Astron. Astrophys. 353, 457-464 (2000)

3. Results

A contour plot of the combined I-band image of the 315x315 arcsec² field surrounding Q1205-30 is shown in Fig. 1. Q1205-30 itself is here marked by a `q' and an arrow. The star from which the central core of the Point Spread Function was defined (see Sect. 3.1) is marked `PSF'. Six candidate Ly emitting galaxies (see Sect. 3.3) are marked "+".

Fig. 1. Contour plot of the I-band frame showing the 315x315 arcsec2 field surrounding Q1205-30. North is up and east to the left. Q1205-30 is shown by the arrow next to the "q". The star used to define the PSF is marked `PSF' (see Sect. 3.1). The positions of six candidate Ly emitting galaxies are marked with a "+" (see Sect. 3.3).

3.1. Objects near the line of sight towards Q1205-30

Q1205-30 was selected for observation because of the high column density absorber seen along the line-of-sight towards it. Previous detections of high column density absorbers (for a recent summary see Moller & Warren 1998) suggest that their absorption cross-section is very small, and that emission from the object therefore likely will be hidden under the quasar PSF. To search for emission from the absorbing object we therefore performed a careful PSF subtraction as detailed in this section.

The basic principles of the PSF subtraction are as described in Moller & Warren (1993a), and Fynbo et al. (1999a). For a first approximation to the PSF we used the star marked `PSF' in Fig. 1. This is the nearest unsaturated point source significantly brighter than Q1205-30, but as is evident from Fig. 1 it has four nearby neighbour objects (projected distances in the range 4-9 arcsec). The signature of those four objects does not affect the core of the PSF. Their effect on the halo of the PSF was masked out and removed via substitution of the masked regions by areas selected from scaled high S/N halos from isolated point-sources brighter than the PSF-star (and hence saturated in the central core). The comparatively large projected distances (4 arcsec and larger) means that the presence of those objects, even if not corrected for, would not in any case have had any effect for the results reported below.

The DAOPHOT-II (Stetson 1997) extension ALLSTAR was used to perform the final PSF model fit and subtraction. Identical procedures were followed for the PSF subtraction in all combined images: Narrow band, B and I.

In Fig. 2 we show a 18x18 arcsec² cut-out of the I-band image centred on Q1205-30 before (a) and after (b) subtraction of the quasar PSF. The quasar is marked "q", and it is clearly seen that two faint galaxies (named g1 and g2) were blended with the quasar PSF. In Fig. 2b we have drawn a circle of radius 1.3 arcsec around the centre of the subtracted PSF. Inside this circle the large residuals (not shown) from the PSF subtraction make it impossible to search for objects. The galaxy g2 is well separated from the PSF-subtraction residuals so its projected distance from Q1205-30 (2.8 arcsec) is well determined. In contrast, g1 is partly embedded in the residuals. The galaxy g1 could therefore in reality be an elongated object extending across the quasar, and the measured projected distance (2.2 arcsec) is hence an upper limit to the true projected distance. The measured I-band centroids of the galaxies g1 and g2 are marked by an "" in Fig. 2b.

Fig. 2a and b. Contour plot of the I-band frame showing an 18x18 arcsec2 field surrounding Q1205-30. North is up and east to the left. The frame has been smoothed by a 3x3 pixel boxcar filter. Left: final combined frame (QSO is marked q). The contour levels are at -9, -6, -3, 3, 6, 9... in units of the background noise, with the dotted contours being negative. Right: same as left but here after subtraction of the quasar PSF, revealing two galaxies marked g1 and g2 at separations 2.2 and 2.8 arcsec (from the QSO) respectively. Their centroids are marked "". A circle of radius 1.3 arcsec has been drawn around the centre of the subtracted PSF (see text for details).

In Fig. 3a,b we show the same 18x18 arcsec² cut-out of the B-band image. Here we have drawn a circle of radius 1.5 arcsec, again to mask out the area where PSF subtraction makes it impossible to search for objects. The ""s here again mark the centroids of the galaxies g1 and g2 as measured on the I-band image. In the B-band image g1 is found at the same projected distance as in the I-band image while the extremely red object g2 is not detected.

Fig. 3a and b. Same as Fig. 2 but here for the B-band image. The contour levels are at -6, -4, -2, 2, 4, 6... in units of the background noise with dotted contours being negative. Right: Here a circle of radius 1.5 arcsec has been drawn around the centre of the subtracted PSF. The ""s mark the centroids of the galaxies g1 and g2 as measured on the I-band image.

In Fig. 4a,b we again show the same 18x18 arcsec² cut-out as above, but here from the combined narrow band frame. The ""s here again mark the centroids of the galaxies g1 and g2 as measured on the I-band image. In this image we clearly see an extended object (named S6) to the north of the quasar, possibly extending all the way to the east of the quasar. Note that g1 here is found at a slightly larger projected distance from the quasar than in the I-band and B-band images. This illustrates the point we made above that the PSF subtraction will tend to "push" the object out at a larger distance, because part of the object is embedded in the non-recoverable central part of the PSF. This will then also impact the photometry, as the total flux of the object will be underestimated. The position of g2 (marked by `' and a dotted circle) corresponds to a faint protrusion on the extended object. Since the narrow band is within the B passband, we would indeed expect to see only very weak emission from the extremely red object g2.

Fig. 4a and b. Left: Same as Fig. 2 but here for the narrow band image. The contour levels are at -6, -4, -2, 2, 4, 6..., in units of the background noise, with the dotted contours being negative. Right: Here the PSF of the quasar has been subtracted, revealing a large extended object (named S6) to the north, possibly extending to the east, of Q1205-30. A circle of radius 1.5 arcsec has been drawn around the centre of the subtracted PSF. The dotted circle marks the extend of g2 in the I-band image, and the ""s mark the centroids of the galaxies g1 and g2 as measured on the I-band image.

3.2. Photometry of objects near the QSO line of sight

In this subsection we describe the photometry on Q1205-30, g1, g2 and S6 in the three filters B, I and narrow band. For g1 we measured the flux inside a circular aperture of diameter 3.5 arcsec. The resulting magnitudes presented in Table 3 will be somewhat underestimated since we miss the flux closer than 1.5 arcsec from the QSO. For g2 we also measured fluxes in circular apertures of diameter 3.5 arcsec. We did not detect g2 in B, and hence provide the 2 detection limit. The galaxy g2 is very red, consistent with being an old stellar population at a redshift of about 0.5 or more. In Sect. 4.1 we discuss the possible effects of g2 on the line of sight due to gravitational lensing. In the narrow band image we can not in an objective way determine whether the flux detected at the position of g2 originates from g2 or from S6. Hence we chose to consider two extreme cases for the photometry of S6 and g2 in the narrow filter.

Table 3. Photometric properties of the QSO, of the faint galaxies g1 and g2 and of the Ly emitter S6. Upper limits to magnitudes are 2 except for g1, where the limit is determined from model A mentioned below.

In model A (minimum Ly flux model; Fig 4a) we subtracted the maximum flux we can possibly ascribe to g2 in the narrow band. We used a model of g2 obtained by smoothing the I band image of g2 to the seeing of the narrow band image, and then determined the maximal scaling of this model which after subtraction left residuals consistent with the noise. The remaining flux, making no correction for flux from S6 within the central core of the subtracted quasar PSF, was assigned to S6. In model B (maximum Ly flux model; Fig 4b) we assumed that all the flux seen north and east of Q1205-30 originates from S6 and not from g2. In order to estimate the flux within the central core of the quasar PSF we assumed that S6 is symmetric about its major axis (PA 120^o east of north). A model of S6 was made as follows. First we flipped the image of S6 around its major axis. In the PSF subtracted image we then replaced the region inside radius 1.5 arcsec from the quasar PSF with the flipped image. The flux of the model was measured using a circular aperture of diameter 4.9 arcsec.

The impact parameter of S6 was found to be 1.8 arcsec in model A and 1.5 arcsec in model B. The Ly flux of S6 is 6.60.6 ergs s^-1 cm^-2 (model A) and 7.90.7 ergs s^-1 cm^-2 (model B).

3.3. Candidate Ly emitting galaxies in the field

The "narrow minus on-band-broad" versus "narrow minus off-band-broad" colour/colour plot technique (Moller & Warren, 1993a; Fynbo et al. 1999a) has proven a powerful tool to identify galaxies in the faintest end of the high redshift galaxy luminosity function (LF). We here describe the search procedure and the photometric measurements carried out with the aim to produce the plot shown in Fig. 6.

Fig. 5a and b. Two "extreme" models of S6. Left: in model A we assume that all the flux seen at the position of g2 is indeed from g2, and we make no allowance for flux from S6 within the central core of the quasar PSF. Right: in model B we assume that no significant flux is seen from g2, and we model S6 within the central core of the quasar PSF assuming that S6 is symmetric about its major axis with position angle 120o east of north.

Fig. 6. Two-colour diagram versus for objects detected at a S/N in the narrow band frame. The dashed line is a line of constant in the continuum (chosen to be that of Q1205-30), but with a spectral feature in the narrow filter ranging from a strong absorption line in the upper right hand corner to a strong emission line in the lower left hand corner. Three of the six candidate emission line galaxies are seen in the lower left corner. The remaining three are detected in neither B nor I. For these galaxies the lines with arrows indicate the 2 allowed region in the diagram. The dotted lines confine the expected region of objects with no special features in the narrow filter. The inserted diagram shows the region of the plot containing Q1205-30 (marked q) and objects with S/N.

Identification and photometric measurements of objects in the field was done with the photometry package SExtractor (Bertin & Arnouts, 1996). We used a minimum object extraction area of 8 pixels and a detection threshold of 1.3 above the background. As our detection image we selected to use a weighted average of the three combined frames. The weights were chosen to be the inverse of the variance in each of the combined frames. As a detection filter we used a Gaussian filter with a full-width-at-half-maximum (fwhm) of 4 pixels similar to the fwhm of point sources in the detection image, which is 3.4 pixels.

In total we detected 473 objects with a signal-to-noise ratio (S/N) in the narrow band. In Fig. 6 we show the colour-colour diagram versus for the entire sample. In this diagram objects with Ly emission in the narrow band filter will be located in the lower left corner while objects with absorption in the narrow filter will be located in the upper right hand corner. Since our narrow filter is centred at 4906Å, which is in the red wing of the B filter, the colour of an object will depend on the slope of the objects spectral energy distribution. Very red objects will, therefore, appear to have a slight excess emission in the narrow band, causing the locus of continuum sources to be slightly tilted towards the lower right of the diagram.

We determined the expected region for objects with no special feature in the narrow band by calculating the position in this diagram for a wide range of galaxy spectral energy distributions taken from the models of Bruzual & Charlot (1993) The galaxy models were calculated for several redshifts in the range to , and were corrected for Ly line blanketing due to intervening absorbers (Moller & Jakobsen 1990). The resulting region is indicated by the dotted line in Fig. 6.

Inspection of Fig. 6 clearly shows that the vast majority of the detected objects indeed conform to the predicted colours. Three objects to the lower left are, however, found to lie significantly outside the locus of continuum objects. All three are found in the region expected for blue objects with an emission line in the narrow filter.

The detection limit in the combined image used for object detection is dominated by the deeper broad band images. Faint objects with large equivalent width Ly emission might therefore be missed by the detection algorithm. To make up for this we repeated the detection procedure, but this time using the narrow band image for detection. SExtractor found six emission line objects with S/N. Three of those were the ones reported already from the combined image detection, while the other three remain undetected in broad band.

In addition to S6 we hence detect a total of six emission line objects at S/N. The positions of the six objects are marked by crosses in Fig. 1. Image cut-outs showing the 6 objects in each of the three bands are reproduced in Fig. 7. Photometric properties of the emission line objects are reported in Table 4. Magnitudes were measured using both SExtractor isophote apertures and large circular apertures. Emission line fluxes corresponding to the measured n(AB) aperture magnitudes range from erg s^-1 cm^-2 to erg s^-1 cm^-2.

Fig. 7. Extractions from the combined narrow band (top), B band (middle) and I band (bottom) for each of the six candidate emission line galaxies detected with S/N (see Sect. 3.3). In each field emission line galaxies have been marked with an arrow. The size of the square fields is 18x18 arcsec2.

Table 4. Photometric properties of the six candidate Ly emitting galaxies. Upper limits to magnitudes and lower limits to equivalent widths (EWs) are 2. SFRs are calculated assuming a =1 and =0 universe, and using the prescription SFR=L(Ly)/1.12 10⁴² erg s^-1.

As seen in the insert in the upper left corner of Fig. 6, Q1205-30 has only a slight excess emission in the narrow band filter which covers Ly of the quasar also. The weakness of this excess emission is the result of the blue wing of the Ly emission line of the QSO being absorbed partly by the Lyman limit absorber.

© European Southern Observatory (ESO) 2000

Online publication: December 17, 1999
helpdesk.link@springer.de