Research Colloquium with Ken Hara (Stanford University)

Dynamic state and parameter estimation of partially ionized plasmas using sequential data assimilation

Ionized gases, also known as plasmas, play an important role in various natural phenomena and engineering applications. While high-fidelity physics-based models are essential to improving our understanding of complex plasma phenomena, an alternative is to use data-driven models to infer unknown properties using measurement data. The application of data assimilation to plasma physics is limited despite its potential impact. The two main challenges for plasma physics problems are that (i) the physical phenomena span across vastly different spatial and temporal scales and (ii) the measurement data may contain large uncertainties and are typically sparsely sampled. In this talk, I will introduce our in-house extend and ensemble Kalman filters that employ 0D global equations (i.e., ordinary differential equations) for collisional-radiative models and 1D hydrodynamic equations (i.e., partial differential equations) for advection-ionization problems.