Astron. Astrophys. 359, 113-130 (2000)

Appendix A: HRI X-ray source detection

Source detection was done using EXSAS (Zimmermann et al. 1997), and the standard command DETECT/SOURCES. This command generates a local source detection by a sliding-window technique followed by a maximum likelihood test which compares the observed count distribution on the full resolution image (pixel of 0.5") to a model of the point spread function (PSF) and the local background (Cruddace et al. 1988). The "likelihood of existence" is defined as , where the probability of the null hypothesis that the observed distribution of counts is only due to a statistical background fluctuation; provides a maximum likelihood measure for the presence of a source above the local background. We take =6.8 as detection threshold (; or for Gaussian statistics) as argued in CMFA.

For source detection, the HRI report (David 1997) advises to screen out lower and higher HRI Pulse Height Analyzer (hereafter PHA) channels, which are found to have the highest background. However, source counts should always be determined using all the 1-15 channels to cancel out the uneven HRI efficiency distribution across the detector area (S. Döbereiner, private communication). We decided to search X-ray sources above a fixed detection threshold in channels 1-15 and 3-8. ¹⁶ EXSAS gave us list of X-ray detections with positions, one sigma error box (), , and count rate. We removed sources detected in channels 1-15 but not in 3-8, considering these detections as spurious. For instance, there are two hot spots in the Core A observation in the South-East corner of the HRI . Since hot spots are usually considered as spurious detections, this criterion automatically removes them.

The astrometry must be corrected from offsets of typically a few arcseconds due to the time-dependent boresight error in the ROSAT aspect system. To do this, IR or optical counterparts in a 10" radius circle around the X-ray sources were searched. We then selected a sample of X-ray sources with an unambiguous counterpart and 1 (half width) error box , comparable to the IR/optical typical position error box (). Then, offsets in right ascension () and declination () were estimated by individual offset weighted mean: , with . We found offsets ranging from -0.5" to 2.5". We subtracted these offsets, and checked the quality of our astrometry by estimating sample residuals mean before and after offsets subtraction: . We found ranging from 0.5" to 1.2". and () were then quadratically added to to obtain an error box radius after astrometric correction (). In the case of the three different core F observations, the images were aligned and merged after astrometric correction to obtain a single deep HRI exposure of 77.2 ks. Source detection was subsequently performed as described in the article.

Table A1 lists the HRI X-ray sources, for which we adopt, in Col. 1, the same acronym as in CMFA: "ROXR" (for Oph X-ray ROSAT source), followed by "A" or "F". ¹⁷ Fig. A.1 indicates the source numbering.

Fig. A1. Sources designations from Table A1-B1. The image of Fig. 2 was smoothed using an edge detection algorithm.

Table A1. HRI X-ray sources in the Oph cloud cores A & F.

Table A1. (continued).
Notes:
gives the error box; is the likelihood of existence, we give the maximum value for the observation set.

In order to allow easier comparisons with previous work, X-ray source positions are listed in both J2000 and B1950 equinoxes, with their error box, in Cols. 2-6. The likelihood of existence is in Col. 7. Count rates are indicated in Col. 8-11. For the core F field, the indicated positions () and values correspond to observation (#1, #2, or #3) where and the position accuracy are the best, i.e. when the count rate is highest. When an X-ray source is detected in one observation above the detection threshold, and not detected in other observations, we have estimated the corresponding count rate upper limits (3.25), using the EXSAS command COMPUTE/UPPER_LIMITS. We have noted that the detection efficiency degrades with increasing angle to the axis, in the same way as the point spread function (this is discussed in 5.2).

Appendix B: optical/IR counterparts of the HRI X-ray sources

We searched stellar counterparts for the 63 ROSAT   HRI X-ray sources on the ESO/SERC second Digitized Sky Survey (DSS2). Fig. B1 gives the finding charts with BKLT IR sources for each of the 63 ROSAT   HRI X-ray sources.

Fig. B1. Optical finding charts of the 63 ROSAT   HRI X-ray sources of Tables A1 and B1. Each map is a extracted from the ESO/SERC sky survey red Schmidt plate using the second Digitized Sky Survey (one pixel=1"); North is at the top, East at the left. Circles show the ROSAT   HRI 90 confidence error boxes (i.e. one sigma error box from Table A1 multiplied by 1.6). Asterisks show the BKLT infrared sources 90 confidence error boxes ().

Fig. B1. (continued)

Fig. B1. (continued)

Fig. B1. (continued)

Table B1 gives identification lists for the two fields, and cross-identification with other surveys. Col. 1 is the ROXR numbering from detection (Table A1). Cols. 2-4 are respectively cross-identification lists with the X-ray sources of CMFA (ROXR1), Casanova (1994; ROXR2), and Kamata et al. (1997). Dots mean "X-ray source undetected", and dash "out of observation field". Col. 5 gives the first name attributed to this counterpart.

Table B1. HRI X-ray sources counterpart identifications in the Oph cloud cores A & F.

Table B1. (continued)
Comments: = see notes below. ROXRA or ROXRF = X-ray source number (this article). ROXR1 = CMFA X-ray source number. ROXR2 = Casanova (1994) X-ray source number. ASCA = Kamata et al. (1997) X-ray source number. Dash = out of observation field. Dots = unobserved source. red = ISOCAM source with IR excess. blue = ISOCAM source without IR excess. nII = new class II. nIII = new class III.? = X-ray detected source for which intrinsic X-ray luminosity cannot be determined. = X-ray undetected source. II-III? = class II or class III candidate (see Appendix B).
Notes:
ROXRA1: (7) The IR index (=dlog()/dlog) is estimated between 2.18-4.69 µm from Walter et al. (1994) and Jensen et al. (1997): we find =-2.5. (8-9) These values are estimated from Walter et al. (1994).
ROXRA2: (7) Martín et al. (1998) classify these stars as CTTS. (8-9) These values are estimated from GWAYL.
ROXRA4: (2) ROXR1-5 was identified with IRS3 and IRS5 (CMFA).
ROXRA8: (5) The position of Chini8 (Chini 1981) is 33" away from this optical star, the good position is given in Table 1 of Strom et al. (1995) by source number 3. (7) The ISOCAM LW2 and LW3 upper limits exclude an IR excess (see Appendix D).
ROXRA9: (2) ROXR1-12 was identified with an anonymous optical star (CMFA).
ROXRA13: (2) ROXR1-15 was neither optical nor IR counterpart (CMFA).
ROXRA16: (3) ROXR2-16 was from ROXRA16 (identified with WSB28; CMFA), ie only two error boxes away, but Martín et al. (1998) identify ROXR2-16 with another star.
ROXRA17: (2) ROXR1-22 was identified with GSS30-IRS1, 2, 3 (CMFA).
ROXRA21: (7) S1 is an embedded B-type star (Wilking et al. 1989; André et al. 1988). (8) The visual extinction comes from the data of Lada & Wilking (1984), see André et al. (1988). (9) This B3-B5 stellar luminosity is taken from André et al. (1988).
ROXRA26: (5) In the BKLT survey this well known emission line star appears to be a separation binary (this can also be suspected from the finding chart): the main component is B162710-241914 (J=8.56), the second one is B162710-241921 (J=11.27).
ROXRF1: (5) The IR star B162623-244308 is also in the 90 confidence error box but its J-band luminosity (15.65) is lower than that of the well known emission line star DoAr25=B162623-244311 (9.29).
ROXRF7: (2) ROXR1-22 was identified with SR24N and SR24S (CMFA). (5) In our observation #2 and #3 the counterpart of ROXRF7 is clearly associated with SR24S. In our observation #1 due to the weakness of the X-ray source the situation is less clear. We associate this source with SR24S.
ROXRF9: (2) ROXR1-36 was ambiguously identified with GY193 and GY194 (CMFA).
ROXRF12: (5) A weak star is visible in the 90 confidence error box on the DSS2 image, but this star is neither detected by the BKLT survey, nor by the PMM USNO-A1.0 catalogue.
ROXRF14: (3) ROXR2-22 was identified with ROXs20A and ROXs20B (CMFA).
ROXRF19: (6) This source is just at the border of the ISOCAM survey. As a part of its flux is lost, Bontemps et al. (2000) do not characterize this source. (7) Martín et al. (1998) classify these stars as CTTS. We thus consider this source as class II.
ROXRF21: (2) ROXR1-43 was identified with GY263 and IRS43 (CMFA). (8) Best value from Grosso et al. (1997). (9) This is the bolometric luminosity (see Grosso et al. 1997).
ROXRF23: (2) ROXR1-46 was identified with an unnamed optical star (CMFA), probably the star B162730-244726. (5) We identified this X-ray with the IR star B162728-244803 just at the border of the X-ray error box.
ROXRF24: (5) This source is red in the PMM USNO-A1.0 Catalogue (Monet et al. 1996) with and .
ROXRF25: (2) ROXR1-48 was identified with GY280, GY290, and GY291 (CMFA).
ROXRF28: (5) The IR star B162735-244532:B is also in the 90 confidence error box but its J-band luminosity () is lower than GY296=B162735-244532:A (12.62).
ROXRF37: (7) The Hipparcos distance is 75 pc: this star is a foreground F2V star.
References:
B = Barsony et al. (1997). BBRCG = Barsony et al. (1989). DoAr = Dolidze & Arakelyan (1959). El = Elias (1978). GSS = Grasdalen et al. (1973). GY = Greene & Young (1992). HD = The Henry Draper catalogue (Draper 1918). IRS = Wilking et al. (1989). ROXC = Montmerle et al. (1983). ROXs = Bouvier & Appenzeller (1992). S = abbreviation for "Source" in Grasdalen et al. (1973). SKS = Strom et al. (1995; Table 1). SR = Struve & Rudkjobing (1949). U = The PMM USNO-A1.0 Catalogue (Monet et al. 1996). VSS = Vrba et al. (1976). VSSG = Vrba et al. (1975). WL = Wilking & Lada (1983). WSB = Wilking et al. (1987). YLW = Young et al. (1986).

In the core A field, we find 26 X-ray sources, of which only one (ROXRA10) remains without optical or IR counterpart. Of the 25 identified X-ray sources, 22 were seen with the ROSAT   PSPC , and 4 are new detections (ROXRA3, 10, 16, 22). In the core F field, we find 37 X-ray sources, including 7 without optical or IR counterpart. Of the 30 identified X-ray sources, 18 were seen with the ROSAT   PSPC , and 12 are new detections (ROXRF3, 8, 12, 15, 18, 19, 24, 26, 28, 32, 35, 36). Altogether, 63 X-ray sources are detected, and 55 are identified. Of the 55 identified X-ray sources 40 are PSPC sources.

For sources with a low statistical significance (, or ; 3.25-3.9 for Gaussian statistics) we find X-ray sources with and without optical or IR counterparts. The X-ray sources without counterparts are always weak sources and may be spurious detections (locally high background), and this may therefore also be the case for weak X-ray sources with counterparts in case of chance spatial coincidence. For instance in the Core A field (respectively Core F) there are 875 (resp. 1173) BKLT sources; this sample is dominated by background sources without detectable X-ray emission. To estimate the number of chance coincidences, we have placed in each field 10⁵ random X-ray source positions, and searched for each whether there is a BKLT source in a circle of 10" radius: we have then an estimate of the probability to find a BKLT counterpart by chance within 10" from a spurious X-ray detection. This probability is 0.044 (resp. 0.049) for Core A (resp. Core F), or approximately 1/20 for both; in other words one (resp. two) spurious source identification are expected for the Core A (resp. Core F) field. As we have for Core A (resp. Core F) two (resp. 15) X-ray sources with out of 26 (resp. out of 37), this implies that one weak X-ray source in Core A (resp. 13 in Core B) is real, which is consistent with the number of identifications. We are therefore confident that the identifications of weak X-ray sources with stellar counterparts are correct.

Appendix C: comparison between HRI and PSPC observations

Within the boundaries of our observation fields (core A and core F), there are 61 X-ray sources detected previously with the ROSAT   PSPC (53 from CMFA, and 8 from Casanova 1994). However, 21 X-ray sources are not detected with the ROSAT   HRI . This difference could be explained by lower observational sensitivity and/or source variability. To elucidate this point, we must estimate HRI count rates for PSPC sources, and compare them with the adopted HRI detection threshold (3.25 ).

We first estimate the conversion factor between the PSPC count rate in the energy range of CMFA (1.0-2.4 keV) and the HRI count rate in the whole energy range (0.1-2.4 keV). We did not select X-ray sources with ambiguous PSPC detection (10 sources with notes in Cols. 2-3 of Table B1). For core F observation, the lowest detection count rate was taken to minimize variability effects (four sources with upper limits are not selected): we kept only 26 X-ray sources. Since many of these sources are variable (as shown by the core F observations), a conversion factor estimator insensitive to extreme values of the sample is needed. This is why we take the median of the PSPC /HRI count rate ratio, instead of the mean. We find count rate = 2.4 HRI count rate.

Fig. C.1 displays the HRI count rate vs. the PSPC count rate. It shows two classes of sources: sources near the median, and sources beyond the median (with error bars). The dispersion of points (within 1 rms) around the median value could be due to X-ray extinction effect on the conversion factor or to a variability factor . Preibisch et al. (1996) calculated the conversion for the whole energy band of the ROSAT   HRI assuming optically thin plasma emission with K and different values for the X-ray extinction: for increasing from cm^-2 to 10²² cm^-2, the conversion factor decreases from 2.5 to 2.0. Our observational estimate is in agreement with these values, which also show that the dependence of the conversion factor on X-ray extinction is small compared to the dispersion of count rates and can be neglected in our plot. We conclude that the dispersion is due to variability: WL20, GSS37, VSS27, and SR9 must have been in a high state during the PSPC observation, as were ROXs4 and SR2 during the HRI observation, the other sources being essentially unchanged in both observations.

Fig. C1. Plot of ROSAT   HRI (this paper) vs. PSPC (CMFA) count rates and sources variability. The dashed line is the median value of the conversion factor between HRI and PSPC count rates (2.4); dotted lines show the dispersion (1 rms) around this median value. Count rate error boxes are shown for sources outward the dotted lines. Sources below (above) the dotted lines presumably flared during PSPC (HRI ) observations (see text for details).

Using this conversion factor, we can estimate the HRI count rate from the PSPC count rate, and compare it with our HRI threshold computed with EXSAS. We find that 14 sources are below this threshold (ROXR1-1, 6, 7, 16, 20, 21, 27, 33, 34, 37, 40, 53, and ROXR2-16, 18), and the other 7 sources were in a high state during the PSPC observation (ROXR1-19, 30, 45, 47, and ROXR2-27, 30, 33). We conclude that the non-detection of the 21 PSPC sources by the HRI can be fully explained by the difference in sensitivity and intrinsic variability.

We can also compare our detections with ASCA (Koyama et al. 1994; Kamata et al. 1997; see observation field in Fig. 1) despite its lower angular resolution. Kamata et al. (1997) detected 19 X-ray sources, of which 10 were previously observed by Koyama et al. (1994). Compared to Einstein Observatory and ROSAT   PSPC observations, 7 new X-ray sources were discovered by ASCA . ¹⁸ These 7 X-ray sources are also not detected with the HRI . On the 12 sources already observed by Einstein Observatory and ROSAT PSPC , we detect 9 sources, the 3 others being below our sensitivity threshold according to the above conversion factor.

Appendix D: optical/IR counterpart without IR classification

Nine X-ray sources have optical/IR counterpart for which the IR classification is not known. Three of these X-ray sources are found in the ISOCAM survey, but with only upper limits in LW2 and LW3 filters. We give here their spectral energy distribution (see Fig. D.1), and discuss their possible IR classification. In case of doubt, the resulting Class II (or Class III) source candidates have not been included in the statistic studies of this article.

Fig. D1. Spectral Energy Distribution of optical/IR counterparts without IR classification. B and R data come from the PMM USNO-A1.0 catalogue (Monet et al. 1996); J, H, K data come from BKLT; LW2 and LW3 are the ISOCAM upper limit at 10 mJy from Bontemps et al. (2000). The dashed line shows the limit between Class II source and Class III source classification (=dlog()/dlog=-1.5 from AM). "out" means that the source is outside the ISOCAM survey.

ROXRA8: The counterpart of this X-ray source is the optical star Chini8=SKS3 (, ). The ISOCAM LW2 and LW3 upper limits exclude an IR excess. We classify this source as a Class III source.

ROXRF12: A weak star is visible in the 90 confidence error box on the DSS2 image, but this star is neither found in the BKLT survey, nor in the PMM USNO-A1.0 catalogue. The low optical/near IR magnitudes imply a low luminosity for this object. We propose this source as a weak Class II or Class III source candidate detected during a strong X-ray flare. This source may also be a brown dwarf.

ROXRF18: The counterpart of this X-ray source is the IR star B162720-243820 (). This star is visible in the DSS2 (red) optical image, but it is not in the PMM USNO-A1.0 catalogue (probably because only stars appearing both in blue and red images were accepted), thus we have no estimate of its B and R magnitudes. The low near IR magnitudes imply a low luminosity for this object. This source may be a weak Class II or Class III source candidate detected during a strong X-ray flare. ISOCAM LW2 and LW3 upper limits do not exclude an IR excess for this object. This object can also be a weak Class I protostar with a strong X-ray flare.

ROXRF23: The counterpart of this X-ray source is the IR star B162728-244803 ; see note in Table B1). The low near IR magnitudes imply a low luminosity for this object. The SED of this source peaks in the H-band. We propose this source as a weak Class II or Class III source candidate detected during a strong X-ray flare.

ROXRF24: The counterpart of this X-ray source is an optical star () named only in the PMM USNO-A1.0 catalogue (Monet et al. 1996). Unfortunately, this object lies outside the BKLT survey. We propose this source as Class II or Class III source candidate.

ROXRF25: The counterpart of this source is BBRCG50 observed only in K-band (). This source is not retrieved in BKLT. The ISOCAM LW2 and LW3 upper limits exclude a strong IR excess. We propose this source as Class II or Class III source candidate.

ROXRF34: The counterpart of this X-ray source is the optical star WSB58=B162800-244819 (, ). Wilking et al. (1987) noted a probable H detection needing confirmation. The SED of this source peaks in the H-band. We propose this source as Class II or Class III source candidate.

ROXRF35: The counterpart of this X-ray source is the optical star B162800-245340 (, ). The SED of this source peaks in the J-band. We propose this source as Class II or Class III source candidate.

ROXRF36: The counterpart of this X-ray source is the optical star B162812-245043 (, ), which appears to be a close binary () in the second Digitized Sky Survey (see Fig. B1). The SED of this source peaks in the H-band. We propose this source as Class II or Class III source candidate.

© European Southern Observatory (ESO) 2000

Online publication: June 30, 2000
helpdesk.link@springer.de