A Neural Network Reconstruction of the Coupled Inner Magnetospheric Environment

Tuesday, 15 December 2015: 15:25
2018 (Moscone West)
Jacob Bortnik1, Wen Li1, Richard M Thorne1, Chao Yue1, Xiangning Chu1, Vassilis Angelopoulos2, Lauren W Blum3, Qianli Ma1, Craig Kletzing4, Geoffrey D Reeves5 and Harlan E. Spence6, (1)University of California Los Angeles, Los Angeles, CA, United States, (2)University of California Los Angeles, Earth, Planetary, and Space Sciences, Los Angeles, CA, United States, (3)University of California Berkeley, Berkeley, CA, United States, (4)University of Iowa, Iowa City, IA, United States, (5)Los Alamos National Laboratory, Los Alamos, NM, United States, (6)University of New Hampshire Main Campus, Space Science Center, Durham, NH, United States
The Earth’s inner magnetosphere represents a dynamic and highly coupled system that is often challenging for physical models to reproduce accurately, but knowledge of this system and its spatiotemporal evolution is crucial for both practical applications in the form of space weather, and scientific insight. In this presentation, we demonstrate the use of a deep neural network in predicting and specifying the global, time-varying distributions of a number of key wave and particle populations including chorus, hiss and magnetosonic wave, and electrion distributions from cold plasma to plasmasheet energies and right through the relativistic and ultra-relativistic populations. We show the temporal and spatial relationships between these waves and particle populations that ultimately lead to the dynamics of the ultra-relativistic particles and discuss implications.