2. Data analysis
The published infrared observations of GRS 1915+105 for which there are simultaneous or quasi-simultaneous (ie. within 2 days) RXTE/PCA data are those from Mirabel et al. (1998), Eikenberry et al. (1998), Fender et al. (1998), Eikenberry et al. (2000), Fender & Pooley, (2000). All observations reveal very variable X-ray light curves (see Table 1), corresponding to classes , and in the classification by Belloni et al. (2000).
Table 1. Log of PCA observations and summary of spectral parameters (see text). Classes in column 3 correspond to the classification from Belloni et al. (2000).
For each observation, we produce light curves at 1s time resolution (from Standard1 data) and isolated the long hard low-flux intervals corresponding to state C (unobservable inner disk) of Belloni et al. (2000). For each interval, we measured its length from the light curve (see Table 1). Then we accumulated spectra on a time scale of 16 seconds from Standard2 data, thus retaining the full energy resolution and coverage of the PCA. From each spectrum, we subtracted the background estimated with pcabackest vers. 2.1b. We did not correct for deadtime effects, but we do not expect this effect to be too important. For each observation in PCA epoch 3 we produced a detector response matrix using pcarsp , while for epoch 4 we used the response provided on line by K. Yahoda 1. We fitted each spectrum with the "standard" model used for black-hole candidates, consisting of the superposition of a multicolor disk-blackbody and a power law. By assuming a distance of 12.5 kpc and a disk inclination of 70o (Mirabel & Rodríguez 1994), we can derive from the fits the inner radius of the accretion disk. Correction for interstellar absorption (fixed to cm-2, see Belloni et al. 2000) and an additional emission line (fixed at 6.4 keV) were also included. A systematic error of 1% was added. The value of the reduced was usually around 1, although some fits were slightly worse. The resulting interesting parameters (inner disk radius and temperature, slope of the power law) as a function of time are shown in Fig. 1 for three of the five observations, for which this automated procedure gave good results. The remaining two observations had to be treated more carefully. The observation from 1997 Sep 15th, the only one from class , resulted in an extremely strong power law component, with a photon index steeper than 3. The softness and intensity of this component made it impossible to obtain sensible values for the disk parameters, although there is evidence of its presence. This enhanced power law is probably the reason of the difference between this class and the others (see Belloni et al. 2000). The observation from 1998 July 10th did not include full state-C intervals: in this case, we measured the length of the intervals from the infrared (Eikenberry et al. 2000). Also, the inner disk radius resulted to be larger and therefore more difficult to measure as this component is softer. In order to estimate the disk parameters, we produced a 32s spectrum corresponding to the bottom of the dip only and obtained the best fit parameters, corresponding to the largest inner radius. This is the reason why there is only one point for this observation in Fig. 2.
© European Southern Observatory (ESO) 2000
Online publication: June 8, 2000