Astron. Astrophys. 340, 351-370 (1998)

2. Observations and data analysis

NGC 3079 was observed from October 20 to 25, 1992 with the ROSAT HRI and from November 13 to 15, 1991 with the ROSAT PSPC, for a total observing time of 20.7 ks (HRI) and 19.0 ks (PSPC). The observations were split into 11 and 9 observation intervals (OBIs) for HRI and PSPC, respectively. To analyse the data for times of high detector background, we investigated rates of the master veto rate counter for the PSPC and the invalid counts for the HRI integrated over 60 s. The rates reach maxima of 280 cts s^-1 (82 cts s^-1), but stay below 200 cts s^-1 (60 cts s^-1) for 98% (99%) of the time for the PSPC (HRI). We therefore decided not to reject events due to times of high background.

2.1. Attitude corrections

To improve on the attitude solution we have adopted two subsequent techniques. For the HRI data we analyzed individual OBIs and aligned the data with the help of 7 bright point sources (namely H1, H2, H8, H9, H15, H20, H23, cp. Sect. 3.1) to the average position. While the offset in the position of these sources was for most OBIs, for OBI 5 the offset was . The photons were corrected for the offsets.

We also looked for possible optical counterparts (see Appendix A.1 and A.2) with the aid of APM finding charts (Irwin et al. 1994), and compared the optical and X-ray positions, to determine a possible systematic error on the absolute attitude solution. We found optical candidates for 9 sources in the HRI and 7 in the PSPC. We determined a systematic shift of 3:004 to the east and 3:009 to the north and an additional counterclockwise rotation of 0:O39 for the HRI observation. The corrected center of the HRI pointing direction is = ,  (J2000.0). For the PSPC, we determined a shift of 8:008 to the east and 6:005 to the south and an additional counterclockwise rotation of 0:O2. The corrected center of the PSPC pointing is = ,  (J2000.0). Source lists and images have been corrected for these systematic effects. The remaining position uncertainty is less than for both instruments.

The attitude corrections determined above for HRI and PSPC are in good agreement with boresight parameters determined from a larger sample of observations that now are used for the SASS re-processing (M. Kürster, private communication).

2.2. Iso-intensity contour maps

For the PSPC data, contour plots have been obtained from images that were the result of the superposition of sub-images with bin size in the 8 standard bands (R1 to R8, cf. Snowden et al. 1994), corrected for exposure, vignetting, and dead time and smoothed with a Gaussian filter having a FWHM corresponding to the on-axis point spread function (PSF) of that particular energy band. The FWHM values used range from to . The average background was calculated from a source free region to the north of NGC 3079.

For the HRI, contour plots have been obtained from images with 2:005 bin size which were corrected for dead time and smoothed with a FWHM Gaussian filter. To reduce the background due to UV emission or cosmic rays we used only those events detected in the HRI raw Pulse Height Amplitude channels 2-8.

The resulting images are discussed in Sect. 3.1.

2.3. Source detection

We performed source detection and position determination with the EXSAS local detect, map detect, and maximum likelihood algorithms (Zimmermann et al. 1992). Maximum likelihood values () can be converted into probabilities () through , so that corresponds to a Gaussian significance of about 3.6 and corresponds to a Gaussian significance of about 3.9 (Cruddace et al. 1988; Zimmermann et al. 1994).

Once the count rates of sources are obtained (see below for the details on how they have been obtained in the two instruments), they are converted into fluxes in the ROSAT band (0.1-2.4 keV) assuming a 5 keV thermal bremsstrahlung (cf. Table 2). To estimate possible errors in the X-ray luminosities due to the selection of a wrong model or temperature, conversion factors for a 0.5 keV thermal bremsstrahlung and for a 0.3 keV and a 3 keV thin thermal plasma spectrum have also been calculated. For this range of temperature and models, the resulting values change by 15%.

Table 2. Conversion factors from count rates to fluxes (0.1-2.4 keV) for the ROSAT HRI and PSPC detector (broad band) in units of  erg cm^-2 cts^-1, corrected for Galactic absorption
Notes:
thermal bremsstrahlung spectrum (we assumed  keV for conversion of point-like sources)
thin thermal plasma (we assumed  keV for diffuse X-ray emission components)

Different considerations were applied to the HRI and the PSPC data to better suit the properties of these instruments.

HRI : Sources were searched for in the inner diameter circle about the field's center. Again, we used only those events detected in the HRI raw Pulse Height Amplitude channels 2-8 (see Sect. 2.2). Sources with a were accepted.

PSPC : Sources were searched for in the inner field centered on NGC 3079 in the five standard ROSAT energy bands: "broad" (0.1-2.4 keV), "soft" (0.1-0.4 keV), "hard" (0.5-2.0 keV), "hard1" (0.5-0.9 keV), and "hard2" (0.9-2.0 keV). Sources for which we obtained in at least one of the bands were considered. We have used a slightly higher L value than in the HRI analysis since for the PSPC we are already getting background limited.Once the existence of the source is established, its position is determined from the band with the highest value. This is then used to derive the net counts in the five standard bands defined above in an aperture corresponding to FWHM of the energy band considered. The background is always taken from the same area in the corresponding background map (see EXSAS manual, Zimmermann et al. 1994). These values are used to calculate hardness ratios and their corresponding errors: HR1 = (hard-soft)/(hard+soft) and HR2 = (hard2-hard1)/(hard2+hard1), where all of the quantities are corrected for the appropriate vignetting correction (see EXSAS manual). If the source is not detected in one of the four bands, however, the corresponding hardness ratio cannot be simply calculated. We have therefore modified the algorithm, and determined to compute the hardness ratio using the 2 upper limit instead of the net counts, when the signal-to-noise ratio in a particular band is less than 2. This allows us to estimate a maximum or minimum value that the hardness ratio can have. When both quantities are upper limits no hardness ratio is calculated.Hardness ratios can be used to have a crude estimate of the spectral parameters that best apply to the energy distribution of the source photons, when the statistics do not allow a more detailed analysis. To show this we have also calculated "theoretical" hardness ratios from the spectral distribution of some standard spectral models (i.e. Raymond & Smith, power law and thermal bremsstrahlung) with a low energy cut-off. These are shown in Fig. 1, and can be used as a comparison to the observed values of HR1 and HR2 (see Sects. 3.3.2 and Appendix A.1).

Fig. 1. Expected distribution of HR1 and HR2 for different spectral models and spectral parameters and observed values. Curves are drawn for different assumed equivalent absorbing column, as given in the figure (in logarithmic value) and for different temperatures/indexes. Left panels: Curves for Raymond & Smith (top) and thermal Bremsstrahlung spectra. Dots along the curves indicate different temperatures, in steps of 0.1 in log(T7), where T7 is T/ K. Some of the points are labeled for easy reference. Upper right pannel: Curves for power law spectra. Dots along the curves indicate different photon index, as indicated. Lower right pannel: Observed hardness ratios and limits

© European Southern Observatory (ESO) 1998

Online publication: November 9, 1998
helpdesk.link@springer.de