Spectral estimation of plasma fluctuations. II. Nonstationary analysis of edge localized mode spectra

Abstract

Several analysis methods for nonstationary fluctuations are described and applied to the edge localized mode (ELM) instabilities of limiter H-mode plasmas. The microwave scattering diagnostic observes poloidal kθ values of 3.3 cm−1, averaged over a 20 cm region at the plasma edge. A short autoregressive filter enhances the nonstationary component of the plasma fluctuations by removing much of the background level of stationary fluctuations. Between ELM’s, the spectrum predominantly consists of broadband 300–700 kHz fluctuations propagating in the electron diamagnetic drift direction, indicating the presence of a negative electric field near the plasma edge. The time-frequency spectrogram is computed with the multiple taper technique. By using the singular value decomposition of the spectrogram, it is shown that the spectrum during the ELM is broader and more symmetric than that of the stationary spectrum. The ELM period and the evolution of the spectrum between ELM’s varies from discharge to discharge. For the discharge under consideration which has distinct ELM’s with a 1 ms period, the spectrum has a maximum in the electron drift direction which relaxes to a near constant value in the first half millisecond after the end of the ELM and then grows slowly. In contrast, the level of the fluctuations in the ion drift direction increases exponentially by a factor of 8 in the 5 ms after the ELM. High frequency precursors are found which occur 1 ms before the ELM’s and propagate in the ion drift direction. These precursors are very short (∼10 μs), coherent bursts, and they predict the occurrence of an ELM with a high success rate. A second detector, measuring fluctuations 20 cm from the plasma edge with kθ values of 8.5 cm−1, shows no precursor activity. The spectra in the ion drift direction are very similar on both detectors, while the ‘‘electron’’ spectrum level is significantly larger on this second detector.