5. Summary and conclusions
We have presented a sample of four untypical flare events and provided a common explanation: Parameters that depend on the size of the emitting plasma volume (e.g. count rate, EM) deviate from the standard exponential decay behavior due to temporary occultation of the flaring volume by the rotating star. This is most evident from the large flare on Algol, for which the data are most abundant and our modeling is therefore most reliable.
The comparatively slow rise of the count rate in the X-ray lightcurve, broad maxima and following exponential decay are well represented by a model that describes emission from a spherical plasma loop that emerges from the back of the star and gradually rotates into the line of sight of the observer. The increasing visible fraction of the loop produces the flat maximum and apparent slow-down of the rise stage.
In our data there is no indication for sine-like modulation of the X-ray lightcurve, since all but one of the lightcurves are clearly asymmetric, and the duration of all events is well below the rotational period of the host star. The only exception is the flare on V773 Tau where Skinner et al. (1997) proposed sinusoidal modulation to reproduce the shape of the ASCA lightcurve. We suggest a different explanation involving a second X-ray emitting region on the star additionally to a rotationally modulated flare to come up for the `convex' shape of the lightcurve. However, due to the lack of pre-flare data, no conclusive evidence is present for either of the interpretations. Evidence for reheating of the plasma at inferred from the increase of the hardness ratio (see Skinner et al. 1997) does not contradict our model, but could be related to the disappearing of a region emitting soft X-rays similar to the one we introduce. No significant change in temperature nor emission measure of the soft component was observed in Skinner's spectral analysis, but we note that the results of spectral fitting depend on the assumption for the abundances and column density.
The decay timescales found from our best fit to the respective flare event are all in the typical range for TTS flares (of a few hours) except for V773 Tau, where the flare lasted extraordinary long (). Comparatively large loop sizes of a considerable fraction of the radius of the star are obtained for all observations analysed here, r spanning between 10 to 65 % of the star radius. These values are in agreement with typical loop sizes for TTS flares inferred from quasi-static loop modeling (see Montmerle et al. 1983, Preibisch et al. 1993).
In view of the large relative size of the ratio between loop and star radius the assumptions we explain in Sect. 3 concerning our model for a rotating flare might seem somewhat oversimplifying. We also note that different solutions of the model seem to describe the data equally well even in the case of the well restrained Algol observation. Therefore, uncertainties in the fit parameters are to be regarded carefully. However, the qualitative description of the scenario is very good and the data are well represented by the model. Other interpretations of the `anomalous' flare events we presented in this paper may not be excluded but are still to be traced.
Clearly, continuous observations of whole flares are needed to verify whether an event could be subject to rotational modulation of the kind we discussed in this paper. Up to date most of the flares observed lack completeness in that either the rise or part of the decay were missed by the observation. In the near future XMM will provide the possibility of long, uninterrupted observations (up to h) that will enable to pursue the development of flares in whole. Better statistics are needed to be able to study the time development of spectral parameters for TTS flares, and try to confirm the `rotating flare model' by use of the spectral information similar to our analysis of the Algol observation.
© European Southern Observatory (ESO) 1999
Online publication: March 10, 1999