During the nominal operation phase of the PSPC between 1990 and 1994 this imaging X-ray detector on board ROSAT was used for more than 200 pointed observations in an area of centered on the position 43 20. The often overlapping images cover in total 58.6 square degrees of the LMC region. A merged image in the energy band 0.5-0.9 keV with a resolution of 15 is presented in Haberl & Pietsch (1998). Statistics about the location and exposures of the PSPC observations can be found in Haberl & Pietsch (1999), hereafter HP99. For our analysis we used the 212 observations with exposure of more than 100 s.
Using merged data for source detection is only useful in areas where data from the central part of the detector is not superposed by off-axis data with degraded spatial resolution. To derive a catalogue of point and point-like X-ray sources in the LMC we therefore started with the analysis of the images from the individual observations. Each of the 212 observations has been analyzed in five energy bands (soft: 0.1-0.4 keV, hard: 0.5-2.0 keV, hard1: 0.5-0.9 keV, hard2: 0.9-2.0 keV and broad 0.1-2.4 keV) with the three detection methods of EXSAS (Zimmermann et al. 1994). For each band the merged detection lists of two sliding box methods (using local background and a spline fit background map) were used as input for the maximum likelihood (ML) algorithm. Sources were only accepted when the likelihood of existence (ML) was larger than 10.0 (corresponding to a probability ML) greater than ), the total number of counts more than 12 and the angle to the telescope axis within 52.The five ML detection lists from the different bands were merged to a single list for each observation. The ML results were used for the final source parameters like likelihood of existence, celestial coordinates, detected counts, source extent and extent likelihood. For all parameters uncertainties are given.
The source extent was determined using a Gaussian approximation for the intensity profile. This is justified as long as the extent is small and the surface brightness distribution is well peaked in the center. However there are well known SNRs in the LMC which do not obey these criteria (in particular ring shaped remnants like N132D). The maximum likelihood algorithm gives too low count rates in these cases. This effect is strongest in the center of the field of view (FOV) of the instrument and decreases with off-axis angle due to the degrading point spread function (PSF) which smears out any structures. As a consequence we re-determined the number of detected source counts by integrating them in a box around the source and a nearby box for background for the following SNRs: N132D, N63A, N49, N103B and N49B.
To determine count rates, exposure maps were created taking instrument maps (vignetting) and the satellite attitude into account. From the count rates in the soft (S), hard (H), hard1 (H1) and hard2 (H2) bands hardness ratios and were derived as indicators for the shape of the X-ray spectrum. A mask of the support structure of the detector entrance window and the detector rim (WSS) was summed up taking the changing attitude (mainly caused by the satellite wobble) into account. This "wobbled mask" was used to calculate the distance of each source to the shadowed structures where the exposure was reduced by more than 10%. The distance in units of the FWHM of the point spread function directly gives an indication for how much source parameters are influenced by the WSS.
A large fraction of the LMC region was observed more than once yielding multiple detections of many sources. To study long-term time variability HP99 cross-correlated the detection lists of all individual observations. They presented 27 X-ray sources which showed flux variations as observed by the PSPC by factors between 3 and more than 1000.
For this work the detection lists were merged to produce a source catalogue of ROSAT
PSPC sources. For sources with multiple detections the one with the smallest position
error (usually the detection nearest to the telescope axis and/or best counting
statistics) entered the catalogue. Two detections were regarded as caused by the same
source when their positions are within the distance d = d90 + , where d90 denotes the summed 90% errors of the two sources and the systematic error of
7 and the average source extent of the two detections.
The source extent was only used when the likelihood for the extent was larger than 20.
The catalogue was finally visually screened to remove spurious detections like
multiple detections in diffuse emission regions and obvious false detections near
the WSS. In particular in the 30 Doradus region the intense diffuse X-ray
emission led to many spurious detections which were screened out. Only
known SNRs were accepted and for a more complete catalogue in that area
HRI data with better angular resolution needs to be analyzed.
The final PSPC catalogue of point-like and weakly extended sources comprises
758 entries. An excerpt with five consecutive entries is shown in
Table 1. The complete table is available electronically.
The table columns give (1) source number, (2) likelihood of existence
(maximum value from the five detection energy bands),
(3) net exposure, vignetting corrected, (4) and (5) celestial
coordinates of X-ray source, derived from energy band with lowest
position error, (6) statistical 90% confidence error on the X-ray
position, the systematic uncertainty is about 7, (7) PSPC
0.1-2.4 keV count rate and error, (8) and (9) hardness ratios, not
calculated in cases where not all required count rates are available, (10) source
extent, from same detection as source position,
(11) likelihood for the extent, (12) ROSAT observation identifier,
(13) off-axis angle, (14) distance to WSS and (15) remark.
Copyright The European Southern Observatory (ESO)