skip to content
ROSAT Guest Observer Facility


K. M. Leighly & H. Marshall
Eureka Scientific,


In 1995, the ROSAT HRI observed the broad-line radio galaxy 3C 390.3 once every 3 days for nine months. The resulting lightcurve, showing substantial structure including flares and quiescent periods which may be indicative of nonlinear variability, raises several questions. Is this variability structure observed in radio-quiet active galaxies (AGN)? Can narrow and broad-line AGN be distinguished by their variability properties? Do the variability properties of 3C 390.3 change after an increase in flux? We propose to address these questions by characterizing the intermediate time scale variability of a small sample of AGN with comparable luminosity as 3C 390.3, but differing by being radio quiet or having narrow emission lines. We also propose follow-up monitoring of 3C 390.3 in order to verify its nonlinear behavior, which places severe constraints on X-ray variability models.


X-ray variability has been observed in active galaxies (AGN) for two decades, but the origin of the variability is comparatively poorly understood. So far, observations of rapid variability have constrained the size of the emission region and efficiency of the accretion process, and it is known that between and Hz the power spectrum is steep and featureless. But nearly nothing is known about the variability on intermediate time scales of weeks to months. These timescales are important since the frequency of the expected turnover of the power spectrum tells the length of the ``memory'' of the system giving an upper limit on the emission region size and constraints on emission models.

In 1995, we observed the broad-line radio galaxy 3C 390.3 using the ROSAT HRI every three days for about nine months, resulting in the first regularly sampled light curve on time scales of days to months (Figure 1). The results are quite startling and contrast markedly with our expectations. We find a large amount of power is apparent on these time scales, and the structure characterized as flares interspersed among quiescent periods suggests that a nonlinear process is the origin. This may provide fundamental new variability property which can cleanly distinguish between models, and indicates that the study of intermediate time scale variability provides a potentially powerful new tool for the study of AGNs. These results raise several questions. Primarily, are the variability properties of objects with similar luminosity but different radio properties or optical line widths similar or different? Second, does the scaling of variability properties with luminosity observed at short time scales carry through to intermediate time scales and for objects with different optical line properties?

We have therefore proposed monitoring observations of a small sample of AGN chosen to complement the observations of 3C 390.3 and approved AO-6 monitoring observations of the narrow-line Seyfert 1 galaxy 1H 0419-557. We chose two objects which have roughly the same X-ray luminosity as 3C 390.3, but differ in that they are radio quiet (Mrk 876) or have narrow optical emission lines (H0707-495). As an orthogonal line of investigation, we propose monitoring observations of two other narrow-line Seyfert 1 galaxies which have lower (H0707-495) and higher (PKS 0558-504) luminosities. It is necessary to use ROSAT to observe faint objects such as Mrk 876 and the steep spectrum NLS1s, and wee will also propose complementary observations using RXTE, which will yield more easily interpreted observations of lower luminosity broad-line Seyfert 1 galaxies.

Intermediate Time-Scale Variability

Linear Properties:

Thus far, the most constraining results from AGN variability studies have come from EXOSAT data, because even sampling was possible on time scales of <3 days. The variability power spectral density (PDS) on these time scales can be described by a power law with slope s < -1 (; e.g., McHardy 1989). Various models have been proposed to explain this steep power spectrum, including shot noise models of various shapes which may describe overlapping flares (e.g., Lehto 1989) or hot spots on a rotating disk with modulation by beaming and occultation (Abramowicz et al. 1991). Recent analysis finds that the index clusters around a value of 1.55 over a large range of luminosity, tending to rule out shot noise models (Lawrence & Papadakis 1993, hereafter LP93). An inverse correlation between power spectral amplitude and X-ray luminosity was also found (LP93), which generally supports the idea that less luminous objects have a lower black hole mass.

Figure 1: 3C 390.3 light curve from the 1995 ROSAT monitoring. Approximately 9 months are spanned by 92 observations. Evidence for nonlinearity is seen in the flaring and quiescent periods. Bottom:) Variance using a sliding box (N=6), highlighting quiescent periods.

On longer time scales, PDS must flatten to low frequencies, or the total power would diverge. At long time scales of years the flux of individual AGN vary (e.g., Abraham & McHardy 1989) but less than implied by extrapolation of the high frequency PDS to low frequencies. The turnover, expected to occur on time scales of months, gives information about the physical process driving the variability; for example, if it is connected to dynamical instabilities in the accretion disk, the viscous time scale would be relevant, days (: black hole mass, : Schwarzschild radius, T: temperature, : disk viscosity, : Eddington luminosity).

Contrary to expectation, no evidence of the break has been found in the few well sampled time series over periods of weeks to months. The power spectrum from the monitoring observations of 3C 390.3 shows no evidence for a break (Figure 2). A remarkable result is that the index is , the same as found by LP93 at higher frequencies, and the amplitude of variability is consistent with the LP93 correlation, even though the range of frequencies sampled are completely disjoint.

Figure 2: The power spectrum from the 1995 ROSAT monitoring campaign. Both the raw and binned power spectrum (Papadakis & Lawrence 1993) are shown. The spectral index is steep, (Leighly et al. in prep.)

Nonlinear Properties:

In general, light curves produced from linear and nonlinear processes are quite distinct (e.g., Figure 2 of Vio et al. 1992) because of the ability of nonlinear processes to produce flares and quiescent periods. Evidence for nonlinearity in light curves provides a strong model constraint. For example, shot noise is a linear process and can be ruled out if nonlinearity is detected (e.g., Vio et al. 1991). Physical models may require modification. For example, Vio et al. (1992) show that the simplest spots-on-a-disk model produces a rather uniformly varying light curve. However, if there is interaction between the spots so that n overlapping spots increase emission to , flares can be produced.

The flares and quiescent periods in the 3C 390.3 light curve are reminiscent of the optical light curve from the OVV blazar 3C 345 (Vio et al. 1991) and suggest the presence of nonlinearity. Analysis of nonlinearity requires special methods, as the power spectrum and related autocorrelation measure only linear properties (e.g., Vio et al. 1992). Using ``surrogate'' data method (Theiler et al. 1992) in which simulated light curves with well defined properties are compared with the real data, we detect nonlinearity with 5 sigma confidence. Quiescent periods in the light curve might be evidence for intermittency, a characteristic of nonlinearity. That is demonstrated by computing variance over a sliding box, shown in Figure 1 (Isliker & Benz 1994). These results are especially exciting because this is the first time that nonlinearity has been rigorously established in X-ray light curves from AGN (Leighly et al. in prep.).

Questions and Answers

An X-ray Jet in 3C 390.3?:

Since 3C 390.3 has a compact superluminal jet (Alef et al. 1996), it is possible that the X-ray emission comes from the jet, and indeed the similarity with the optical light curve from the OVV blazar 3C 345 (Vio et al. 1991) perhaps suggests this. Other evidence, including the lack of radio variability and the evidence for reprocessing including a broad iron line (Eracleous et al. 1996), reflection component (Nandra & Pounds 1994) and double peaked, broad optical lines (Perez et al. 1988), would seem to favor isotropic emission. Therefore, the large scale structure and nonlinearity observed in the 3C 390.3 light curve may not be applicable to AGN in general. To test this, we need to perform similar quality intermediate time scale monitoring observations of a similar AGN that is instead radio quiet.

To address this question, we have proposed observations of the Seyfert 1 galaxy Mrk 876 every 3.65 days for a year. Located near the NEP, it is always visible to ROSAT. The X-ray luminosity of this object is comparable to that of 3C 390.3 so a direct comparison of properties can be made. Coordinated V and R band photometric observations provided by M. Dietrich will be used to test reprocessing models for continuum emission (e.g., Clavel et al. 1992).

Narrow-line vs. Broad-line Seyfert 1 Galaxies

Recent ROSAT observations of an optically defined class of objects, the Narrow-line Seyfert 1s (NLS1s; Goodrich 1989), discovered properties distinct from broad-line Seyfert 1s, including an extremely steep soft X-ray spectrum (). The most interesting and significant result is that the soft X-ray photon index is inversely correlated with the optical line width. Some NLS1s show exceptionally rapid or large amplitude variability. It seems reasonable that the extreme properties of NLS1s result from an extreme value of a physical parameter, possibly the black hole mass, the accretion rate or geometry (e.g., Boller, Brandt & Fink 1996).

While it has been suggested that rapid, large amplitude variability should also be considered a characteristic property of NLS1s (Forster & Halpern 1996), it has never been systematically tested. In fact, there are several examples of broad-line Seyfert 1s, including NGC 7314 (Yaqoob et al. in prep.) and MCG-6-30-15 (Reynolds et al. 1995), which have comparable doubling time scales and luminosities. Doubling time scales can be misleading, however, and the fact that two of the three NLS1s in LP93 sample lie substantially above the correlation is suggestive. A larger comparative amplitude of variability might support a relatively smaller mass black hole and larger accretion rate (e.g., Boller, Brandt, & Fink 1996). A lower frequency turnover of the PDS would also support this. Finally, we note that the light curves of some NLS1s show flare activity, suggesting nonlinearity (e.g., Mrk 478, Marshall et al. 1996; IRAS 13324-3804, Otani et al. 1996).

During 1996, we are already scheduled to monitor the high luminosity NLS1 1H0419-577 daily for two months. To complement this campaign, we have proposed daily monitoring observations of the NLS1 H0707-495 which is a factor of 5 lower in luminosity, and is thus close to the luminosity of the broad-line radio galaxy 3C 390.3 and radio quiet object Mkn 876. Further, we have proposed monitoring observations of the peculiar radio-loud NLS1 PKS 0558-504, a factor of 2.5 higher in luminosity than 1H 0419-577. These observations will allow us to compare variability properties of two samples of objects. The first has similar X-ray luminosities, but differing radio and optical line properties, and the second has similar optical line properties (all NLS1s) but differing luminosities. With proposed RXTE observations, we will also form a sample of broad-line objects with differing luminosities. Since variability properties likely reflect the primary properties of the system including the X-ray emission mechanism, accretion rate or black hole mass, differentiation on the basis of these properties will allow a major step forward in our understanding of these objects.


Alef, W., et al. 1996, A&A, 308, 376

Abramowicz, M. A., et al. 1991, A&A, 245, 454

Abraham, R. G., & McHardy, I. M. 1989. Proc. 23rd ESLAB Symp., eds. J. Hunt and B. Battrick (Paris: ESA) p. 865

Boller, Th., Brandt, W. N. & Fink, H., 1996, A&A, 305, 53

Clavel, J., et al. 1992, ApJ, 393, 113

Eracleous, M., et al. 1996, ApJ, 459, 89

Forster, K., & Halpern, J. P., 1996, ApJ, in press

Goodrich, R. W., 1989, ApJ, 342, 224

Isliker, H., & Benz, A. O., 1994, A & A, 285, 633

Lawrence, A., & Papadakis, I., 1993, ApJ, 414, L85 (LP93)

Lehto, H. 1989. Proc. 23rd ESLAB Symp., eds. J. Hunt and B. Battrick (Paris: ESA), 499

Marshall, H., et al. 1996, ApJ, 457, 169

McHardy, I. M. 1989. Proc. 23rd ESLAB Symp., eds. J. Hunt and B. Battrick (Paris: ESA) p. 1111

Nandra, K. & Pounds, K. A., 1994, MNRAS, 268, 405

Otani, C. et al. 1996, in Proc. of ``Röntgen-strahlung from the Universe'', Würzburg, Germany, Sept. 25-29, 1995

Papadakis, I. E., & Lawrence, A. 1993, MNRAS, 261, 612

Perez, E., et al. 1988, MNRAS, 230, 356

Reynolds, C., et al. 1995, MNRAS, 277, 901

Theiler, J., et al. 1992, Physica D, 58, 77

Vio, R., et al. 1991, ApJ, 380, 351

Vio, R., et al. 1992, ApJ, 391, 518

Return to the Title Page
About this document ...

This document was generated using the LaTeX2HTML translator Version 96.1 (Feb 5, 1996) Copyright © 1993, 1994, 1995, 1996, Nikos Drakos, Computer Based Learning Unit, University of Leeds.

The command line arguments were:
latex2html -split 0 marshall.tex.

The translation was initiated by Michael Arida on Mon Jun 3 15:35:46 EDT 1996

Michael Arida
Mon Jun 3 15:35:46 EDT 1996