AAS 196th Meeting, June 2000
Session 54. Applications of Statistics
Special Contributed Display, Thursday, June 8, 2000, 9:20am-4:00pm, Empire Hall South

## [54.07] Active Stellar Coronae: Lots of Little Flares?

J.J. Drake, V.L. Kashyap (SAO), M. Audard, M. Guedel (LA-PSI)

One of the outstanding questions in stellar X-ray astronomy is the nature of the apparently quiescent emission on active stars -- does this emission actually arise as a superposition of a multitude of impulsive events (such as microflares or nanoflares), or from truly quiescent plasma? This question has gone unanswered due to the difficulties associated with the analysis: [1] the vast majority of the flares that would make up the emission are by definition too weak to be detected, let alone be distinguished from each other; [2] the low count-rates coupled with the small duty cycles of the telescopes increase the deadtime'' due to strong flares, masking the superposed weaker flares; and [3] because of the stochastic nature of the onset of the flares the light-curve cannot be simply fit with a suitably parameterized model.

The distribution of microflares and nanoflares in the Sun are known to follow a power-law of the form dN/dE=k~E-\alpha (e.g.,Hudson 1991,Sol.Phys.133,357; Parnell & Jupp 2000,ApJ529,554). We adopt a similar model for stellar coronae, and have developed a method to determine the value of \alpha for the given dataset.

Because the model is stochastic, we cannot directly compare a simulated light-curve with the observed. Instead, we work directly on event lists (which has the additional advantage of being unbinned) and compare the distribution of photon arrival-time differences between the data and simulated model. Data gaps and deadtime corrections derived for the data set are applied to the simulations before making the comparisons.

Many criteria are available to characterize the level of agreement of a model and data. We have used the Kolmogorov-Smirnoff statistic, and the reduced \chi2 between the distributions of arrival-time differences as possible measures. We find that the KS statistic is unreliable for smaller values of \alpha, where the spread in the flare intensities is large, but works very well for \alpha >> 2. In contrast, the reduced \chi2 provides more reliable error-bars.

We show examples of application to EUVE Deep Survey observations of several late-type stars and discuss the results in the context of flare contributions to coronal heating.