Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 354, 847-852 (2000)

Previous Section Next Section Title Page Table of Contents

2. Data selection

In the present study photon-event files from the Position Sensitive Proportional Counter (PSPC) instrument onboard the ROSAT satellite were used. Data corrected for instrumental effects were obtained from HEASARC archives.

Since the spacial resolution of the instrument decreases towards the lowest energies, only photons with energies above 0.5 keV were used. Furthermore, there is always a problem of interference from nearby sources with completely different statistical properties. One example is 3C 345 which is surrounded by a number of close quasars (Arp 1997). Other example is NGC 6251 where a X-ray jet and halo have been identified close to the central source (Mack et al. 1997). For that reason, for each source, a centrum of gravity of the image was determined and then a circular section of the image, limited by 1/e of the centrum intensity was selected.

In the present paper we are studying 19 AGN of Seyfert 1 (S1) type and 9 quasars (QSOs). Table 1 lists those sources together with the ROSAT files we analyze.


Table 1. Objects analyzed and their ROSAT files

In order to carry on the present study, it was necessary to find a proper method of extracting the information from the photon series. A combined approach using both wavelet spectra and probability density distributions of time intervals between photons was chosen.

We derive wavelet spectra of temporal variations of X-ray counts from the photon event files using the Morlet wavelet transform (see Paper I). As the wavelet spectrum does not change its character when the length of sampling bins is changed, photon counts were sampled in 1-second bins, even for weak sources. An example of wavelet spectrum for NGC5548, ROSAT observation request number (ROR) 701242, bin lengths of 1, 2, 4 and 8 seconds, is shown in Fig. 1.

[FIGURE] Fig. 1. An example of wavelet spectrum for NGC5548, ROR 701242, bin lengths of 1, 2, 4 and 8 seconds. The frequency scale in Hz.

With long sampling bins it is obvious that high frequency information is lost. For that reason 1 second bins were used for all data in the present study. For weak sources it results, of course, in very low counting rates. In order to perform the non-linear filtering of the data, as many as 128 dilations (equivalent to frequency steps) in the Morlet wavelet transform were needed. Only observation periods covering at least 1024 seconds, without interruption, were used. Observation periods covering a multiple of 1024 seconds were divided into several samples. A uniform sample length for all analyzed data was used so the same dilation (frequency) scale could be used for all sources and all observations.

A survey of the data has been performed using low-resolution wavelet spectra, with only 15 dilations. The principal component analysis of the data shows that the dominant component in the wavelet spectra is related to the apparent brightness of the source described here by the average counting rate during all analyzed observation periods (cf. Fig. 2).

[FIGURE] Fig. 2. The PC1 of the survey spectra as a function of the average counting rate.

Thus, the apparent brightness of the sources is a factor dominating the variability properties of the analyzed data. It appears to be responsible for about 25% of the total variance. The influence of apparent brightness therefore obscures the high order, source specific effects. As the first step in the analysis, that influence may be removed from the data matrix. It may be done using a Principal Component Analysis (PCA)-based decomposition technique described in Paper I. After the removal of the most significant principal component (PC1) a new data matrix, hopefully corrected for effects of the apparent brightness, is obtained. However, it has been found that the principal component analysis alone is not able to eliminate all the effects due to the apparent brightness.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 2000

Online publication: February 25, 2000