Additional targets in the first data release

Table 3.4 shows a breakdown of the various 6dFGS-NVSS/SUMSS additional targets and the fractions which have been observed as part of the first data release. The observational completeness for additional targets is never expected to be 100% as these targets are observed serendipitously. It is expected the observed sample will be an unbiased subset of the complete target list. Very few observations south of δ = −50◦ are available as

part of the first data release, so as yet only a small number of SUMSS radio sources have been observed. However, the 17% of NVSS additional targets which have been observed represent a sizable enough sample for analysis. Chapter 4 presents an analysis of the 6dFGS-NVSS additional targets found in the first data release.

3.6

The Database

The construction of a useful database of radio source identifications to be studied in this thesis is discussed in this section. After constructing the master list of 6448 candidate radio source identifications observed in the 6dFGS First Data Release, data for each identification were downloaded from the 6dFGS public database at www- wfau.roe.ac.uk/6dFGS/. For each object the data obtained were as follows:

3.6. The Database 75

Table 3.4 The Sample of SUMSS/NVSS additional targets observed in the 6dFGS first data release.

Sample Ntotal Nobserved %

Name Extendeda Stellarb Total Extendeda Stellarb Total Observed

NVSS 3759 3224 6997 580 609 1191 17%

SUMSS 827 1787 2614 3 3 6 0.2%

NOTES:

a“Extended” additional targets are those given optical morphology class 1 in the Super-

COSMOS Sky Survey.

b“Stellar” additional targets are those given optical morphology class 2 in the SuperCOS-

MOS Sky Survey.

Table 3.5 Summary of the spectral classes visually assigned to 6dFGS-NVSS objects. Class Type of spectrum

Aa Pure absorption line (early-type) spectrum.

Aae Absorption lines and narrow LINER-like emission lines. Ae Conventional Type I or II emission line AGN.

SF Star-forming galaxy or extragalactic HIIregion.

star Galactic star of any spectral type.

? Unclassifiable or unknown (usually because of low S/N). 1. A 6dF spectrum of the object.

2. A bJ and rF image of the object from SuperCOSMOS.

3. J, H & K images of the object from 2MASS (if available).

4. Data for each object, including J2000 position from the 2MASS XSC, redshift, quality flag & bJ, rF, J, H & K magnitudes.

Images of radio contours overlaid onto SuperCOSMOS bJ images were made and

each radio-optical identification was verified by the method discussed in Section 3.4.2 for primary targets and Section 3.5.2 for additional targets, the bJ images were used for

verification as not all objects in the database (eg. many of the additional targets) were located in the 2MASS XSC. Removing invalid radio-optical identifications resulted in a list of true radio identifications containing 5571 objects. Next, all spectra of candidate radio sources were inspected visually and classified. The classification scheme used is discussed in Section 3.6.1. Section 3.6.2 outlines how the data were collated into a final database.

3.6.1

Spectral Classification

The 6dFGS spectrum of each of the 5571 accepted radio source identifications was in- spected visually both to determine the validity of the redshift and to determine the physical

4000 5000 6000 7000 Wavelength (angstroms) 0 2500 5000 7500 10000 12500 Counts g1005480−172604 bJ=13.87 z=0.017 [SII] Hα [OIII] Hβ G H Mg Na 5000 6000 7000 8000 Wavelength (angstroms) 0 1000 2000 3000 Counts g1326188−305738 bJ=11.11 z=0.048 Hα Na Mg G H K Hβ 3930 4930 5930 6930 Wavelength (angstroms) 0 2000 4000 6000 Counts g0020419−042559 bJ=10.48 z=0.021 [NII] Na Mg Hβ G K H 5000 6000 7000 8000 Wavelength (angstroms) 0 1000 2000 3000 Counts g1537070−244727 bJ=17.42 z=0.139 Hα Hβ Hγ Hδ [0III]

Figure 3.12 Example 6dFGS first data release spectra of 4 spectral classes. In each spectrum common

redshifted absorption/emission lines are labelled at the measured redshift, as are the positions of sky emis- sion lines. Top left: An ‘SF’ galaxy, with characteristic strong and narrow Balmer lines. Top right: An ‘Aa’ galaxy, containing broad Mg absorption as well as a distinct break close to the H & K lines at 4100 ˚A.

Bottom Left: An ‘Aae’ galaxy. It looks similar to an ‘Aa’ galaxy but has a weak [NII] line at 6700 ˚A.

3.6. The Database 77

Table 3.6 Comparison of spectral classification between two human classifiers for a subset of ∼ 1800 6dFGS-NVSS sources. The numbers along the diagonal indicate agreement and those off the diagonal indicate disagreement between the two human classifiers.

Elaine Sadler

SF SF? Aa Aa? Aae Aae? Ae Ae? star ?

SF 792 65 2 3 11 19 10 8 2 10 SF? 15 17 3 1 10 10 8 3 0 5 Aa 0 2 427 15 18 1 1 0 0 5 Aa? 0 1 33 9 8 2 0 0 0 9 Aae 0 0 1 0 6 2 2 0 0 0 Aae? 0 2 2 0 24 6 7 2 0 3 Ae 1 0 0 0 2 0 30 0 0 1 Ae? 1 3 0 0 4 2 18 4 0 1 star 0 0 1 0 0 0 0 0 55 5 Thomas Mauch ? 13 8 14 6 2 2 1 5 2 104

process responsible for the radio emission of each galaxy. Each spectrum was classified as either ‘AGN’ for spectra indicative of galaxies harbouring an active galactic nucleus or ‘SF’ for spectra indicative of galaxies with ongoing star formation. AGNs can have a pure absorption line spectrum like that of a giant elliptical galaxy (classed as Aa); a spectrum with absorption lines and weak narrow LINER-like emission lines (classed as Aae); a conventional Type II AGN spectrum which has nebular emission lines such as [OII] , [OIII] or [NII] which are stronger than any Hydrogen Balmer emission lines (Hα

or Hβ) (classed as Ae); or a conventional Type I AGN spectrum with strong and broad Hydrogen Balmer emission lines (also classed as Ae). Star-forming galaxies have spectra typical of HII regions with strong narrow emission lines of Hα and Hβ dominating the

spectrum. Figure 3.12 shows an example of each type of spectrum. Spectra of galac- tic stars were classified ‘star’ and spectra for which a classification could not be made (mostly due to a low S/N spectrum) were classed ‘?’. A ‘?’ was also appended to the end of any classification which was not certain.

To check the reliability of this visual classification scheme a subsample of ∼ 1800 spectra was classified by Elaine Sadler and the classifications between us compared. Ta- ble 3.6 shows this comparison. On the whole the two human classifiers agreed well, with disagreement in 10% of cases. If only the AGN and SF classes are considered then the two human classifiers disagree only 6% of the time. The main cause for disagreement comes when comparing the classifications of emission-line AGN (Aae and Ae classes) and the SF class, and tends to occur for spectra which have line ratios which are border- line between the AGN and SF. These spectra could indicate a class of galaxy which is a composite source, having ongoing star formation as well as harbouring a radio-loud active galactic nucleus. The spectra for which the two human classifiers were in disagreement were examined in detail before a final decision was made on their class. Results from this table were kept in mind during classification of the remaining ∼ 3500 spectra. In their

Table 3.7 The number of spectra of each class outlined in the text for 6dFGS-NVSS primary targets and 6dFGS-NVSS additional targets.

Class Primary sample Additional target sample Total

SF 2644 280 2924 Aa 1268 148 1416 Aae 187 13 200 Ae 162 84 246 star 8 394 402 ? 237 146 383

analysis of 2dFGS spectra using the same classification scheme described here, Sadler et al. (2002) found that this scheme agreed well with methods such as principal compo- nent analysis (PCA; Folkes et al., 1999; Madgwick et al., 2002). This method was also found to agree well with classification based upon diagnostic emission line ratios (Jackson & Londish, 2000).

Spectra covering all 6dFGS quality classes were checked during the classification, and it was found that, in a small fraction of cases (∼ 1%) spectra with Q < 3 were classifiable with believable redshifts. These spectra were re-redshifted using therunzpackage and

included in the final sample. Classifiable spectra with Q < 3 are caused by errors in the redshifting process. Because of this the 6dFGS quality class was generally ignored during the analysis of the data. Table 3.7 shows how the spectra in the 6dFGS-NVSS sample were classified. Spectra of galaxies in the SUMSS sample are discussed in later chapters.

3.6.2

Sample Studied in this Thesis

After classification of all the spectra, data were compiled into a database which was com- piled in both ASCII text format and as anhtmltable accessible through a web browser

for easy access. A search engine was also written to make selecting subsamples of the database simple. The columns of the database are as follows:

1. NVSS or SUMSS catalogue name of the source. NVSS sources start with ’N’ and SUMSS sources start with ’S’.

2. 6dFGS database name of the source.

3. J2000 Right Ascension in hours, minutes and seconds. 4. J2000 Declination in degrees, arcminutes and arcseconds. 5. The 6dF measured redshift of the source.

6. The Heliocentric corrected redshift of the source.

7. The 6dFGS database quality flag for the redshift. Some sources had their redshifts remeasured; these were all assigned a redshift quality 3.

3.6. The Database 79 9. rF magnitude from the SuperCOSMOS database.

10. J magnitude from the 2MASS XSC. Additional targets are assigned a J magnitude of 0.

11. H magnitude from the 2MASS XSC. Additional targets are assigned an H magni- tude of 0.

12. K magnitude from the 2MASS XSC. Additional targets are assigned a K magni- tude of 0.

13. bJ − rF colour.

14. bJ − K colour.

15. Source identification from the NASA extragalactic database.

16. Source classification from the SuperCOSMOS database. 1=Extended, 2=Stellar. 17. Position offset between the NVSS catalogue and the 2MASS XSC for primary tar-

gets and between the radio catalogue and the SuperCOSMOS database for addi- tional targets.

18. 1.4 GHz NVSS flux density in mJy. Values were obtained from the NVSS catalogue for point sources and slightly extended sources. The flux densities of very extended sources with complex radio morphology were determined by integrating the radio emission inside a polygon drawn by hand around each source using thecgcurs

routine inmiriad(Sault et al., 1995).

19. Position offset between the SUMSS catalogue and the 2MASS XSC for primary target sources and between the radio catalogue and the SuperCOSMOS database for additional targets.

20. 843 MHz SUMSS flux density in mJy. Values of 843 MHz SUMSS flux density were determined using the method described for 1.4 GHz NVSS flux density above. 21. 60 µm flux density in Jy.

22. Flag denoting if the source is an additional target. ‘Y’ denotes an additional target and ‘N’ denotes a primary target.

23. Spectral classification according to the scheme in Section 3.6.1.

24. Flag denoting if the source was accepted as a true radio identification by the method outlined in Section 3.4.2 for primary targets and Section 3.5.2 for additional targets. 25. Comments on the source made during spectral classification and radio-optical iden-

tification.

The htmlversion of the table also contains links to an image of the 6dF spectrum,

2MASS J,H and K images of each source if available and a SuperCOSMOS bJ image of

3.7

Summary

This chapter has presented an introduction to the 6dF instrument and the 6dF Galaxy Survey and has described the method by which radio sources have been identified in the galaxy survey’s primary sample, as well as how radio sources to be observed serendipi- tously as additional targets have been selected. Radio source spectra were then classified and a database containing data at many wavelengths was created for further analysis. In general, for both crossmatching and spectral analysis, the human eye was found to be preferable and more reliable than machine-based methods for classification. 16% of pri- mary targets in the galaxy survey were identified with NVSS radio sources. These sources comprise 2.5% of the input radio catalogue.

The database described here comprises the largest and most homogeneous set of red- shifts and spectra of radio sources in the local universe ever obtained. Results from analy- sis of this sample are presented in the remaining chapters of this thesis. Chapter 4 contains analysis of the additional targets, Chapter 5 presents analysis of the NVSS identified pri- mary targets including a determination of local 1.4 GHz luminosity function and Chapter 6 presents results from a determination of the 2-point correlation function using a sub- sample of 6dFGS-NVSS primary targets in the first data release.

Chapter 4