Reconstruction of Zeff profiles at TEXTOR through Bayasian source separation

Abstract

The understanding of the behaviour of impurities is a critical issue in tokamak physics. The ion effective charge Zeff provides a measure for impurity concentration. On the TEXTOR tokamak (Julich, Germany), we run a diagnostic to determine Zeff from the bremsstrahlung emissivity E. From radial profiles of E, electron density ne and temperature Te, profiles for Zeff can be reconstructed. However, their interpretation is difficult outside the plasma centre, because of various uncertainties in E, ne and Te at the edge, which render the radial matching of the different profiles problematic. Conversely, if it were possible to obtain a set of line-integrated values for Zeff directly from the line-integrated measurements of E, ne and Te, then these problems would be avoided. Now, recent advances in the field of statistical signal processing allow the extraction of an unknown signal from a signal mixture. In particular, we describe a procedure for the single-channel Bayesian source separation of a line-integrated Zeff signal from a line-integrated emissivity source, using as a forward model a linearized version of the known functional dependence of Zeff on E, ne and Te. Here, a line-integral over a traditionally obtained Zeff profile may serve as a prior for the line-integrated Zeff signal. In this way, precise information on the electron density and temperature may even become superfluous for the determination of Zeff.