A solar wind-parameterised, probabilistic model of ULF waves in Earth’s magnetosphere

Abstract:

Ultra-low frequency (ULF) waves are involved with the diffusion and energisation of radiation belt electrons. Current models of ULF wave power are deterministic, producing a single output for each set of input parameters. Meanwhile, weather and climate models are increasingly using stochastic parameterisations to account for the effects of sub-scale processes and model uncertainty. Given the large spatial and temporal scales involved in magnetospheric physics and the difficulty in accurately specifying initial conditions, including this approach in radiation belt models could lead to improved characterisation of physical processes. To apply stochastic parameterisation to radial diffusion in the radiation belts, we require a probabilistic forecast of the power in ULF waves in order to estimate diffusion coefficients.

We present a new statistical model of ULF wave power parameterised by solar wind conditions. Parameter reduction on the highly interdependent solar wind properties indicates that solar wind speed $v_{sw}$, variance in proton number density $var(N_p)$ and southward interplanetary magnetic field $B_z$ are the most ULF-effective solar wind parameters and are therefore suitable candidates for an instantaneous empirical model of ULF wave power. While $v_{sw}$ is the dominant driver, $B_z$ and $var(N_p)$ still account for significant amounts of power. We test the hypothesis that power spectral density distributions are consistently lognormal and use this assumption to construct a model of ULF wave power dependent on the incoming solar wind parameters and magnetic local time (MLT). We discuss metrics to validate such a model, including forecasting skill.