-
Opening the Black Box of the Radiation Belt Machine Learning Model
Authors:
Donglai Ma,
Jacob Bortnik,
Xiangning Chu,
Seth G. Claudepierre,
Adam Kellerman,
Qianli Ma
Abstract:
Many Machine Learning (ML) systems, especially neural networks, are fundamentally regarded as black boxes since it is difficult to grasp how they function once they have been trained. Here, we tackle the issue of the interpretability of a high-accuracy ML model created to model the flux of Earth's radiation belt electrons. The Outer RadIation belt Electron Neural net model (ORIENT) uses only solar…
▽ More
Many Machine Learning (ML) systems, especially neural networks, are fundamentally regarded as black boxes since it is difficult to grasp how they function once they have been trained. Here, we tackle the issue of the interpretability of a high-accuracy ML model created to model the flux of Earth's radiation belt electrons. The Outer RadIation belt Electron Neural net model (ORIENT) uses only solar wind conditions and geomagnetic indices as input. Using the Deep SHAPley additive explanations (DeepSHAP) method, we show that the `black box' ORIENT model can be successfully explained. Two significant electron flux enhancement events observed by Van Allen Probes during the storm interval of 17 to 18 March 2013 and non storm interval of 19 to 20 September 2013 are investigated using the DeepSHAP method. The results show that the feature importances calculated from the purely data driven ORIENT model identify physically meaningful behavior consistent with current physical understanding.
△ Less
Submitted 19 January, 2023; v1 submitted 17 August, 2022;
originally announced August 2022.
-
Modeling the dynamic variability of sub-relativistic outer radiation belt electron fluxes using machine learning
Authors:
Donglai Ma,
Xiangning Chu,
Jacob Bortnik,
Seth G. Claudepierre,
W. Kent Tobiska,
Alfredo Cruz,
S. Dave Bouwer,
J. F. Fennell,
J. B. Blake
Abstract:
We present a set of neural network models that reproduce the dynamics of electron fluxes in the range of 50 keV $\sim$ 1 MeV in the outer radiation belt. The Outer Radiation belt Electron Neural net model for Medium energy electrons(ORIENT-M) uses only solar wind conditions and geomagnetic indices as input. The models are trained on electron flux data from the Magnetic Electron Ion Spectrometer (M…
▽ More
We present a set of neural network models that reproduce the dynamics of electron fluxes in the range of 50 keV $\sim$ 1 MeV in the outer radiation belt. The Outer Radiation belt Electron Neural net model for Medium energy electrons(ORIENT-M) uses only solar wind conditions and geomagnetic indices as input. The models are trained on electron flux data from the Magnetic Electron Ion Spectrometer (MagEIS) instrument onboard Van Allen Probes, and they can reproduce the dynamic variations of electron fluxes in different energy channels. The model results show high coefficient of determination $R^2 \sim $ 0.78-0.92 on the test dataset, an out-of-sample 30-day period from February 25 to March 25 in 2017, when a geomagnetic storm took place, as well as an out-of-sample one year period after March 2018. In addition, the models are able to capture electron dynamics such as intensifications, decays, dropouts, and the Magnetic Local Time (MLT) dependence of the lower energy ($\sim <$ 100 keV )electron fluxes during storms. The models have reliable prediction capability and can be used for a wide range of space weather applications. The general framework of building our model is not limited to radiation belt fluxes and could be used to build machine learning models for a variety of other plasma parameters in the Earth's magnetosphere.
△ Less
Submitted 6 June, 2022; v1 submitted 17 September, 2021;
originally announced September 2021.
-
Solar Wind Driving of Magnetospheric ULF Waves: Pulsations Driven by Velocity Shear at the Magnetopause
Authors:
S. G. Claudepierre,
S. R. Elkington,
M. Wiltberger
Abstract:
We present results from global, three-dimensional magnetohydrodynamic (MHD) simulations of the solar wind/magnetosphere interaction. These MHD simulations are used to study ultra low frequency (ULF) pulsations in the Earth's magnetosphere driven by shear instabilities at the flanks of the magnetopause. We drive the simulations with idealized, constant solar wind input parameters, ensuring that any…
▽ More
We present results from global, three-dimensional magnetohydrodynamic (MHD) simulations of the solar wind/magnetosphere interaction. These MHD simulations are used to study ultra low frequency (ULF) pulsations in the Earth's magnetosphere driven by shear instabilities at the flanks of the magnetopause. We drive the simulations with idealized, constant solar wind input parameters, ensuring that any discrete ULF pulsations generated in the simulation magnetosphere are not due to fluctuations in the solar wind. The simulations presented in this study are driven by purely southward interplanetary magnetic field (IMF) conditions, changing only the solar wind driving velocity while holding all of the other solar wind input parameters constant. We find surface waves near the dawn and dusk flank magnetopause and show that these waves are generated by the Kelvin-Helmholtz (KH) instability. We also find that two KH modes are generated near the magnetopause boundary. One mode, the magnetopause KH mode, propagates tailward along the magnetopause boundary. The other mode, the inner KH mode, propagates tailward along the inner edge of the boundary layer (IEBL). We find large vortical structures associated with the inner KH mode that are centered on the IEBL. The phase velocities, wavelengths, and frequencies of the two KH modes are computed. The KH waves are found to be fairly monochromatic with well defined wavelengths. In addition, the inner and magnetopause KH modes are coupled and lead to a coupled oscillation of the low-latitude boundary layer. The boundary layer thickness, $d$, is computed and we find maximum wave growth for $kd$ = 0.5--1.0, where $k$ is the wave number, consistent with the linear theory of the KH instability. We comment briefly on the effectiveness of these KH waves in the energization and transport of radiation belt electrons.
△ Less
Submitted 19 October, 2010;
originally announced October 2010.
-
Solar wind driving of magnetospheric ULF waves: Field line resonances driven by dynamic pressure fluctuations
Authors:
S. G. Claudepierre,
M. K. Hudson,
W. Lotko,
J. G. Lyon,
R. E. Denton
Abstract:
Several observational studies suggest that solar wind dynamic pressure fluctuations can drive magnetospheric ultra-low frequency (ULF) waves on the dayside. To investigate this causal relationship, we present results from Lyon-Fedder-Mobarry (LFM) global, three-dimensional magnetohydrodynamic (MHD) simulations of the solar wind-magnetosphere interaction. These simulations are driven with synthetic…
▽ More
Several observational studies suggest that solar wind dynamic pressure fluctuations can drive magnetospheric ultra-low frequency (ULF) waves on the dayside. To investigate this causal relationship, we present results from Lyon-Fedder-Mobarry (LFM) global, three-dimensional magnetohydrodynamic (MHD) simulations of the solar wind-magnetosphere interaction. These simulations are driven with synthetic solar wind input conditions, where idealized ULF dynamic pressure fluctuations are embedded in the upstream solar wind. In three of the simulations, a monochromatic, sinusoidal ULF oscillation is introduced into the solar wind dynamic pressure time series. In the fourth simulation, a continuum of ULF fluctuations over the 0-50 mHz frequency band is introduced into the solar wind dynamic pressure time series. In this numerical experiment, the idealized solar wind input conditions allow us to study only the effect of a fluctuating solar wind dynamic pressure, while holding all of the other solar wind driving parameters constant. We show that the monochromatic solar wind dynamic pressure fluctuations drive toroidal mode field line resonances (FLRs) on the dayside, at locations where the upstream driving frequency matches a local field line eigenfrequency. In addition, we show that the continuum of upstream solar wind dynamic pressure fluctuations drives a continuous spectrum of toroidal mode FLRs on the dayside. The characteristics of the simulated FLRs agree well with FLR observations, including a phase reversal radially across a peak in wave power, a change in the sense of polarization across the noon meridian, and a net flux of energy into the ionosphere.
△ Less
Submitted 19 October, 2010;
originally announced October 2010.
-
Magnetospheric Cavity Modes Driven by Solar Wind Dynamic Pressure Fluctuations
Authors:
S. G. Claudepierre,
M. Wiltberger,
S. R. Elkington,
W. Lotko,
M. K. Hudson
Abstract:
We present results from Lyon-Fedder-Mobarry (LFM) global, three-dimensional magnetohydrodynamic (MHD) simulations of the solar wind-magnetosphere interaction. We use these simulations to investigate the role that solar wind dynamic pressure fluctuations play in the generation of magnetospheric ultra-low frequency (ULF) pulsations. The simulations presented in this study are driven with idealized s…
▽ More
We present results from Lyon-Fedder-Mobarry (LFM) global, three-dimensional magnetohydrodynamic (MHD) simulations of the solar wind-magnetosphere interaction. We use these simulations to investigate the role that solar wind dynamic pressure fluctuations play in the generation of magnetospheric ultra-low frequency (ULF) pulsations. The simulations presented in this study are driven with idealized solar wind input conditions. In four of the simulations, we introduce monochromatic ULF fluctuations in the upstream solar wind dynamic pressure. In the fifth simulation, we introduce a continuum of ULF frequencies in the upstream solar wind dynamic pressure fluctuations. In this numerical experiment, the idealized nature of the solar wind driving conditions allows us to study the magnetospheric response to only a fluctuating upstream dynamic pressure, while holding all other solar wind driving parameters constant. The simulation results suggest that ULF fluctuations in the solar wind dynamic pressure can drive magnetospheric ULF pulsations in the electric and magnetic fields on the dayside. Moreover, the simulation results suggest that when the driving frequency of the solar wind dynamic pressure fluctuations matches one of the natural frequencies of the magnetosphere, magnetospheric cavity modes can be energized.
△ Less
Submitted 19 October, 2010;
originally announced October 2010.