This paper presents an algorithm to decompose the modulated RHESSI light curves into periodic functions and a smooth function, representing the true (demodulated) time profile of an impulsive source. The decomposition is achieved by
optimizing a trade-off between the Poisson likelihood, a smoothness constraint, and conditions on the average grid transmission and the (modulated or non-modulated)
background. The algorithm, which operates on the level of count rates and does not require imaging information, is verified by numerical simulations and applied to some early RHESSI data, where - as a preliminary result - several impulsive features on time scales < 4 s may have been identified.
keywords: Signal processing; inverse methods; solar flares
Authors: Kaspar Arzner
Publication Status: Solar Physics, in press
Last Modified: 2002-10-09 10:39