SpringerLink
Forum Springer Astron. Astrophys.
Forum Whats New Search Orders


Astron. Astrophys. 349, 97-107 (1999)

Previous Section Next Section Title Page Table of Contents

4. Structural properties

The presence of substructure in the ABCG 194 cluster of galaxies has been a subject of debate:

- Using an adaptive kernel algorithm, Kriessler & Beers (1997) determined the presence of substructure in the central core of this cluster with a probability level of 95%.

- By applying a multiscale analysis which couples kinematic estimators with wavelet transforms (Escalera et al. 1994), Girardi et al. (1997) classify the ABCG 194 cluster as unimodal with a very condensed core and two small-scale subgroups within the cluster.

- Beers & Tonry (1986) have suggested that this cluster shows good evidence for multiple X-ray substructure.

At a smaller scale - thus not really corresponding to substructures - from a 40´[FORMULA]40´ Digital Sky Survey image around the ABCG 194 cluster center and applying a wavelet analysis technique to the corresponding ROSAT PSPC image, Lazzati et al. (1998) detected 26 X-ray sources; four of these coincide with known galaxies in the cluster: NGC 547, NGC 541, NGC 564 and NGC 538.

4.1. Optical structure

An overall description of the Dressler map has been given in Sect. 3.1. Besides the large scale morphological aspect discussed above, in particular the bright extension along PA=50 which accounts for the "linear" aspect of the cluster, we also see a fainter extension towards the south-east.

This tends to indicate the presence of substructure. In order to qualify the degree of significance of this substructure, we have applied the spatial test developed by Salvador-Solé et al. (1993), which has been shown to be well suited to the detection of substructure in systems (Scodeggio et al. 1995). By applying this method to the total sample of 97 galaxies, we estimate two density profiles: the first one, [FORMULA], is obtained by inverting the density of numbers of pairs of galaxies, and the second one, [FORMULA], by inverting the density of numbers of galaxies at projected distances from the center of symmetry. [FORMULA] is therefore insensitive to correlations in galaxy positions. A difference among these two quantities is interpreted as indicating the presence of substructure in the system.

The distributions of [FORMULA] and [FORMULA] are displayed in Fig. 11; they are indistinguishable within error bars (estimated with a Monte Carlo technique). We have applied the statistical method proposed by Salvador-Solé et al. (1993) to test the significance of substructures at a scale chosen a priori. We find that already at a scale of 0.25 Mpc the probability to have substructure is less than 30%, and this probability strongly decreases with increasing scale. Therefore this method detects no significant substructures in the optical sample (see Salvador-Solé et al. 1993 for details). Note that the result is the same if we apply this method to a more complete magnitude limited subsample.

[FIGURE] Fig. 11. [FORMULA] and [FORMULA] profiles (see Sect. 4.1).

4.2. Substructures derived from kinematics

The previous analysis considers the cluster globally but does not take into account the dynamical properties of substructures (if any). We will therefore apply the Serna & Gerbal (1996) method, coupled with a study of the galaxy velocity distributions in the cluster.

This method sorts the galaxies according to their total energy (i.e. the sum of the potential and kinetic energies), leading to a dendogram where the total energy appears vertically. Pairs and groups of galaxies then appear with a lower total energy. The velocity density distributions were obtained using profile reconstructions based on a wavelet technique (instead of a histogram). The features obtained with this method are significant at various chosen levels above the noise (Fadda et al. 1998).

We applied this method to a sample of 97 galaxies defined as follows: the CGH velocity sample of 74 galaxies with both velocities in the cluster range and magnitudes; the BCG sample of 22 galaxies for which magnitudes were not available except for two which we found in the NED database; for the remaining 20 we assigned each galaxy the mean magnitude of the CGH sample: 15.2; one Seyfert galaxy (Knezek & Bregman 1998) to which we also assigned a magnitude of 15.2.

A first dendogram reveals the presence of six galaxies which appear to be only very loosely linked to the cluster. The velocities of these six galaxies are quite similar, but they are distant from each other in projection on the sky. The velocity distribution indicates a bump in the 6500 km s-1 region, well separated from the general distribution. We therefore chose to eliminate these objects in the following analysis, and are left with a sample of 91 galaxies; this sample leads to the velocity interval 4300-6200 km s-1 for galaxies belonging to the cluster, that is a somewhat narrower range than found by CGH. This sample will be referred to as the "All" sample, hereafter.

The velocity distribution of this new sample at various significance levels is displayed in Fig. 12a. The overall shape of this distribution is obviously not gaussian, but shows an asymmetry with an excess at high velocities. This is confirmed by the values of the skewness and kurtosis: see values for "All" in Table 4. Its mean velocity (which is also the median) is equal to that of the second brightest galaxy (NGC 541).

[FIGURE] Fig. 12. Left: velocity density distribution of galaxies. Top panel a : galaxies belonging to the "All" sample (see text); the three densities correspond to confidence levels: 3[FORMULA] (full line), 2.6[FORMULA] (dotted line) and 2.2[FORMULA] (dot-dashed line). Bottom panel b : galaxies belonging to the All (full line) and Main (dashed line) samples at a 3[FORMULA] level. c Right: dendogram obtained by the HTree method for 91 galaxies. The numbers at the bottom correspond to those of the groups defined in Table 4.


[TABLE]

Table 4. Characteristics of the subgroups identified from the dendogram in Fig. 12c. Columns have the following meaning: (1) group name, (2) number of members in group, (3) average radial velocity of group in km s-1, (4) velocity standard deviation in km s-1, (5) skewness, (6) kurtosis.


The dendogram obtained after excluding these six galaxies (see Fig. 12c) reveals the presence of subgroups. Groups 1 to 5 (in order of increasing mean velocity) come out of the sample easily. We have constructed a "Main" structure by taking out these five groups from the "All" sample. The statistical characteristics for these various subgroups are presented in Table 4 and a map of group positions in the overall field is shown in Fig. 13.

[FIGURE] Fig. 13. Positions of subgroups in the overall cluster field. The numbers of the groups indicated correspond to those in Table 4.

Group 3 appears to be the central subsystem, at the bottom of the gravitational potential well of the cluster. It contains the three brightest galaxies (including NGC 541 and NGC 547 which are both radiosources) which show a high level of boundness. Notice the presence of three other galaxies which appear to be bound together, among which the X-ray galaxy NGC 538, and also that of the Seyfert galaxy.

Groups 1 and 4 are well defined both spatially and in velocity space (their velocity dispersion is small). Group 4 contains the X-ray galaxy NGC 564.

Groups 2 and 5 contain only a few galaxies with small radial velocity dispersions and average intergalactic distances. Note that group 2 is far from the cluster center and appears to be weakly bound to the bulk of the cluster, as seen on the dendogram.

In comparison with the "All" sample, the "Main" structure velocity density distribution appears more gaussian, with a mean velocity equal to that of the overall sample within the error bars, but a smaller velocity dispersion and skewness (see Fig. 12b). The center of the "Main" sample is displaced towards the north east relatively to the "All" sample.

Our results can be compared to those found in the literature and based on various methods. Using a technique combining wavelets and kinematical information, Girardi et al. (1997) determined the small-scale structure in the ABCG 194 cluster and found the same group as our group 2, and a group which, by coordinates and radial velocity coincides with our group 3. Using two other different methods to extract groups, Garcia (1993) also found our group 3. This group therefore appears to be at the bottom of the cluster potential well, while the other groups described in Table 4 appear more like pairs of galaxies with possible satellites.

Notice that the "linear" structure of the cluster corresponds to our group 3, while the extension observed towards the south east in the Dressler map (Fig. 5) may correspond to our group 4.

The velocity dispersion profile (VDP) is shown in Fig. 14 for the "All" sample. The binning has been performed in ellipses with their major axis along the direction PA=50 and with an axial ratio [FORMULA]; the result is shown for bins of equivalent radii 400 and 800 arcsec. The shape of this profile does not change (within the error bars) if the subclusters are excluded. The VDP "inverted" shape shows an increase with radius up to [FORMULA] arcsec, then a decrease and is comparable to that derived by den Hartog & Katgert (1996) for this cluster. These authors interpret such a profile as originating from a relaxed region.

[FIGURE] Fig. 14. Velocity dispersion as a function of radial distance to the cluster center; the equivalent radius of each bin is equal to 400 arcsec for the bold curve and to 800 arcsec for the light one.

4.3. X-ray structure

We now compare these results with the X-ray features derived first from the wavelet analysis and from the "detect" software developed by Snowden, and second from the pixel by pixel fit.

The wavelet analysis of the X-ray image reveals no substructure at middle and large scales. At the two smallest scales (2 and 4 pixels, or 30 and 60 arcsec), 9 components (at least) are detected at a 3[FORMULA] level (see Fig. 6b). The Snowden method gives 38 sources in the same field, among which 9 in common with ours. The difference between the numbers of detections reached by both methods raises the question of their validity. The Snowden method is based on a convolution by the PSF (which varies with radius), and is therefore close to the wavelet method. However, the statistics on which the detection thresholds are determined for both methods are not the same. Our wavelet software estimates the image characteristic statistics by analyzing the whole image (see e.g. Slezak et al. 1994) while Snowden considers only circles containing between 90% and [FORMULA] of the encircled energy radius of the off-axis PSF. The average background is taken into account only in the cases where there are less than 4 counts per annulus. This may explain the difference between the numbers of sources. Note also that the purpose of both methods is not the same: Snowden's software is aimed at detecting everything that is not diffuse X-ray emission, in order to eliminate these point-like sources, and therefore it tends to find a higher number of small sources (including possible cosmic rays), while our purpose is to detect only X-ray sources above a certain significance level.

Source A corresponds to one of the two "dumbbell" galaxies, which are the first and third brightest cluster galaxies; however the X-ray PSPC pixel size is too large to be able to discriminate between both galaxies. Source B can be identified with the second brightest galaxy in the cluster. Source C coincides with the Seyfert galaxy reported by Knezek & Bregman (1998). The radiosources 0123-016AB are also strong X-ray emitters (Burns et al. 1994). Note that these sources have velocities in the cluster. Source F appears to be a star (it is star-like in the POSS and there is no quasar at this position in the Véron-Cetty & Véron 1993 catalogue). The other sources do not coincide with any galaxies with available redshifts. Note that in the part of the X-ray field shown in Fig. 6b, we detect all the sources reported by Lazzatti et al. (1998) except one (their source 26).

We have estimated the X-ray luminosity of the nine detected sources, proceeding in three steps:

  • we fit the whole data by a [FORMULA]-model as explained in Sect. 3.2;

  • we subtracted to the real image the synthetic image corresponding to the values obtained by the fit;

  • in the image thus obtained, we have counted the photons around the peaks corresponding to the detected sources and give the corresponding results in Table 5; we have checked that outside the sources, the background is around zero, as expected;

  • the luminosities were derived from the counts assuming a temperature of 1 keV, following the method described in Sect. 3.2.

The luminosities found for these sources are much lower than those given by Lazzati et al. (1998), but the reason for such a discrepancy is not clear. We have compared our source counts with the results given by the ROSAT SASS data processing pipeline, and find that they agree within 35% except for sources B and C. Note that the specific [FORMULA]-model used for this estimate is not very important.


[TABLE]

Table 5. Luminosity of the sources detected in the hard band (0.44-2.04 keV). The X-ray luminosity is computed with a 1 keV mekal model.


Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: August 25, 1999
helpdesk.link@springer.de