CAPÍTULO III: ANÁLISIS
3.6 Utilización de tablas cruzadas
vey
There were a number of key guaranteed time projects proposed for the Herschel mission proposed in the planning stage that were selected as legacy projects. HerMES, the Herschel Multi-tiered Extragalactic Survey (Oliver et al.,2012), was given the most amount of time on Herschel , with 900 hours and covering an area of 100 sq. deg with the SPIRE and PACS instruments, and 270 sq. deg. in one shallow field (HeLMS) with SPIRE alone. The broad science goal of the survey was to provide insight into galaxy evolution using far-infrared wavelengths.
Figure 2.1: The locations and approximate footprints of many HerMES fields shown in galactic plane projection on top of IRAS dust maps. The warmer the colours the more intense the galactic cirrus emission. HerMES fields were placed to avoid the bulk of this galactic emission. This figure was produced by the HELP collaboration as discussed in a later chapter and so also includes fields from other surveys as indicated by asterisks.
To diversify the possible science results varying field sizes and depths were selected. The smaller fields reach a greater depth than the wider fields with many scan repeats and the shallower, wider fields have fewer scans contributing towards the final image. This leads to the so-called tiered “wedding-cake” design of the survey, ranging from the so- named Level-1 cluster fields of 0.08 sq. deg. to Level 6 region of XMM-LSS-SWIRE at 19 sq. deg. with HeLMS as the only Level 7 field. Fields were designed to sample a broad range of redshifts and luminosities of galaxies, the smaller fields reaching deeper to find fainter galaxies, and the wider, shallower fields finding the brighter, rarer objects. Binning by redshift and luminosity, the survey was designed to find 75 galaxies in each ∆ log L = 0.5 times ∆z=0.2 bin, between redshifts 0 ≥ z ≥ 3. At 250µm, the luminosity depths were approximated at z = 2 to be 2 × 1012L for deep fields like COSMOS, to
1 × 1013L in HeLMS using the Lagache et al. (2003) models. Table2.1gives the names,
AOR (astronomical observing request) numbers as well as the sizes and number of scans for all HerMES fields. Bolded field names are discussed in some capacity in this thesis. Many of these fields are so-called legacy fields that include data at other wavelengths from past surveys. This allows multi-wavelength science to be conducted, with examples given below.
Set Level Target Mode Ωgood(deg2) Scan num. 1 CD Abell 2218 Sp. Nom. 0.10 100 2 CD Abell 1689 Sp. Nom. 0.08 48 3 CD MS0451.6-0305 Sp. Nom. 0.08 48 4 CS RXJ13475-1145 Sp. Nom. 0.08 48 5 CS Abell 1835 Sp. Nom. 0.08 48 6 CS Abell 2390 Sp. Nom. 0.08 48 7 CS Abell 2219 Sp. Nom. 0.08 48 8 CS Abell 370 Sp. Nom. 0.08 48 9 CS MS1358+62 Sp. Nom. 0.08 48 10 CS Cl0024+16 Sp. Nom. 0.08 48 11 CH MS1054.4-0321 Sp. Nom. 0.16 16 12 CH RXJ0152.7-1357 Sp. Nom. 0.16 16 13 L1 GOODS-S Sp. Nom. 0.35 76 14 L2 GOODS-N Sp. Nom. 0.55 30 15 L2 ECDFS Sp. Nom. 0.58 19 22 L2 COSMOS Sp. Nom. 2.82 8
17 L3 Groth Strip Sp. Nom. 0.60 7
18 L3 Lockman-East ROSAT Sp. Nom. 0.57 7
18B L3 Lockman-East Spitzer Sp. Nom. 1.40 7
19 L3 Lockman-North Sp. Nom. 0.65 7
23 L4 UDS Sp. Nom. 2.02 7
24 L4 VVDS Sp. Nom. 2.02 7
22B L5 COSMOS HerMES Sp. Nom. 4.38 4
27 L5 CDFS SWIRE Sp. Fast 11.39 20
28 L5 Lockman SWIRE Sp. Fast 17.37 2
28B L5 Lockman SWIRE Sp. Fast 7.63 2
29 L5 EGS HerMES Parallel 2.67 7
30 L5 Bootes HerMES Parallel 3.25 5
31 L5 ELAIS N1 HerMES Parallel 3.25 5
32 L5 XMM VIDEO1 Parallel 2.72 4
32B L5 XMM VIDEO2 Parallel 1.74 4
32C L5 XMM VIDEO3 Parallel 2.73 4
Set Level Target Mode Ωgood(deg2) Scan num.
33 L5 CDFS SWIRE Parallel 10.89 5
34 L5 Lockman SWIRE Parallel 16.08 4
39B L5 ELAIS S1 VIDEO Parallel 3.72 4
35 L6 ELAIS N1 SWIRE Parallel 12.28 2
36 L6 XMM-LSS SWIRE Parallel 18.87 2
37 L6 Bootes NDWFS Parallel 10.57 2
38 L6 ADFS Parallel 7.57 2
39 L6 ELAIS S1 SWIRE Parallel 7.86 2
40 L6 FLS Parallel 6.71 2
41 L6 ELAIS N2 SWIRE Parallel 7.80 2
42 L7 HeLMS Sp.Fast 270 2
Table 2.1: The set, standard name, scan mode, area of the field and number of scans. Each set has a corresponding number of AORs and the AOR lists are given below. Bolded field names are fields that are explicitly discussed in some capacity in this thesis.
The type of science possible was very much dependent on the confusion noise of the maps. The Herschel telescope beam is much wider than similar generation optical and near-infrared telescopes. Galaxies that are resolvable in the optical and near-infrared with SDSS or Spitzer may in the Herschel maps occupy the same telescope beam or even pixel, becoming unresolved point sources that are heavily blended together. Thus catalogues produced on Herschel maps are considered catalogues of sources rather than galaxies as one cannot assume that the extracted flux can be attributed to only one galaxy. Due to the number of distant galaxies emitting in the infrared across the sky the background to the map can be thought of as consisting of light from many galaxies instead of a flat black background. The combination of blended sources and a background consisting of a fluctuation of sources contributes to uncertainty in the determination of a source’s flux. This is referred to as the confusion noise and is a hard limit on the detectability of sources. Above the confusion noise (constrained by Nguyen et al. (2010) at 5σ=24.0, 27.5, 30.5mJy at 250, 350, and 500µm respectively) catalogues of sources can be constructed by finding peaks in the map corresponding to the shape of the instrument’s point spread function (PSF) scaled to the flux of the source. These catalogued sources allow population statistics to be calculated. Stacking on known sources (B´ethermin et al., 2012) to find average fluxes for unresolved sources or direct analysis on the map with a model of the
number counts (Glenn et al., 2010) enabled the number counts to reach down to 2mJy and to account for over 50% of the CIB from known sources alone. Viero et al. (2013) find galaxies of mass 9.5 log(M/M) to 10 log(M/M) contribute the most to the CIB
at 250µm by simultaneously stacking sources binned on e.g. mass and colour to prevent biasing by flux from nearby sources.
Fields were chosen for the wealth of ancillary data already present. This allowed a matching between HerMES sources to spectroscopic redshifts (spec-zs) or to enable the determination of photometric redshifts (photo-zs) using the multi-wavelength data available by fitting model spectral energy distributions (SEDs). SEDs for HerMES sources were determined by combining SPIRE and PACS data from the PEP (PACS Evolutionary Probe,Lutz et al.(2011)) survey inElbaz et al.(2010), and more recently inHuang et al.
(2014). These photometric results were fitted to empirical templates in Rowan-Robinson
et al. (2010) providing evidence for a cold dust component to the SED. These template
fits would allow the determination of a redshift distribution for obscured star-forming galaxies. Given the redshift and the Herschel fluxes, the infrared luminosity of a source can be determined and thus the time evolution of the infrared luminosity function found as in Vaccari et al.(2010) andEales et al.(2010b) showing strong evolution out to z ≈ 1. More recently, Gruppioni et al. (2013) combine HerMES and PEP data to confirm that the infrared luminosity function evolves out to z = 1, flattens between 1 < z < 3 and drops off after z = 3, consistent with the peak of star formation occurring at z = 2.
The link between star formation and environment could also be explored in wide fields that cover a broad range of densities. As the most massive halos are more strongly clustered (Kaiser,1984) and there is strong correlation between the mass of a dark matter halo and a hosted galaxy, determining the clustering of HerMES sources and therefore the clustering of star-forming galaxies will give clues as to the type of environment that hosts a star forming galaxy. The smplitude of clustering at particular scales will also determine whether star formation occurs in the largest galaxies within central halos or smaller, orbiting galaxies in the sub-halos. Cooray et al.(2010) found the brighter HerMES sources were in dark matter halos above (5 ± 4) × 1012M, with contributions towards both the 1-
halo and 2-halo clustering strengths. These results are in contention with H-ATLAS, (Eales
et al., 2010a) another Herschel legacy project, finding no clustering strength at 250µm
(Maddox et al., 2010). A follow-up study by Mitchell-Wynne et al. (2012) constrained
the redshift distribution N (z) of sources in the B¨ootes field. The correlation function was cross-correlated with ancillary data in the near infrared to constrain the redshift
distribution but in the process found a clustering strength less than Cooray et al.(2010) at 250µm in the process. This discrepancy is still unresolved.
With such a wide area covered by HerMES, rare objects can also be discovered. Colour- cutting the catalogues and maps (choosing sources or pixels that have a ratio between two wavelengths greater than some value) to select sources that are extremely red and therefore more distant was performed by Dowell et al. (2014) and found the extreme object HFLS3. HFLS3 was followed up with a variety of telescopes (Riechers, 2013) and was spectroscopically confirmed at z = 6.34. This was at the time considered the most distant unlensed star forming galaxy then identified, and its star formation rate (SFR) then calculated at approximately 3000Myr−1. At those star formation rates an object would
be rare enough that current galaxy formation models would not tolerate finding more than a few similar objects (Riechers,2013). Subsequent follow-up observations (Cooray et al.,
2014) have found that the system is lensed and have revised the star formation rate to 1320Myr−1. Whilst this is not a paradigm-shifting value, these rarer systems are still of
great value for galaxy evolution models.
Distant objects can also be found through gravitational lensing as with HFLS3. Ex- tremely high mass galaxies or galaxy clusters strongly warp space-time around them, forcing light from objects behind them to change path as they pass through the gravit- ational field, allowing the otherwise hidden object to be observed. The lensed object is magnified and can be multiply imaged dependent on the alignment of the lens and object, aiding identification of lensed sources. Lensed sources allow us to probe further in redshift and/or to fainter sources. Wardlow et al. (2013) identified 13 lensed source candidates across the L2-6 fields with nine confirmed in telescope follow-ups on the Hubble Space Telescope or W. M. Keck Observatory. In addition over a hundred sources were flagged within the Lockman fields as being potential lensed sources by Rowan-Robinson et al.
(2014) using SED template fits to photometry.
The diverse scientific results above all utilise data products created by the HerMES telescope data processing teams and will be the subject of discussion for the rest of this chapter and chapter3. To process the telescope data, the HerMES team was divided into four, PMAP and PCAT to process the PACS maps and catalogues respectively, and SMAP and SCAT for the SPIRE data. The majority of the work in this thesis falls within the SCAT team’s remit, with work in this chapter directly contributing to the HerMES second data release (Wang et al.,2014). The SMAP process is also described below as simulation maps were made using the pipeline for this and a later chapter and a comparison between
the HerMES and H-ATLAS map-making process is made in chapter 6. The PMAP and PCAT processes are not discussed in this thesis.