next previous
Up: ROSAT HRI catalogue of region


2 ROSAT HRI data

2.1 Data analysis

The LMC was observed by the ROSAT HRI in more than 500 pointings during the operational phase of ROSAT between 1990 and 1998. 543 observations with exposure times of 50 to 110000 s (Fig.1) in a field of 10$^\circ$ $\times$10$^\circ$ around RA = 05$^{\rm h}$ 25$^{\rm m}$ 00$^{\rm s}$, Dec = -67$^\circ$ 43$^\prime$ 20 $^{\prime\prime}$ (J2000.0) were used for the analysis. The analysis was carried out using three detection methods available in EXSAS (Zimmermann et al. 1994). For each pointing X-ray sources were searched using the sliding window methods with local background and with a spline fitted background map. The resulting detection lists were merged and a maximum likelihood algorithm was performed on this list. Sources were accepted if their likelihood of existence was larger than 10.0, i.e. the existence probability was higher than $P = 1 -
\exp$(- $ML_{\rm exi}$) = $1 - 4.5 \ 10^{-5}$, and their telescope off-axis angle smaller than 15$^\prime$ during the observation.

For point and point like sources the source extent was determined by the maximum likelihood technique fitting the source intensity distribution with a Gaussian profile. The count rates resulting from this calculation are correct only for sources with small extent and a brightness profile peaking in the center. For extended sources like SNRs with ringlike structure the net count rates were determined interactively by integrating the counts within a circle around the source. For the background the counts were averaged in a ring around the source distant enough not to be influenced by the source emission.

In order to increase the sensitivity HRI observations with pointing directions within a radius of 1$^\prime$ were merged after adjusting their position. This was possible for 56 different regions in the LMC. Source detection was also performed on these data and additional faint sources were found which were not detectable in single pointings.

The final source lists obtained for each pointing and co-added observations were merged to one list and multiple detections of a source were reduced to one detection for each source. For this purpose the detection with the smallest positional error was chosen. After screening manually in order to delete spurious detections like knots in extended emission, the catalogue finally contains 397 distinct sources.

  \begin{figure}{\psfig{,angle=270,width=9cm} }
\end{figure} Figure 1: Histogram of HRI pointing exposure times

2.2 Positional corrections and error

ROSAT observations suffer from a systematic positional uncertainty of about 7 $^{\prime\prime}$ (Kürster 1993). For minimizing this systematic error the coordinates of identified objects were compared to high accuracy positions available in the TYCHO catalogue obtained from the ESA Hipparcos space astrometry satellite (Hoeg et al. 1997) or in the literature. First the X-ray position was corrected to TYCHO coordinates. For sources without any TYCHO counterpart, but identified on the ESO Digitized Sky Survey (DSS) frame with other stars on this frame which were listed in the TYCHO catalogue, more accurate coordinates were calculated for HRI sources by determining the offset between the TYCHO and DSS positions and between the HRI and DSS position. Other sources could be identified with objects in the SIMBAD data base operated at the Centre de Données astronomiques de Strasbourg or in the literature and their positions were corrected after checking their positions on DSS frames. Correction of coordinates for one source implied improved coordinates for all detections of this source in different pointings and for other sources in same pointings. Those secondary corrections again allowed correction of further pointings if the sources were detected several times. Finally for 254 out of 397 sources improved coordinates were determined.

In cases where positional correction was possible the remaining systematic error consists of the error in former optical measurements and the statistical error of the identified source. For not corrected sources the systematic error was set to 7 $^{\prime\prime}$. The positional error was finally computed as a composite of the statistical uncertainty with 90% confidence and the systematic error. It is used throughout the paper for the error circle. After the source detection procedure the mean positional error was 8 $.\!\!^{\prime\prime}$3. The coordinate correction reduced the mean positional error of all sources to 6 $.\!\!^{\prime\prime}$4. For position corrected sources the mean positional error is 5 $.\!\!^{\prime\prime}$1.


Table 1: HRI sources with significant flux variability

2 3 4 5 6 7   8

Rate Rate $\frac{ F_{\rm max}}{F_{\rm min}}$ Red. $\chi^2$ DOF No Remarks
  HRI PSPC       PSPC  
  [cts s-1] [cts s-1]          

*4.6e-1 *2.3e-3 598.1 376.1 1 331 HMXB RX J0502.9-6626 (CAL E)
20 5.3e-2 7.7e-2 4.5 163.5 2 380 AGN RX J0503.1-6634, z=0.064 [SCF94]
23 2.4e-2 2.1e-2 3.4 61.9 1 715  
49 *4.9e-3 *7.5e-3 4.3 9.4 4 559 <XB> or <AGN>
65 2.2e-3 1.9 25611.5 663.9 45 1030 SSS RX J0513.9-6951
103 4.1e-3 2.8e-3 10.5 10.2 6   foreground star HD 35862
124 1.8e-2 1.0e-1 4.9 19.6 20 1094 AGN RX J0524.0-7011, z=0.151 [SCF94]
155 *8.3e-3 *4.2e-3 132.1 18.7 30   Nova LMC 1995 [OG99]
167 1.4e-3 1.2e-1 256.1 51.6 26 1039 SSS RX J0527.8-6954
180 *1.9 *7.5 11.2 691.6 74 122 foreground star K1III& HD 36705 (AB Dor)
193 3.4e-2 1.7e-1 2.6 7.0 8 749 foreground star G5 HD 269620 [CSM97]
202 2.0e-3 8.5e-2 1022.5 11.9 20 204 HMXB Be/X RXJ0529.8-6556 [HDP97]
218 8.4e-3 3.7e-1 344.9 99.9 10 252 HMXB Be/X EXO053109-6609 [HDP95a], [DHP96]
233 6.0e-3 2.2e-2 66.1 7.9 13 184 HMXB RX J0532.5-6551 (Sk-65 66) [HPD95b]
239 *8.9e-2 *4.9 367.1 10247.6 25 316 HMXB LMC X-4, HD 269743 O8III
293 4.4e-2 1.6e-1 2.1 9.4 18 902 foreground star dMe CAL 69 [CSM97]
300 5.4e-3   18.9 414.6 2   <stellar>, source not resolved by the PSPC
306 *6.0 *23.4 2.5 1838.8 22 41 HMXB LMC X-3
311 3.5 13.5 1.6 576.1 29 1001 HMXB LMC X-1, O8III
313 *6.3e-3 *8.3e-3 3.7 9.3 3 668 <stellar>
348 4.3e-2   284.0 775.9 6 654 SSS CAL 83 [SCF94], one PSPC point., source near rim
349 3.0e-2 1.0e-1 19.5 7.2 5 61 foreground star? [HP99a]
352 1.3e-3 1.2e-2 3.1 10.0 2 1225 HMXB RX J0544.1-7100 [HP99b]
363 6.1e-2 1.3e-1 1.5 36.8 3 1240 SSS CAL 87
364 1.2e-2   30.5 158.4 1 747 <XB> or <AGN>, one PSPC pointing, source near rib
375 3.3e-3 3.3e-2 4.5 6.1 4 1127 foreground star F3/F5IV/V HD 39756


Notes to columns 2 and 3: For point and point like sources count rates are the mean of output values from maximum likelihood algorithm for single pointings. For extended sources and bright sources with apparent extent (see text) the average of integrated count rates in single pointings was taken (* in front of the number).
Notes to column 6: Degrees of freedom.
Notes to column 7: Source number from HP99b.
Notes to column 8: Sources classified in this work are put in < >. Abbreviations for references in square brackets are given in literature list.

...hspace{1mm} T = 10.0 -- 200.0 eV \\
\end{figure} Figure 2: PSPC/HRI conversion factor as function of $N_{\rm H}$ for power law, Raymond & Smith, and black body spectra

2.3 Correlation with existing catalogues

The catalogue was cross-correlated with the SIMBAD data base and the TYCHO catalogue in order to identify HRI sources. The HRI catalogue contains samples of known SSSs, X-ray binaries, SNRs, Galactic foreground stars, and background AGN. The catalogue was also cross-correlated with the source list from the pointed PSPC observations (HP99b). 138 HRI sources are identical with sources which were detected in PSPC data and thus for most of them the hardness ratios (HR1, HR2) are known. Since the HRI had no spectral resolution no information on the X-ray spectrum could be obtained for HRI sources which are completely new detections. A total of 94 HRI sources were identified with known objects like SSSs, X-ray binaries, SNRs, stars, and background AGN.

With the help of their X-ray properties like extent, extent likelihood, PSPC hardness ratios, X-ray to optical flux ratio (see Sect.3.2), and X-ray variability 14 previously unknown HRI sources and 7 sources also listed in the PSPC catalogue were newly classified.

The whole source catalogue from HRI observations with the corrected coordinates, final positional error, existence likelihood, HRI count rate, extent, extent likelihood, PSPC count rate and the corresponding PSPC source number with hardness ratios (HP99b) is given in Table 4. For each HRI and PSPC count rate the results for the pointing with the smallest positional error, determined by the maximum likelihood algorithm, were selected. Therefore HRI count rates in the table are representative for one single observation for each source. For extended SNRs the given count rate may correspond to a knot within the source. PSPC count rates are taken from the PSPC catalogue (HP99b) if available. For HRI sources without PSPC detection we derived 2$\sigma$ upper limit from the pointing with the highest exposure time. If the source was too close to the rim or the window support structure of the PSPC detector, no count rate is given in Table 4. Neither was it possible to determine PSPC count rates or upper limits for sources located in regions with diffuse emission.

2.4 Flux variability

About 80% of HRI sources were observed more than once and allow time variability studies. For point and point like sources longterm lightcurves were produced with observation-average count rates or upper limits determined by the maximum likelihood algorithm, whereas for extended sources integrated count rates within a circle were used (see Sect.2.1). For some very bright sources the count rates were integrated in the same way, because an apparent extent resulted from the maximum likelihood algorithm. An apparent extent is computed if the high photon statistics of the bright sources cause a significant deviation from the assumed model for the point spread function.

A $\chi^2$-test for a constant count rate was performed and the factor between the maximum and minimum flux was computed for each lightcurve. Together with the reduced $\chi^2$ this flux factor was used to characterize variability on long time scales of days to years (see also HP99a). For SNRs we expect constant integrated flux, however the flux factor was in the range of 1.0 to 1.8. This may be caused by different off-axis angles and/or different extraction of the extended source. Therefore variations below a factor of 2.0 should be handled with care as they might indicate no real variability but false integration of the source flux because of the extent or existence of a nearby bright source.

In order to obtain a complete lightcurve of the ROSAT observations, also PSPC count rates and upper limits were calculated for the HRI sources. In Figs. 2a-c the PSPC to HRI count rate conversion factor is plotted over $N_{\rm H} =
10^{20} - 10^{23}$ cm-2 for three different spectral models. SSSs with a soft black body spectrum can be modeled with T = 10.0 - 50.0 eV and galactic $N_{\rm H} = 10^{20} - 10^{21}$ cm-2 in the direction of the LMC. XBs in general show a power law spectrum with $N_{\rm H}$ up to 1022 cm-2 because of intrinsic absorption $N_{\rm H}$. So for most of the point and point like X-ray sources PSPC count rates can be converted into HRI count rates by dividing by a typical value of 3, though for very soft sources this scale factor can be larger. Sources in regions with extended emission (e.g. 30 Dor or N44) or close to another source can not always be resolved in PSPC data and may result in false large converting factor.

$\chi^2$ and the flux factor were again calculated for all lightcurves including PSPC count rates (divided by 3.0) and upper limits. Finally 26 sources show significant variability with reduced $\chi^{2} >$ 5 corresponding to a probability > 0.9999 (see Table 1). Four of them are new classified HRI sources (for sources Nos. 49 and 364 see Sect.3.2.4, for Nos. 300 and 313 see Sect.3.2.2). As example the lightcurve of source No. 49, a new HRI candidate for a variable X-ray binary or AGN is shown in Fig.3.

PSPC count rates were determined in as many pointings as possible. The mean value was calculated from these count rates and compared to the HRI mean count rates (see Fig.4). The resulting conversion factor is close to 3.0, only variable sources marked with dots show bigger deviation.

  \begin{figure}{\psfig{,angle=270,width=9cm} }
\end{figure} Figure 3: Lightcurve of source No. 49. Crosses for converted PSPC count rates, dots for HRI count rates. Zero point of the space craft clock is 1990, June 21 21:06:50 UT

  \begin{figure}{\psfig{,angle=270,width=9cm} }
\end{figure} Figure 4: Mean of observation-averaged count rates from PSPC pointings over mean of observation-averaged count rates from HRI pointings for ROSAT sources. Squares indicate SNRs, lozenges stars, hexagons XBs, triangles AGN, and asterisks SSSs. Crossed symbols are already known candidates. Variable sources are additionally marked with filled dots. The line indicates a PSPC/HRI conversion factor of 3

next previous
Up: ROSAT HRI catalogue of region

Copyright The European Southern Observatory (ESO)