), is approximately volume limited toz ≈ 0.38. This sample consists mostly of brightest cluster galaxies and have spectroscopic redshifts up tor ∼ 19.5. Finally, the quasar sample described in Richards et al. (2002) includes objects extending up toz∼ 6.
There are several sources of SDSS value added catalogues for galaxies exists other than those available in the main survey database. The emission–line data used in this thesis are taken from the MPA–JHU DR7 database5, and the derivation of these measurements is detailed in Brinchmann et al. (2004) and Tremonti et al. (2004). A brief discussion on the derivation of different properties for SDSS galaxies can be found in Chapter 2.
1.3.3 Galaxy And Mass Assembly survey
The work done in this thesis is largely based on the Galaxy And Mass Assembly (GAMA) survey data. GAMA is a spectroscopic survey undertaken at the Anglo–Australian Telescope (AAT). GAMA spectroscopic targets were selected from the SDSS Data Release 6 (DR6, Adelman-McCarthy et al. 2008) to limiting Petrosian magnitudes ofr < 19.4 in two fields, and r < 19.8 in the third field. The GAMA input catalogue is drawn from the SDSS and UKIRT Infrared Deep Sky Survey (UKIDSS), with an initial aim to measure redshifts for galaxies in three 48 deg2 regions at9, 12 and 14.5 hr, on the celestial equator, with magnitude selections r < 19.4, z < 18.2 and
KAB < 17.6 over all three regions, and r < 19.8 in the 12–hr region. Moreover, the GAMA survey implements
a highly complete star–galaxy separation that jointly uses an intensity-profile separator and a colour separator. Additional details related to the construction of the GAMA input catalogue is given in Baldry et al. (2010). The tiling of sources is performed using a “greedy” algorithm (Robotham et al. 2010) that ensure the main survey requirements will be met even under worse than typical weather conditions.
The data in the GAMA survey regions used for this thesis consists of measurements taken during the GAMA observations together with those from the SDSS, 2–degree field Galaxy Redshift Survey (2dFGRS) and Millen- nium Galaxy Catalogue (MGC) sources, as GAMA does not re–observe the majority of SDSS, 2dFGRS and MGC galaxies in the three GAMA regions. The GAMA spectra are obtained from the AAT with the 2–degree Field (2dF) fibre feed and AAOmega multiobject spectrograph. AAOmega provides a resolution of 3.2 ˚A full width at half maximum (FWHM) with complete spectral coverage from 3700–8900 ˚A (Sharp et al. 2006; Driver et al. 2011). The GAMA spectroscopic data set used in the analysis presented in this thesis is over98% complete in spectroscopic followup (Driver et al. 2011), the small spectroscopic incompleteness likely due to low–luminosity, low surface–brightness galaxies (Loveday et al. 2012). The determination of corrections to account for the survey incompleteness is described in detail in Chapter 2.
The GAMA flux calibration process, described in detail in Hopkins et al. (2013) and in Liske et al. (in prep.), is essentially a two–step process. In the first instance, an initial flux calibration is achieved for each 2dF plate to correct for the wavelength–dependence of the system throughput. This is then supplemented by an absolute flux correction.
Three fibres on each 2dF plate are assigned to standard stars. For each star a flux correction vector is derived by
5
16 INTRODUCTION
taking the ratio of the observed to its best fit model, the average between the three provides an unique wavelength– dependent correction for a given plate. Any lower–order shape in the continuum is removed by dividing the standard stellar spectrum by the unique correction vector. A fit to the residuals achieves an initial curvature correction that accounts for the poor CCD response at blue and red extremes of the spectrum. An absolute flux calibration is obtained by tying the spectrophotometry directly to ther–band petrosian magnitudes from the SDSS photometry.
The standard strong optical emission lines are measured from each curvature corrected and flux calibrated spectrum assuming a single Gaussian approximation and a common redshift and line–width within an adjacent set of lines (e.g. Hα and the [NII]λλ6548, 6583 doublet), and simultaneously fitting the continuum local to the set of lines (Hopkins et al. 2013). Further details related to the application of corrections for the underlying Balmer stellar absorption, dust obscuration and fibre aperture effects are discussed in Chapter 2. Finally, the SDSS photometry in u,g,r,i,z filters is available for each GAMA galaxy.