next up previous contents
Next: Discussion Up: Spectral Analysis Previous: Time-averaged Analysis

Time-resolved Analysis

We use the 2-channel, 1-s ratemeters to gain information on short time scales spectral variation. Our results can be presented either in terms of an hardness ratio HR (counts in the 100-700 keV band divided by those in 40-100 keV) or of the equivalent temperature kT of an OTTB law. We remind that this is only a functional exponential spectral shape, since the object cannot be thin due to the compactness inferred from the fast variability. The first two seconds can be described by the extraordinary high kT=3200 keV evolving to 60 keV, or with a value of HR=4.06 decreasing to HR=0.75, respectively. (It should be noted, however, that the peak values of kT and HR could be affected by the approximation mentioned in paragraph 2 and therefore it could be significantly smaller.) The subsequent evolution is shown in Fig. 5.20 (bottom panel) for the time interval for which also the high time resolution data are available, both in terms of HR and of kT. We note that the HR and kT vary with the same $\sim$5-s periodicity observed in the light curve, with the hardness of the pulse increasing almost linearly with phase, in a ramp-like fashion, with distinct maxima correlated with the dips in the light curve.

Alternatively, we have also described the spectral evolution using a PL, deriving a variation of the photon index with time. We find that the first two seconds the 2-channel spectrum can be approximated by a photon index $\alpha$ = 0.7, evolving to about 3 in the next two seconds and then varying from 4 to 5 in the subsequent $\sim$100 s and between 5 and 6 in the rest of the event.


next up previous contents
Next: Discussion Up: Spectral Analysis Previous: Time-averaged Analysis
Lorenzo Amati
8/30/1999