Spatiotemporal multiscaling analysis of impurity transport in plasma turbulence using proper orthogonal decomposition

• Published in: Physics of plasmas 2009-04, Vol.16 (4)
• Main Authors: Futatani, S., Benkadda, S., del-Castillo-Negrete, D.
• Format: Article
• Language: English
• ISSN: 1070-664X; 1089-7674
• DOI: 10.1063/1.3095865

The spatiotemporal multiscale dynamics of the turbulent transport of impurities is studied in the context of the collisional drift wave turbulence. Two turbulence regimes are considered: a quasihydrodynamic regime and a quasiadiabatic regime. The impurity is assumed to be a passive scalar advected by the corresponding E × B turbulent flow in the presence of diffusion. Two mixing scenarios are studied: a freely decaying case, and a forced case in which the scalar is forced by an externally imposed gradient. The results of the direct numerical simulations are analyzed using proper orthogonal decomposition (POD) techniques. The multiscale analysis is based on a space-time separable POD of the impurity field. The low rank spatial POD eigenfunctions capture the large scale coherent structures and the high rank eigenfunctions capture the small scale fluctuations. The temporal evolution at each scale is dictated by the corresponding temporal POD eigenfunctions. Contrary to the decaying case in which the POD spectrum decays fast, the spectrum in the forced case is relatively flat. The most striking difference between these two mixing scenarios is in the temporal dynamics of the small scale structures. In the decaying case the POD reveals the presence of “bursty” dynamics in which successively small (high POD rank) scales are intermittently activated during the mixing process. On the other hand, in the forced simulations the temporal dynamics exhibits stationary fluctuations. Spatial intermittency or “patchiness” in the mixing process characterizes the distribution of the passive tracer in the decaying quasihydrodynamic regime. In particular, in this case the probability distribution function of the low rank POD spatial reconstruction error is non-Gaussian. The spatiotemporal POD scales exhibit a diffusive-type scaling in the quasiadiabatic regime. However, this scaling seems to be absent in the quasihydrodynamic regime that shows no scaling (in the decaying case) or two different superdiffusive-type scaling regimes (in the forced case).