• No se han encontrado resultados

4. Interpretación de la información

4.1 Centro Educativo Esperanza Galicia

4.1.4 Maestra 4

We have recently started to extend MUNICS to higher redshift. This new project, called MUNICS-Deep, will image one square degree to a depth two magnitudes fainter than MUNICS. While MUNICS is limited to studying galaxy evolution at z<

∼1.2, the

new dataset will allow to probe the changes in the field galaxy population in the im- portant redshift range 1<

z<∼2, thus building a bridge between medium-deep surveys

and the regime of the Lyman-break galaxies (see Section 1.2 for an overview). Ob- servations for MUNICS-Deep are still in progress, but promise exiting new results in observational cosmology.

Chapter 3

A Near-Infrared–Selected Galaxy

Redshift Survey: Design and

Observations

The main scientific results of my PhD thesis rely on a near-infrared selected spectro- scopic redshift survey. This chapter describes how the targets for spectroscopy were selected and how the observations were carried out.

3.1

Selection of the Spectroscopic Sample

Objects for spectroscopic observations were chosen from the K-band selected photo- metric catalogue of MUNICS in ten survey fields, the details of which can be found in Table 2.2 (see Chapter 2 for a description of the field nomenclature). The K-magnitude selected field galaxy spectroscopy programme was carried out in the fields S2F1, S2F5, S5F1, S6F5 and S7F5, while the spectra available in the other fields originate from a dedicated search for faint red AGN.

Object selection for spectroscopy was based on two criteria. Firstly, we aimed at a K-band magnitude-limited sample. Due to the use of optical spectrographs, the ap- propriate K-band limit is determined by the typical colours of red galaxies (roughly

RK'4, see Figures 5.4 and 5.5) and the limits of the optical spectrographs at the telescopes we used. Trying to keep the K-band completeness of the spectroscopic sam- ple as high as possible yields sample limits of K≤17.5 for spectroscopic observations at the Calar Alto 3.5-m telescope and K <19.0 for observations at the Very Large Telescope (VLT). Obviously, a small fraction of very red objects will be lost, but their number density is comparatively small anyway (see, for instance, Martini (2001) and references therein). Results on the few very red objects identified spectroscopically can be found in Section 5.8.

Secondly, in selecting objects for spectroscopy, we have tried to exclude stars. This was done using the image-based classification of objects into point-like objects and ex- tended sources as described in MUNICS I. For the classification the point-spread func- tion (PSF) for each image and each filter is determined. Then the intensity distribution

36 CHAPTER3. A NEAR-INFRARED–SELECTED REDSHIFT SURVEY

Table 3.1: Summary of results of the test of the image-based classification method for a sample of objects with K016.5 in two survey fields (upper part of table) and for the whole spectroscopic sample (lower part of table). The numbers quoted in the table give the number and fraction of objects with stellar or galactic spectrum among objects classified as point-like or extended, respectively.

Morphological classification Spectral classification

Star Galaxy

Sparse sample with K016.5:

Point-like 39 100 per cent 0 0 per cent

Extended 11 20 per cent 44 80 per cent

Complete spectroscopic sample:

Point-like 64 79 per cent 17 21 per cent

Extended 28 7 per cent 397 93 per cent

of every object in the images is compared to the PSF, and the objects are classified into stars (PSF-like sources) and galaxies (extended sources). Finally, this classifica- tion information, which is available for all objects in all six filters, is weighted by the signal-to-noise ratio of the object in each filter, and the final classification is done.

To test this classification procedure, we have built up a small spectroscopic test sample of bright objects which was purely magnitude selected and therefore contains both stars and galaxies. The results of the comparison between image-based and spec- troscopic classification show that our morphological approach is able to distinguish stars and galaxies with reasonable reliability. The results of this test are summarised in Table 3.1.

On a purely magnitude-limited sample containing all objects with K016.5, all objects classified as point-like are indeed stars, and the stellar contamination is 20 per cent only. On the other hand, we can also use the complete spectroscopic catalogue to investigate the reliability of our classification method, yielding also very good agree- ment between spectral and morphological classification.

Thus our method to classify objects into point-like or extended sources can reliably reduce stellar contamination of the spectroscopic catalogue as long as the objects are not too faint. Pre-selecting our spectroscopic sample by image morphology results in a loss of less than 4 per cent of all galaxies which are classified as point-like, and in a 7 per cent contamination by stars.

The quality of the classification procedure is also demonstrated in Figure 3.1, where one can see that the vast majority of all objects on the stellar sequence are indeed classified as point-like. In Chapter 5 we will show that the fraction of stars in our spectroscopic sample is indeed much lower than in comparable surveys.

3.2. SPECTROSCOPIC OBSERVATIONS 37 12 14 16 18 20 0 1 2 3 4 5 K [mag] J−K [mag] 0 5 10 15 20 25 0 5 10 15 20 25 χ2 galaxy χ 2 star

Figure 3.1: Illustration of resulting object classification in point-like (filled symbols) and extended sources (open symbols). Left-hand panel: The locus of objects in the

JK vs. K diagram for two MUNICS mosaic fields. The sequence of stars at JK'1

indeed contains mostly point-like sources. Right-hand panel: A comparison of the morphological classification to aχ2criterion based on fitting template spectral energy distributions of stars and galaxies to the broad-band photometry of the objects in the MUNICS field S2F1. Again, the agreement is very good.

Documento similar