Confidential manuscript submitted to JGR 1 Pitch Angle Dependence of Energetic Electron Precipitation: 2 Energy Deposition, Backscatter, and the Bounce Loss Cone 1 2 3 R. A. Marshall and J. Bortnik 1 4 Ann and H. J. Smead Department of Aerospace Engineering Sciences, University of Colorado Boulder, Boulder, CO 5 80309, USA. 2 6 Department of Atmospheric and Oceanic Sciences, University of California Los Angeles, Los Angeles, CA 90095, USA. 7 Key Points: • 8 We characterize energy deposition and atmospheric backscatter of radiation belt 9 electrons as a function of energy and pitch angle • 10 We use these simulations to characterize the bounce loss cone and show that it is 11 energy dependent • 12 The simulated backscatter of precipitation is characterized by field aligned beams 13 of low energies which should be observable Corresponding author: R. A. Marshall, [email protected] –1– Confidential manuscript submitted to JGR 14 Abstract 15 Quantifying radiation belt precipitation and its consequent atmospheric effects re- 16 quires an accurate assessment of the pitch angle distribution of precipitating electrons, as 17 well as knowledge of the dependence of the atmospheric deposition on that distribution. 18 Here, Monte Carlo simulations are used to investigate the effects of the incident electron 19 energy and pitch angle on precipitation for bounce-period time scales, and the implica- 20 tions for both the loss from the radiation belts and the deposition in the upper atmosphere. 21 Simulations are conducted at discrete energies and pitch angles to assess the dependence 22 on these parameters of the atmospheric energy deposition profiles and to estimate the 23 backscattered particle distributions. We observe that the atmospheric response is both 24 energy and pitch angle dependent. These effects together result in an energy-dependent 25 bounce loss cone angle, which can vary by 2–3 degrees with particle energy when consid- 26 ered at low-Earth orbit. This modeling also predicts that a significant fraction of the input 27 electron distribution will be backscattered, and should be observable by low-Earth-orbiting 28 satellites as field aligned beams emerging from the atmosphere at energies lower than the 29 input distribution, and having pitch angles distributed just inside the loss cone. 30 1 Introduction 31 Energetic Particle Precipitation (EPP) from the inner magnetosphere into the Earth’s 32 atmosphere is both a source of energy for the upper atmosphere, and a sink for the radia- 33 tion belts and ring current. Both the upper atmospheric and magnetospheric communities 34 require an accurate assessment of EPP fluxes and spectra to understand the role played by 35 EPP in their respective domains. 36 EPP can significantly change the properties, dynamics, and chemical composition 37 of the upper and middle atmosphere. The chemical changes induced by EPP have long 38 been found to have implications for the production of atmospheric nitric oxides (NOx) 39 and reactive hydrogen oxides (HOx) [e.g., Sinnhuber et al., 2012], both of which can lead 40 to significant ozone losses in the stratosphere and mesosphere [e.g., Randall et al., 2007; 41 Rozanov et al., 2012; Seppälä et al., 2015]. HOx produced in the mesosphere has a short 42 lifetime, but can cause days-long ozone depletion of up to 90% [Andersson et al., 2014]. 43 The energy deposited by EPP in the mesosphere and lower thermosphere (MLT) can also 44 result in massive production of NOx. During the polar winter, these EPP-induced NOx 45 descends in the polar vortex into the stratosphere, leading to catalytic ozone destruction 46 [Callis et al., 1998; Randall et al., 2007]. 47 Despite the overwhelming evidence of EPP’s influence on the atmosphere, numeri- 48 cal models are incapable of capturing the effects satisfactorily. For example, in the Arctic 49 spring of 2004, an enormous influx of EPP-induced NOx was observed to descend from 50 the MLT into the polar stratosphere [Natarajan et al., 2004]. NOx mixing ratios in the 51 upper stratosphere increased by as much as a factor of 4, causing localized catalytic re- 52 ductions in ozone of more than 60% [Randall et al., 2005]. Randall et al. [2016] com- 53 pared the observed increase in NO2 in the upper stratosphere with results calculated using 54 the National Center for Atmospheric Research (NCAR) Whole Atmosphere Community 55 Climate Model (WACCM) [e.g., Marsh et al., 2007]. Modeling results were inconsistent 56 with the observed flux of NOx that descends into the stratosphere and the discrepancy was 57 attributed to a combination of inaccurate specification of the EPP flux and distribution, 58 and inadequate simulation of the vertical transport [Randall et al., 2016]. In particular, 59 WACCM simulation underestimated NOx in the mesosphere between about 60 and 80 km, 60 where electrons with energies near 50–300 keV typically deposit their energy [Codrescu 61 et al., 1997]. 62 Underestimation of EPP-induced NOx by WACCM is not entirely surprising, since 63 only those electrons with auroral energies (i.e. a few keV) were included in this model, –2– Confidential manuscript submitted to JGR 64 and higher energies (which are known to play a role) were omitted. This underestima- 65 tion is partly attributable to the lack of a satisfactory data set of higher-energy precip- 66 itation electron to be included in the modeling. Precipitating electrons in the relevant 67 energy range are mainly measured by the Medium Energy Proton and Electron Detector 68 (MEPED) [Evans and Greer, 2004] onboard the NOAA POES spacecraft. More recently, 69 observations have been made by the Balloon Array for Radiation-belt Relativistic Electron 70 Losses (BARREL) balloon mission via X-ray observations. The inversion from BARREL 71 measurements to precipitating electrons is extremely difficult to obtain uniquely as dif- 72 ferent precipitating electron distributions can fit the observed X-ray spectra equally well 73 [Halford et al., 2015; Clilverd et al., 2017]. On the other hand, MEPED measurements suf- 74 fer from proton contamination and poor spectral resolution, and do not adequately sample 75 the loss cone [Nesse Tyssøy et al., 2016; Peck et al., 2015; Rodger et al., 2010]. While the 76 Arctic winter of 2004 was remarkable with regard to the amplitude of the NOx descent 77 [Randall et al., 2005], such processes occur every winter at some level. 78 From a magnetospheric perspective, EPP is known to be one of the key loss mecha- 79 nisms for radiation belt enhancements following geomagnetic storms and substorms [e.g., 80 Tu et al., 2010; Thorne, 2010]. Radiation belt fluxes can be enhanced by orders of magni- 81 tude during these events [e.g., Baker et al., 2012], and high-energy particles above a few 82 hundred keV can be particularly damaging to high-altitude spacecraft [Baker et al., 2004]. 83 Understanding and predicting the loss rates of radiation belt fluxes during the recovery 84 phase of storms thus has direct implications for the overall understanding of radiation belt 85 dynamics, its impact on technology and hence better decision making in spacecraft opera- 86 tions. 87 Currently, uncertainties in the theoretical precipitation loss rates lead to large dis- 88 crepancy in electron lifetimes used in radiation belt models. The uncertainty in lifetimes is 89 attributed to a number of factors, including a lack of adequate observation of the waves 90 [e.g., Engebretson et al., 2008], the validity of the linear approach for modeling wave- 91 particle coupling [e.g., Bortnik et al., 2008], and uncertainty in the pitch angle distribu- 92 tion of electrons near the loss cone [e.g., Friedel et al., 2002; Millan et al., 2007; Tu et al., 93 2009]. The electron loss rates or lifetimes used in current papers vary by an order of mag- 94 nitude. For example, Shprits et al. [2005] used a constant lifetime of 10 days inside the 95 plasmapause and an empirical function of Kp (i.e., τ = 3/Kp) outside the plasmapause. 96 Conversely, Barker et al. [2005] used an L-dependent lifetime varying from 3 days at 97 L = 6 to 29 days at L = 4. Thorne et al. [2005] determined that the effective lifetime 98 is about one day in the outer radiation belt, based on microburst observations. Clearly, the 99 radiation belt electron lifetimes and their dependencies on magnetospheric conditions are 100 poorly understood. 101 Progress has been made in the area of wave distributions and the resulting diffusion 102 coefficients. Meredith et al. [2012] used multiple satellite observations to build a model of 103 chorus waves based on satellite observations; Glauert et al. [2014] used diffusion coeffi- 104 cients for whistlers, chorus, and hiss to build a global model of radiation belt dynamics. 105 Lifetimes based on wave distributions have been estimated by Agapitov et al. [2014] based 106 on Akebono wave data for the inner belt and slot region, Orlova and Shprits [2014] based 107 on chorus waves, and by Orlova et al. [2016] based on a model of hiss distribution derived 108 by Spasojevic et al. [2015]. These studies determined lifetimes ranging from ∼0.1–1000 109 days, varying with L-shell, energy, and Kp. It should be noted, however, that all of these 110 studies considered only quasi-linear (weak) diffusion and incoherent scattering, neglect- 111 ing the coherent interactions between intense waves that can have dramatic effects on pitch 112 angles within a single bounce period [e.g., Meredith et al., 2012]. 113 The precipitation of energetic particles in the upper atmosphere has been extensively 114 studied using parameterization methods [e.g., Roble and Ridley, 1987; Lummerzheim, 115 1992; Fang et al., 2008, 2010] and physics-based Monte Carlo simulations [e.g., Solomon, 116 2001; Xu et al., 2018].