See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/316471118

A Monte Carlo Model of Crustal Field Influences on Solar Energetic Particle

Precipitation into the Martian Atmosphere

Article in Journal of Geophysical Research: Space Physics · April 2017

DOI: 10.1002/2016JA023781

CITATIONS READS

13 222

9 authors, including:

Rebecca Dawn Jolitz Chuanfei Dong

University of California, Berkeley Boston University
7 PUBLICATIONS 642 CITATIONS 248 PUBLICATIONS 4,288 CITATIONS

SEE PROFILE SEE PROFILE

Christina O. Lee Robert Lillis

125 PUBLICATIONS 2,859 CITATIONS

University of California, Berkeley
271 PUBLICATIONS 7,467 CITATIONS
SEE PROFILE
SEE PROFILE

All content following this page was uploaded by Chuanfei Dong on 20 November 2017.

The user has requested enhancement of the downloaded file.

 PUBLICATIONS
Journal of Geophysical Research: Space Physics
RESEARCH ARTICLE A Monte Carlo model of crustal field influences
10.1002/2016JA023781
on solar energetic particle precipitation
Key Points:
• Solar energetic particle precipitation
into the Martian atmosphere
can ionize and heat atmospheres of R. D. Jolitz1, C. F. Dong2 , C. O. Lee3, R. J. Lillis3 , D. A. Brain1 , S. M. Curry3 , S. Bougher4 ,
weakly magnetized planets
• A simulation was developed to predict C. D. Parkinson4, and B. M. Jakosky1
atmospheric effects due to solar 1
energetic particle precipitation into Laboratory for Atmospheric and Space Physics, University of Colorado Boulder, Boulder, Colorado, USA, 2Department of
the Martian atmosphere Astrophysical Sciences and Princeton Plasma Physics Laboratory, Princeton University, Princeton, New Jersey, USA, 3Space
• Less atmospheric ionization occurs Sciences Laboratory, University of California, Berkeley, California, USA, 4Department of Atmospheric, Oceanic, and Space
near crustal field anomalies than
Sciences, University of Michigan, Ann Arbor, Michigan, USA
unmagnetized areas of Mars

Abstract Solar energetic particles (SEPs) can precipitate directly into the atmospheres of weakly
Correspondence to: magnetized planets, causing increased ionization, heating, and altered neutral chemistry. However, strong
R. D. Jolitz, localized crustal magnetism at Mars can deflect energetic charged particles and reduce precipitation. In
[email protected]
order to quantify these effects, we have developed a model of proton transport and energy deposition in
spatially varying magnetic fields, called Atmospheric Scattering of Protons and Energetic Neutrals. We
Citation: benchmark the model’s particle tracing algorithm, collisional physics, and heating rates, comparing
Jolitz, R. D., C. F. Dong, C. O. Lee,
against previously published work in the latter two cases. We find that energetic nonrelativistic protons
R. J. Lillis, D. A. Brain, S. M. Curry,
S. Bougher, C. D. Parkinson, and precipitating in proximity to a crustal field anomaly will primarily deposit energy at either their stopping
B. M. Jakosky (2017), A Monte Carlo altitude or magnetic reflection altitude. We compared atmospheric ionization in the presence and absence
model of crustal field influences on solar
of crustal magnetic fields at 50°S and 182°E during the peak flux of the 29 October 2003 “Halloween storm”
energetic particle precipitation into the
Martian atmosphere, J. Geophys. Res. SEP event. The presence of crustal magnetic fields reduced total ionization by ~30% but caused ionization
Space Physics, 122, 5653–5669, to occur over a wider geographic area.
doi:10.1002/2016JA023781.

Received 23 DEC 2016 1. Introduction

Accepted 25 APR 2017
Accepted article online 28 APR 2017 Solar energetic particles (SEPs) are charged particles accelerated by the moving shock source ahead of a cor-
Published online 20 MAY 2017 onal mass ejection or by the local shock source at the solar flare site. They primarily consist of protons and
electrons with energies from a few keV to up to several GeV. SEPs that precipitate into planetary atmo-
spheres can have a substantial effect on ionospheric and neutral chemistry. In the terrestrial atmosphere,
precipitating SEPs are known to cause heating, ionospheric outflow, and increased atmospheric ionization
[Moore et al., 1999; Usoskin et al., 2009]. This in turn leads to enhanced electrical conductivity in the atmo-
sphere, changes in lower atmosphere electrochemistry such as increased ozone concentrations via hydroxyl
ion formation, and formation of aerosol particles between altitudes of 10 and 25 km [Kokorowski et al., 2006;
Damiani et al., 2010; Mironova and Usoskin, 2014]. However, all of these effects are typically confined to the
poles, as Earth’s strong global magnetic field shields equatorial latitudes from energetic particles [Mironova
et al., 2015].
Mars, on the other hand, lacks a strong intrinsic magnetic field. Charged particles can precipitate into the
atmosphere if they are accelerated past the compressed magnetic fields of the magnetosheath and bow
shock formed by solar wind plasma draping over the dayside ionosphere [Acuña et al., 2001]. Unlike solar
wind particles, SEPs are mostly unaffected by this barrier. Phobos-2 measurements showed energetic particle
fluxes recorded during a SEP event crossed the region without substantial change [McKenna-Lawlor et al.,
1998]. Subsequent modeling found that the barrier does not attenuate SEP protons with energies greater
than 50 keV [Leblanc et al., 2002].
There are two main obstacles to SEP entry into the atmosphere. The first is physical obstruction by the planet
or “shadowing,” which most strongly affects extremely energetic (>1 MeV) SEPs that are not significantly
influenced by magnetic fields. Measurements from Mars Odyssey found that a combination of planetary
shadowing, anisotropy of SEPs, and detector field of view caused orbital modulations of 20–200 MeV SEPs
©2017. American Geophysical Union. [Luhmann et al., 2007]. The second obstacle is the localized crustal magnetic anomalies in the southern hemi-
All Rights Reserved. sphere that present strong magnetic fields capable of reflecting incident SEPs [Acuña et al., 2001]. Prior

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5653

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

modeling studies on crustal anomaly influences on SEP fluxes suggest that the strongest field anomalies can
attenuate SEP fluxes by over 50% at altitudes of 200 km [Leblanc et al., 2002].

Compared to Earth, information on the atmospheric effects caused by SEP precipitation is relatively limited.
Up until the Mars Atmosphere and Volatile EvolutioN (MAVEN) spacecraft’s arrival at Mars, evidence of SEP
precipitation was inferred from indirect observations during SEP events. SEP fluxes were determined by using
background count rates in plasma analyzers as proxies for 20–100 MeV protons [Delory et al., 2012; Futaana
et al., 2008]. Radar ground reflections observed by the Mars Advanced Radar for Subsurface and Ionospheric
Sounding (MARSIS) instrument on Mars Express disappeared during SEP events, suggesting that SEP precipi-
tation increased ionization [Morgan et al., 2008; Espley et al., 2007]. The Electron Reflectometer on Mars Global
Surveyor measured increased upward fluxes of superthermal electrons linked to SEP events [Lillis et al., 2012].
However, further analysis of radio occultation data during periods of high SEP activity only found small
increases of electron density below altitudes of 100 km in four out of six SEP events and no increase in elec-
tron densities between altitudes of 100 and 200 km in all six events [Ulusen et al., 2012]. SEP events were
linked to ionization and electron production, but the exact mechanism remained unknown.
In 2014, MAVEN arrived at Mars carrying a suite of instruments designed to study the upper atmosphere and
plasma environment, including a detector called solar energetic particle, or SEP, to directly monitor SEPs. The
SEP instrument has been used to further characterize the near-Mars energetic particle environment.
Measurements of anisotropic high-energy (>200 keV) SEP flux attenuation have confirmed Mars Odyssey
observations of planetary shadowing. SEP has also observed complex low-energy shadowing governed by
local magnetic field geometry [Lillis et al., 2016]. The SEP instrument has also made measurements of SEP
electrons and ions during solar flare and coronal mass ejection events [Lee et al., 2017]. Coincident measure-
ments of SEPs and atmospheric emission have led to the discovery of SEP electron-linked diffuse aurora
[Schneider et al., 2015].
SEP events are frequently coincident with changes in solar X-ray and extreme ultraviolet fluxes. Modeling stu-
dies can decouple these changes and isolate the causation of atmospheric effects like electron production
and aurora.
SEP precipitation has been previously modeled using approximations. Leblanc et al. [2002] used a 1-D
stopping-and-range-of-ions-in-matter (SRIM) model to calculate the total heating by SEPs in the atmosphere
between altitudes of 80 and 170 km. Their model predicted SEP heating rates that were less than a tenth of
heating rates due to solar ultraviolet fluxes, meaning SEP precipitation only increased atmospheric heating
by ~30%. Sheel et al. [2012] used a stopping approximation that assumed all energy is deposited at a fixed
altitude for a specific SEP energy. Once they predicted the total energy deposited for each trace species in
the atmosphere, they determined ion densities by assuming an ion-electron pair is produced for every
35 eV deposited. They predicted that electron densities could reach 3000 cm
3 at 80 km during a SEP event
that was 100 times weaker than the total flux measured during the 29 September 1989 SEP event, which
would be sufficient to cause the MARSIS radar reflection disappearance. These approaches are mutually
exclusive, as Sheel et al. [2012] assume that SEP energy deposition is solely imparted to ionization and
Leblanc et al. [2002] assume that SEP energy deposition is imparted either to heating or electronic excitation.
SEP transport to the surface has also been modeled to predict the surface radiation environment during an
event. Leblanc et al. [2002] estimate that only protons with energies greater than 83 MeV can make it to the
surface. Because only extremely energetic particles can reach the surface, surface flux models need to
account for nuclear processes. Planetocosmics and HZETRN (High-charge (Z) and Energy TRaNsport) are
two such models [Desorgher et al., 2005; Norman et al., 2013; Gronoff et al., 2011]. Both models are in close
agreement, predicting ionization rates of ~1000–2000 cm
3 s
1 below 60 km during the 29 September
1989 SEP event [Gronoff et al., 2015].
Existing models of SEP energy deposition do not account for differential energy deposition into heating, exci-
tation, ionization, and charge transfer like the model presented in Kallio and Barabash [2001]. Others do not
consider possible effects caused by proton gyromotion in crustal magnetic anomalies.
In this work, a three-dimensional Monte Carlo model of SEP precipitation called Atmospheric Scattering of
Protons and Energetic Neutrals (ASPEN) is introduced to track transport and energy deposition of SEP protons
in the Martian atmosphere. We will present a verification of tracing particle motion in magnetic fields and

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5654

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

validation of collision modeling. We will next discuss transport processes as a function of SEP proton energy,
focusing on influences by crustal fields. Finally, we will use the ASPEN model results to predict ionization alti-
tude profiles, maps, and secondary electron spectra during a specific SEP event.

2. Model Description
ASPEN uses Monte Carlo methods to predict energy deposition into a planetary atmosphere via charged
particles, accounting for both particle motion in three-dimensional inhomogeneous electromagnetic
fields and collisions with neutral molecules in the atmosphere. While existing GEANT4-based models like
Planetocosmics are capable of this, we did not use them because we did not require their in-depth modeling
of relativistic effects such as bremsstrahlung. ASPEN steps an incident particle through a planetary atmo-
sphere and electromagnetic field until the calculated likelihood of collision along its path exceeds a randomly
chosen collision probability. When that occurs, the consequences of the collision are determined (e.g., energy
imparted to the atmosphere), and the particle begins stepping again until the particle energy falls below a
given energy threshold or the particle exits the simulation boundaries.
The model has three key assumptions. First, neutral atmospheric molecules are stationary relative to each
simulated particle. Second, simulated charged particles never interact with other charged particles. These
assumptions restrict ASPEN to operating in altitudes where the density of simulated charged particles does
not exceed the density of neutral targets, e.g., below 500 km. Third, electromagnetic fields and atmospheric
densities do not temporally vary over the lifetime of a simulated particle.
To study atmospheric effects due to precipitating SEPs near crustal field anomalies, protons with energies up
to 5 MeV are simulated and allowed to scatter between the altitudes of 70 km and 500 km. We focused on
SEPs with energies below 5 MeV since they were most likely to be affected by crustal magnetic fields. By
approximating the likelihood of an elastic collision as the ratio of the elastic cross section to the total cross
section, we assumed that protons with energies that fall below 10 eV are no longer traced and are assumed
to be thermalized, as a 10 eV proton has a 99.989% chance of imparting its energy to heat.
2.1. Model Dynamics
Individual protons are traced and allowed to collide with the atmosphere until they leave the preset altitude
range or fall below the energy threshold. Each has an assigned fixed three-dimensional trajectory and loca-
tion. They move according to the Lorentz force equation for a particle of charge q and mass m with position
!s and velocity !v:

d!  !
 !
!
¼q !v B !
v
m s þ E s
dt

A fourth-order Runge-Kutta method is used to solve this differential equation and approximate the change in
v ) and position (d!
velocity (d! s ) after elapsed time (dt). By repeatedly solving this equation using a time step
dt, ASPEN determines a charged particle’s Lorentz path.
To ensure the time step traces fine particle motion in complex field configurations, dt is set equal to
 
dl=!v  where ! v is the particle’s velocity and dl is a spatial step length. Using a step length that is less
than half of the shortest simulated proton’s gyroradius in the strongest crustal fields ensures that the
simulation correctly predicts gyromotion by all simulated protons. The lowest-energy particle simulated
in this study is a 1 keV proton, which has a gyroradius of 5.9 km at the greatest magnetic field value
(0.7 μT) in the multifluid magnetohydrodynamics (MF-MHD) magnetic field mesh. Thus, we require that
step length dl does not exceed 1 km.
The step length dl is also constrained by the distance between collisions. While particle motion is determinis-
tic, particle collisions with the atmosphere are stochastic. The probability of a projectile colliding with a sta-
tionary target is random. However, a particle is more likely to collide if it travels in a denser medium or has a
higher interaction cross-section area with a target. When applied to a particle beam of intensity I, this relation
is known as Beer’s law:

dI ¼
nσIdl

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5655

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

where n is the density of scattering targets, σ is the cross-sectional area presented by each scattering target, I
is the beam intensity, and dl is the distance over which a beam attenuates. Following an approach for
electron-neutral collisions from Lillis et al. [2008], we solved Beer’s law for intensity and set the collision
probability equal to the ratio of attenuated beam intensity to the initial beam intensity. For a particle traveling
distance l and encountering N atmospheric species with spatially varying densities, this gives a probability of
scattering of

" #
X
N
∫ σ ðUðl ÞÞn ðl Þdl
l
0 0 0
Pðl; UÞ ¼ exp
0 i i
i

where σ i(U(l0 )) is the cross section presented to a proton with energy U(l0 ) and ni(l0 ) is the number density of
the ith atmospheric neutral species encountered by the proton after traveling distance l0 between 0 and l. By
choosing a random probability for the left-hand side of the equation, we can use the equation to determine
the distance l a particle travels before colliding, which is effectively a weighted random number. Because
atmospheric densities and cross sections for each species may vary along a particle’s path (i.e., along a cur-
ving path in magnetic field), we cannot directly solve this equation to find distance l for a random collision
probability P. Instead, ASPEN implicitly solves for distance l between collisions by selecting a random number
between 0 and 1 and incrementally moving the particle by dl0 until the calculated probability of scatter
exceeds the random probability on the left-hand side.
When a collision occurs, the name of the impacted neutral species is determined by comparing a new ran-
dom number to a distribution weighted by collision probability per species. Likewise, the most likely process
that occurred as a result of the collision is selected. ASPEN currently considers five processes: elastic collision,
excitation (emission of a photon), ionization (ejection of an electron from the impacted neutral atom), charge
exchange (capture of an electron from the impacted neutral), and electron stripping (ionization of a fast
neutral hydrogen projectile). Each process causes the projectile to lose energy and scatter. ASPEN tracks
where and how much energy is deposited by each collisional process, including the location and energy
of any produced secondary electrons.
Following a collision, the particle’s velocity is recalculated to account for lost energy and a new scattering
direction before resuming tracing. The particle identity can also change, as the collision can change the state
of the proton to a neutral hydrogen atom (charge exchange) and back (electron stripping). If the particle
becomes a neutral hydrogen atom, the particle travels in a straight line unaffected by electromagnetic fields,
which we trace using the same step length as a proton with the same energy.
A step length that accommodates gyromotion is not short enough to approximate distances between
collisions in dense regions of an atmosphere. A 1 km step length is sufficient in the Martian atmosphere
above an altitude of 300 km, since the highest neutral densities are ~107 cm
3 and the maximum cross
section presented by atmospheric constituents is approximately 10
15 cm2, giving a distance of 1000 km
before a single collision occurs. However, a particle traveling in the atmosphere at an altitude of 80 km
encounters neutral densities approaching 1014 cm
3, reducing the approximate distance between collisions
to 10 cm. Using the larger distance as a step length undercounts collision but, since the simulation runtime is
proportional to the number of steps taken because it solves the Lorentz force equation after each step, using
the shorter step length increases simulation runtime by 105. To account for both situations, we use an
adaptive step length. If the cumulative collision likelihood increases too rapidly (e.g., over 20% in a single
step), ASPEN reduces the step length by half and tries the step again. Likewise, if the cumulative collision like-
lihood increases too slowly, ASPEN doubles the step length up to the maximum step length required to
simulate gyromotion.

2.2. Model Inputs

ASPEN requires four inputs. The first two are specific to a planetary atmosphere: neutral densities and
magnetic fields. Electric fields can be optionally included. All should correspond to the same solar wind
condition and orientation of Mars. The other two inputs are specific to the simulated projectile ion and the
target neutral species: inelastic and elastic cross sections as a function of projectile energy and energy losses
associated with each inelastic process, including secondary particle production for inelastic processes (eV).

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5656

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

2.2.1. Planetary Atmospheric Densities and Electromagnetic Fields

Number densities of each atmospheric constituent are either read from an altitude-varying profile or are
approximated by a fixed value. Likewise, magnetic and electric fields are imported as an altitude-varying pro-
file, globally varying profile, or a fixed value.
Planetary inputs can be toggled on or off to study specific cases. To examine particle motion in electric and
magnetic fields ignoring atmospheric collisions (e.g., for verifying Lorentz trace accuracy in section 4.1),
ASPEN can run in a trace-only mode that requires only electromagnetic fields. To study energy deposition
in the absence of planetary electric and magnetic fields (e.g., for comparison to results in previously pub-
lished studies on SEP energy deposition), ASPEN can run in a collision-only mode that only requires a
density profile.

In this paper we use an altitude-varying density profile from MTGCM (Mars Thermosphere Global Circulation
Model) of dayside Mars during solar minimum for the four most common species: CO2, O, N2, and CO
[Bougher et al., 2000, 2006]. We also use spatially varying 3-D magnetic and electric fields from the MF-
MHD model [Dong et al., 2014] including crustal fields with the strongest field region on the dayside of
Mars during solar minimum. The MF-MHD simulation domain is a nonuniform spherical grid structure with
a radial resolution varying from 5 km near the inner boundary to 1000 km near the outer boundary.
2.2.2. Collisional Cross Sections
ASPEN requires cross sections for each proton-impact process to model atmospheric effects. Because the
maximum simulated energy is 5 MeV, relativistic collisional physics can be ignored. Protons with energies
below 5 MeV undergo four collisional processes: excitation, ionization, charge transfer, and elastic colli-
sions. From the ratio of the elastic cross section to the total cross section, we know that low-energy
(~200 eV) protons undergo elastic collisions over 50% of the time. Protons with energies exceeding
200 eV are capable of imparting their energy to atomically interact with the impacted neutral molecule:
capturing an electron to become a neutral (charge exchange), ripping an electron off the neutral (ioniza-
tion), and emitting a photon (excitation). Protons that become neutral hydrogen are capable of the same
processes except electron capture. Instead, the previously captured electron in neutral hydrogen can be
ejected in a charge-stripping process.

ASPEN uses elastic collision, charge exchange, charge stripping, and ionization cross sections presented in
Kallio and Barabash [2001], which were originally presented in Basu et al. [1987]. It also uses the excitation
cross sections presented in Basu et al. [1987], which generalized all excitation processes under one cross sec-
tion. As in Kallio and Barabash [2001], proton- and hydrogen-impact cross sections with carbon dioxide are
approximated to be identical to proton- and hydrogen-impact cross sections with molecular oxygen. While
several proton-impact cross sections with carbon dioxide have been discovered since the study done in
Kallio and Barabash [2001], most cross sections for hydrogen-impact cross section with carbon dioxide
remain unknown.
2.2.3. Collisional Energy Loss, Scattering, and Secondary Production
Protons lose energy when they impact neutral species. Energy loss in elastic processes is traditionally
neglected because protons are light compared to the neutral molecules, as a proton has 2% the mass of a
carbon dioxide atom. However, if a higher-energy proton undergoes an elastic collision, the energy imparted
to heating is a fraction of that high energy. Likewise, particles are traditionally assumed to scatter forward.
However, there is backscatter associated with elastic collisions. ASPEN therefore uses the scattering distribu-
tion presented in Kallio and Barabash [2001], based on modeled differential scattering cross sections from
Noël and Prölss [1993], and calculates the energy lost using the conservation of energy and momentum
equations for a two-body collision.
Energy losses associated with inelastic processes are documented in Basu et al. [1993]. To recap, energy lost in
a charge exchange or excitation collision is fixed. In a charge exchange it is the difference between the
impacted neutral ionization threshold and the proton ionization threshold. In an excitation, it is the upper
threshold excitation energy of the impacted neutral. Energy lost in an electron ejection process is the sum
of the ejected (secondary) energy and the energy cost to remove the electron. This can happen in two ways.
The cost in an ionization collision is the ionization threshold of the impacted neutral. In a charge-stripping
collision, the cost is the ionization threshold of the projectile neutral hydrogen. In both, the ejected secondary
electron energy depends on projectile energy.

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5657

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

There are two approaches to predict

secondary electron energies. In Basu
et al. [1993] and Kallio and Barabash
[2001], ionizations and electron-
stripping collisions produce a sec-
ondary electron with a fixed energy
depending on the primary energy
(e.g., a 10 keV proton impacting N2
in a charge-stripping collision pro-
duces a 6.4 eV electron). The fixed
energy approach uses secondary
electron energy data from Basu et al.
[1993]. Alternatively, secondary elec-
tron production can be modeled
using differential cross sections as in
electron precipitation codes [e.g.,
Lillis et al., 2008] allowing for the
production of a range of secondary
energies for a given projectile energy
using random numbers. The random
energy approach uses proton-impact
Figure 1. Schematic of proton transport in the ASPEN simulation. Particles ionization double differential cross
are traced from 500 km and allowed to collide with neutral atmospheric sections with CO2, O2, O, and N2 from
species until they magnetically reflect without colliding, reflect after some Rudd [1988]. Unfortunately, there are
collisions, collide until they have an energy below 10 eV and thermalize, or no double differential cross sections
reach 70 km while still having an energy greater than 10 eV.
for electron production in electron-
stripping collisions.
ASPEN uses a fixed secondary approach for electron-stripping collisions and supports both fixed and double
differential secondary production for ionization collisions. The approaches will be compared in section 5.2.
2.3. Model Outputs
ASPEN logs every simulated proton’s energy and location when they exit the simulation, allowing us to verify
the simulation logic by ensuring protons are not traced after physically leaving the simulation shell or falling
below the energy threshold. The log is also useful to understand if and how injected protons deposit their
energy. ASPEN particles “magnetically reflect” without depositing any energy, deposit their entire energy
and “thermalize,” or deposit a fraction of their energy before turning away and “scatter reflect” (Figure 1).
Only particles with energies exceeding 5 MeV can reach the lower boundary of the simulation before depos-
iting all their energy.
The primary output of ASPEN is a database containing detailed information on every simulated collision: lati-
tude, longitude, and altitude of collision; species impacted; process occurred; whether the impacting particle
was a proton or neutral hydrogen; energy deposited; secondary energy if secondary produced. There exists a
single data structure for each incident SEP proton energy, location, and trajectory. The data structure contains
the collisions for all simulated particles. As ASPEN is a Monte Carlo simulation, increasing the number of par-
ticles simulated improves model predictions. Optionally, ASPEN can be used to build a database containing
each simulated particle’s traced path, including positions and energies. These paths can be plotted to better
understand particle behavior in the presence or absence of magnetic fields.
The data structures are aggregated over all simulated particles to predict observable properties. For
example, CO2+ production rates for 1 keV solar wind protons can be plotted as a function of altitude.
Furthermore, by integrating histograms of specific atmospheric processes over measured proton fluxes,
ASPEN can predict rates of atmospheric effects caused by precipitating protons for any SEP event. This
paper will focus on SEP-induced atmospheric ionization either in the absence of strong magnetic fields
or nearby a specific crustal magnetic anomaly, including altitude profiles, geographical maps, and second-
ary electron spectra.

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5658

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

3. Verification and
Validation
Because each procedure in the simu-
lation code was written de novo,
ASPEN was systematically tested to
ensure the model predicts reason-
able results. In this section, the verifi-
cation of the Lorentz tracing and
adaptive step algorithms will be
described. Also described is the vali-
dation of collisional physics used in
the model via comparison with two
Figure 2. Traces of protons with different energies traversing an artificial distinct atmospheric proton simula-
1 μT z-directed field between altitudes 50 km and 300 km. Each particle is
tions. Finally, neutral heating rates
incident at particle zenith angle 60° (particle clock angle 0°) with energies
100 eV (dark blue), 1 keV (blue), 10 keV (cyan), 100 keV (green), 1 MeV calculated during a sample SEP event
(yellow), and 10 MeV (orange). are compared to rates presented in
Leblanc et al. [2002].
3.1. Model Dynamics Verification
Charged particles gyrate around magnetic field lines with a fixed gyroradius in a uniform magnetic field. If a
proton is traveling transverse and parallel to the magnetic field direction, it will travel in a helix. To verify that
the Lorentz tracing algorithm predicted particle motion correctly, we used ASPEN’s trace-only mode to gen-
erate the paths of protons with energies between 100 eV and 10 MeV traveling at a 60° angle to a 1 μT vertical
magnetic field (Figure 2). The predicted gyroradius of each particle matches the calculated gyroradius within
an error of up to 1.6% at the lowest-energy simulated particle.
ASPEN’s adaptive step algorithm accounts for short distances between collisions without the computational
overhead of using the shortest distance between collisions as a fixed step length. To confirm that the adap-
tive step algorithm is as effective as a small fixed step, we disabled Lorentz tracing to simulate one thousand
1 keV incident protons using either four fixed step lengths between 1 m and 1 km or the adaptive algorithm
(Figure 3). Using the adaptive algorithm predicts similar energy deposition to energy deposition predicted
using a step length shorter than
50 m. The adaptive algorithm took
25 min to predict that the atmo-
sphere absorbs 80% of the 1 keV
proton beam’s energy via collisional
processes and scatters the rest, while
using a 1 m fixed step length took
10 h to predict the same result.
We also observe that fixed step
lengths longer than 10 m predict
energy depositions that do not
match either adaptive or short step
length results. Using these step
lengths predicts less energy deposi-
tion around the peak deposition alti-
tude, as particles miss collisions
occurring below 120 km and instead
deposit energy far below the
stopping altitude.

3.2. Collisional Physics Validation

Figure 3. Comparison of energy deposition by a thousand 1 keV protons in
the Martian atmosphere assuming no magnetic fields using different The collisional physics in ASPEN were
step length algorithms in the ASPEN simulation. validated by running ASPEN in a

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5659

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

collision-only mode and comparing

our predictions with reported results
from two different models of light
ion energy deposition in the Martian
atmosphere: the stopping and range
of ions in matter model (hereafter
called “SRIM”) [Leblanc et al., 2002]
and a model of fast neutral hydrogen
atoms (hereafter called “K&B”) [Kallio
and Barabash, 2001].
Predictions from SRIM and K&B were
previously compared in Leblanc et al.
[2002] using the test case of incident
800 eV neutral hydrogen atoms. SRIM
that predicted 16% of the energy was
deposited into neutral heating, com-
pared to 14% in K&B. ASPEN predicts
Figure 4. Comparison of neutral heating rates predicted by ASPEN and the that 13% of energy is deposited into
SRIM results presented in Leblanc et al. [2002]. heating, in close agreement with the
prior two results. ASPEN also found
that 32% of the atoms were backscattered, which settles between SRIM’s prediction of 10% and K&B’s esti-
mate of 58%.
We exclusively compare ASPEN and K&B predictions to validate energy deposition by ionization and excita-
tion because SRIM does not distinguish inelastic processes. ASPEN predicts 26% of energy is deposited into
ionization, agreeing with the 27% estimated from K&B. However, the energy deposited into excitation pro-
cesses differs, as ASPEN predicts only 6% excitation unlike 30% in Kallio and Barabash [2001]. This is most
likely because we approximated many different excitation mechanisms as a single process due to the lack
of available excitation cross sections. Because we are focusing on ionization in this work and our results
match the ionization results from previous work, this should not significantly impact the results in this paper.

3.3. SEP Event Comparison

ASPEN predicts ionization and heating rates during a SEP event. We can compare our rate calculation to the
SEP proton heating rate presented in Leblanc et al. [2002]. The heating rate in Leblanc et al. [2002] was calcu-
lated by multiplying atmospheric densities in a thick slab by both SEP proton fluxes entering the exobase and
energy deposited by protons in each slab. Using an atomic-equivalent atmosphere model, they approxi-
mated all atmospheric atoms as monatomic oxygen and determined the equivalent atmospheric density.
Proton energy deposition was then determined using SRIM predictions of proton energy deposition in a
column of monatomic oxygen.
Despite how ASPEN’s operating range is limited to below 5 MeV due to lacking relativistic corrections and
nuclear processes, extrapolating our nonrelativistic cross sections to 100 MeV enables ASPEN to predict a rate
that can be compared to Leblanc et al. [2002]. Using the same SEP proton fluxes at 14 energies between 1 keV
and 1 MeV as used in Leblanc et al. [2002], we calculate and compare heating rates using MTGCM dayside
densities described in section 2.2.1 (plotted in green) and the same atomic-equivalent atmosphere densities
used in Leblanc et al. [2002] (plotted in blue) (Figure 4). We also show heating rates using atomic-equivalent
atmosphere densities integrated over the operating range of ASPEN (plotted in red).
ASPEN predicts up to 2 orders of magnitude more heating below 115 km and as low as 2 orders of magnitude
less heating above 115 km. The missing relativistic processes cannot account for this, as we see the same
trend for heating rates calculated using ASPEN operating range energies above 85 km. The effect is caused
by particles in ASPEN depositing energy at lower altitudes than SRIM does. While Leblanc et al. [2002] predict
a heating peak at 120 km caused by particles with energy between 0.5 and 10 keV, neither ASPEN prediction
shows a peak. At 120 km, ASPEN predicts 20–500 keV proton fluxes deposit as much energy as 10 keV proton
fluxes. These discrepancies can be owed to differences in cross sections, as our result is consistent with our

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5660

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

cross-section database. For example,

using the ratio of the elastic cross sec-
tion to the total cross section, the
approximate probability of a 5 MeV
proton elastically colliding with O is
4%. Since 5 MeV particles can have
over 20,000 collisions before therma-
lizing, up to 1000 collisions can be
elastic and heat the atmosphere.
That is roughly the same number of
elastic collisions that a 2 keV proton
would have, as 2 keV protons can
have 1500 collisions and a 70%
chance of elastically scattering.
However, the energy imparted by an
elastic collision is proportional to the
particle’s incident energy. Thus, pro-
tons should impart energy to heating
in amounts proportional to their
Figure 5. Comparison of stopping altitudes between a stopping power
energy, causing higher protons to
approximation (plotted in red) and the ASPEN simulation using dayside
(plotted in black) and nightside (plotted in blue) atmospheric densities. heat the atmosphere more than
ASPEN dayside stopping altitudes are also plotted for the Rudd secondary low-energy protons and preventing
(circles) and fixed secondary models (crosses). the existence of a visible heating
peak between 80 and 170 km.
We can also compare the fraction of energy imparted by heating to all altitudes between 80 and 170 km to
the total energy imparted by the SEPs with energies between 1 keV and 100 MeV. ASPEN predicts 22% of the
SEP energy flux is imparted to heating using MTGCM densities. Using the atomic-equivalent atmosphere
shifts this heating efficiency to 31%, which is consistent with the heating efficiency of atomic oxygen.
However, SRIM predicts only 2% of the input SEP energy is imparted to heating in this altitude range despite
how SRIM predicts a heating efficiency of 16%. Because the energy cannot be imparted to other collisional
processes, SRIM either underestimates heating above 80 km or predicts more heating energy is imparted
below 80 km.

4. Proton Transport and Escape

Before examining the effects of a specific SEP event, we study proton transport and energy deposition as a
function of proton energy. In the first subsection, we predict proton thermalization altitudes in the absence
of magnetic fields and compare to previous model predictions. Next we investigate backscattering in a sam-
ple case of 10 keV protons. We then quantify rates of backscattering and thermalization for protons with
energies between 2 keV and 5 MeV. Finally, we show energy deposition as a function of proton energy.
In the next two sections, we will compare ASPEN results from two distinct simulation runs. In both runs, a
thousand protons with energies between 2 keV and 5 MeV are released from an altitude of 500 km travel-
ing directly toward the dayside atmosphere of Mars during solar cycle minimum. In the “crustal field anom-
aly” run, particles are released over a crustal anomaly located at 50°S and 182°E. In the “absent magnetic
fields” run, particles are released from the same location while magnetic fields are set to 0 nT.
Secondary electrons produced by ionization have energies that are predicted using differential cross
sections from Rudd [1988].

4.1. Thermalization Altitudes

Particles deposit more energy at lower altitudes because of increasing atmospheric densities. The maximum
energy deposited is limited by the projectile’s energy, causing protons with specific energies to be therma-
lized or “stopped” at a specific altitude. Prior studies have predicted stopping altitudes using the continuous
slowing-down (CSD) approximation [Henriksen, 1978; Sheel et al., 2012]. While ASPEN is a Monte Carlo model

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5661

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

that uses a random variable to calculate distances between individual collisions, the CSD approximation
assumes particles with the same energy travel the same distance before thermalizing.
In Figure 5, we compare predictions of proton stopping altitude in the absence of magnetic fields. Stopping
altitudes from ASPEN were taken by averaging the altitudes of protons with energies that fell below our 10 eV
energy cutoff and are assumed to have thermalized in place. The CSD results used stopping ranges in carbon
dioxide from proton stopping range tables provided by the National Institute of Standards and Technology
[Berger et al., 2005]. We solve for stopping altitudes by taking the atmosphere in hydrostatic equilibrium and
balancing the gravitational force with atmospheric collisions, or P(z) = P0e
z/H = R(E)g for stopping range R(E),
gravitational constant g, scale height H, and surface pressure P0. As a first-order approximation, we used an
isothermal carbon dioxide atmosphere with a scale height of 11.1 km.
In order to separate the influence of using different densities from the differences between the models, we
gauge ASPEN’s dependence on different neutral density profiles by determining stopping altitudes for both
dayside solar minimum and nightside solar moderate MTGCM neutral density profiles. Higher stopping
altitudes in a dayside atmosphere are expected, given higher atmospheric densities in the upper dayside
atmosphere. We found that ASPEN stopping altitudes were 2–7 km higher in a dayside atmosphere than
in a nightside atmosphere.
CSD-predicted stopping altitudes of protons with energies less than 2 MeV are up to 30 higher than ASPEN-
predicted stopping altitudes. The difference is greatest for protons with energies below 300 keV. However,
CSD also predicts lower stopping altitudes for protons with energy above 2 MeV. The CSD approximation pre-
dicts that 2 keV protons stop at 152 km and ASPEN predicts they stop at ~122 km, while 5 MeV protons stop
at 67 km using the CSD approximation and at 78 km using ASPEN. These differences are greater than the
variability observed between different atmospheres, meaning they could be caused by variation in cross sec-
tions or different approaches. The stopping power data used in the CSD approximation average over back-
scattered protons, which can decrease stopping range. In contrast, ASPEN determines stopping altitude by
only averaging altitudes of thermalized particles. The difference between ASPEN and CSD results decreases
for higher energies because the CSD approximation predicts less backscattering at high energies than
ASPEN does.

4.2. Backscattering: Case Study

Not all protons that precipitate into an atmosphere are absorbed. Backscattering can attenuate and distribute
the energy-depositing particle population. There are two causes of proton backscatter: atmospheric collisions
and magnetic reflection. Atmospheric collisions stochastically turn particles back upward and away from the
lower atmosphere. They are the only source of backscatter in the absence of magnetic fields. Magnetic reflec-
tion turns away protons in the upper atmosphere before they can deposit any energy. Atmospheric reflection
and magnetic reflection can also act in tandem, allowing protons to deposit some energy in the atmosphere
before being magnetically reflected.
Backscatter mechanisms are observable in traces of protons depositing energy (Figure 6). Figure 6a shows
the paths of 10 identical 10 keV protons depositing energy in the absence of magnetic fields. Differences
between the paths are caused by stochastic scattering events predicted by ASPEN. Six protons thermalize,
losing all energy in the lower atmosphere. Four protons atmospherically scatter, losing substantial energy
before reflecting with 10–50% of their initial energy. Atmospheric scattering is characterized by particles
escaping with less than half of their initial energy.
Figure 6b shows the paths of ten 10 keV protons depositing energy into the atmosphere at a crustal anomaly
located at 177°E and 7°S. Six protons magnetically reflect, taking the Lorentz path (the path of the same pro-
ton moving in without atmosphere, plotted in black) and escaping with 100% of their original energy. One
proton atmospherically scatters exactly as in the previous case, escaping with less than 50% of its initial
energy. Three protons have a different escape mechanism that is a combination of atmospheric scattering
and magnetic reflection. Each scatters from 265 km, the altitude of closest approach in the Lorentz path
(hereafter called the “magnetic reflection altitude”), and escapes with over 50% of their initial energy.
Increased scattering at the magnetic reflection altitude is caused by particles moving parallel to the surface
at this altitude, encountering the same neutral density for tens of kilometers which rapidly increasing the col-
lision likelihood. The traces show that this mechanism distributes energy deposition over a greater area. Two

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5662

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

Figure 6. Traces of eleven 10 keV protons simulated in ASPEN entering at 177°E and 7°S injected directly (α = 0°) in a
(a) purely collisional atmosphere with no magnetic fields and a (b) collisional atmosphere with magnetic and electric fields
from MF-MHD. A noncollisional Lorentz path is plotted in black in both plots. Contours of the total magnetic field in
the right figure are plotted in gray between 0 and 150 nT. Since the 10 traces are very similar, only ones that significantly
scatter can be distinguished.

of the three traces show protons that become neutral hydrogen near the magnetic reflection altitude,
causing them to travel in straight lines from the magnetic reflection altitude down to 240–250 km over
5–10°E from the launch location. These particles deposited energy along these paths before they were
charge stripped, becoming again susceptible to magnetic fields, and then escaped.

4.3. Energy Dependence of Backscatter and Thermalization

Protons with energies less than 5 MeV either thermalize or backscatter. Atmospheric backscattering requires
the occurrence of a nonforward scattering event, which is favored at different proton energies. Likewise, the
likelihood of magnetic reflection depends on proton energy. Protons with lower energy are more susceptible
to deflection by magnetic fields, meaning low-energy protons reflect higher in the atmosphere than high-
energy protons. Magnetic reflection is also dependent on the magnetic topology of the region, but that is
beyond the scope of this paper.
The energy dependence of backscattering (magnetically and atmospherically) is shown for absent magnetic
fields and the studied crustal anomaly in Figure 7. In the absent magnetic fields case (Figure 7a), all protons
deposit energy. Up to 75% of protons with energies below 80 keV deposit their entire energy in the atmo-
sphere. However, between 100 keV and 1 MeV, the chance of thermalization drops to 50%. Protons with ener-
gies over 1.6 MeV are more likely to atmospherically backscatter.
The greater likelihood of higher-energy protons backscattering may seem counterintuitive. Due to the lack of
angular cross-section data for inelastic processes, a key model assumption is that particles only change tra-
jectory in elastic collisions. Elastic collisions only occur frequently for protons with energies below 500 eV.
However, particles with energies over 1 MeV have over 10,000 collisions before thermalizing. Even a 5%
chance of an elastic collision would cause over 500 elastic collisions, raising the likelihood of backscattering.
In the crustal field anomaly case (Figure 7b), protons with energy less than 400 keV are either magnetically
reflected or backscattered. Over 95% of protons with energies below 10 keV will magnetically reflect without
depositing any energy. The dominant proton backscattering process for energies below 119 keV is magnetic
reflection without energy deposition. The dominant process switches to atmospheric scattering for protons
with energies from 119 keV to up to 5 MeV. There are more backscattered 120–800 keV protons nearby a crus-
tal anomaly than in the absence of magnetic fields. Most protons with these energies impart energy to the
atmosphere before being magnetically reflected, like the three traces presented in section 5.2. These protons
are counted as “scatter reflected” instead of “magnetically reflected” in Figure 7b because they can impart
energy before reflecting. Protons with energies above 1 MeV are not as likely to atmospheric scatter as pro-
tons with energies between 100 keV and 1 MeV. Protons with energies greater than 300 keV have an over 5%
chance of thermalizing, which increases to 50% for protons with energies between 1 MeV and 1.2 MeV.

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5663

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

Figure 7. Proton escape as a function of energy for protons traversing through a collisional atmosphere (a) with no
magnetic fields and (b) with magnetic and electric fields from MF-MHD. Monte Carlo error bars (1/√N) are shown for
each energy where N is the number of simulated particles per energy (1000 in Figure 7a and 100 in Figure 7b).

Similar to how high-energy (>1 MeV) protons are mostly reflected in the absent fields case, the probability of
thermalization plummets for protons with energies above 2 MeV.

4.4. Energy Dependence of Altitude-Varying Energy Deposition

The previous sections have considered energy deposition by thermalization and backscattering mechanisms
in isolation. In this subsection, total energy deposition is aggregated regardless of mechanism, both overall
and as a function of altitude, to study energy absorption as a function of proton energy and magnetic fields.
In the absence of magnetic fields, protons with energies below 454 keV deposit 80–91% of their energy in the
atmosphere. Energy deposition by protons with energies above 454 keV steadily declines, as 963 keV protons
deposit 70% and 5 MeV protons deposit only 51% of their energy. The lowered energy deposition coincides
with increased elastic scattering.
Nearby a crustal field anomaly, protons with energies below 454 keV only deposit as much as 1% of their
energy. Opposite to the absent fields case, energy deposited by protons with energies above 454 keV drama-
tically increases with primary energy. The 1.5 MeV protons deposit approximately 60% of their energy.
Percent energy deposition slowly decreases for protons with energies above 2.3 MeV, with 5 MeV protons
depositing 51% of their energy. In net, protons with energies below 1 MeV deposit nearly no energy in the
vicinity of a crustal anomaly, while protons with energies above 1 MeV deposit as much energy near a crustal
anomaly as they would in the absence of magnetic fields.
Regardless of the presence of magnetic fields, the net energy deposition aggregated by altitude between 70
and 300 km is inversely proportional to the altitude, smoothly increasing as altitude decreases (Figure 8).
Higher-energy protons deposit more energy at lower altitudes.
Protons deposit 99.94% of their energy below 150 km in the absence of fields (Figure 8a). The average max-
imum deposited energy above 150 km is 1.7 eV/km. Occasionally, energies over 50 eV/km are deposited
above 150 km, seen as speckles in Figure 8a. These seemingly anomalous large energy depositions at higher
altitudes are caused by elastic collisions. Although elastic collisions are infrequent for particles with energies
greater than 500 eV, the energy lost in a single elastic collision is proportional to the proton’s initial energy at
the time of scatter.
The energy deposition distribution by protons near a crustal field anomaly shows overall much less energy
deposition compared to the absent fields case (Figure 8b). The most notable feature of Figure 8b is that pro-
tons with energies between 50 keV and 800 keV deposit energy in two regions: in a narrow altitude range
around the proton stopping altitudes in the absence of magnetic fields (plotted in red) and at the magnetic
reflection altitude (plotted in black). More energy is deposited at the magnetic reflection altitude than

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5664

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

Figure 8. Energy deposited per primary by a protons with energies between 2 keV and 5 MeV injected in a (a) purely
collisional atmosphere with no magnetic fields and a (b) collisional atmosphere with magnetic and electric fields from
MF-MHD. The altitude of magnetic reflection is plotted in black, and the stopping altitude in a purely collisional atmosphere
is plotted in red.

between the stopping altitude and magnetic reflection altitude. This is because the particle is traveling
laterally, so the path length in that altitude bin is higher, increasing the energy deposited. When the
altitude of magnetic reflection falls below 110 km (for 700 keV), the energy deposition profile is similar to
the absent fields case except with a slightly higher altitude of maximum energy deposition. For example,
1.5 MeV protons deposit the maximum amount of energy at 87 km near a crustal field anomaly compared
to 85 km in the absence of magnetic fields.
Energy deposition in the lower atmosphere by protons with low energies (<500 keV) is much less in crustal
fields than in the absence of magnetic fields. These protons must be both elastically scattered and charge
stripped at the magnetic reflection altitude in order to deposit energy below 150 km. Protons with energies
below 100 keV deposit ~1 eV/km below 150 km, compared to 190 eV/km in the absence of magnetic fields.

5. SEP Event Predictions

Using the data sets generated in the previous section, we can determine rates of ionization near a crustal
anomaly and in the absence of magnetic fields. For the flux F(E) and collision rate R(E, a) at altitude a for
SEP protons with energy E, the process rate during a SEP event is

∫
Ef
IðaÞ ¼ Ei
F ðE ÞRðE; aÞdE

Following the method for determining rates in Leblanc et al. [2002] by using 14 energies between 1 keV and
100 MeV, we use 1 keV as our lower energy bound. The 5 MeV is our upper energy bound, as 5 MeV is the
highest-energy ASPEN can simulate. Our results are relatively accurate from 300 km to 70 km. We use fluxes
from the peak SEP spectrum measured on 29 October 2003 5:59 (Figure 9) by the Electrons, Protons, and
Alpha Monitor (EPAM) on the ACE spacecraft, located at 1 AU [Chiu et al., 1998]. This event, colloquially known
as the “Halloween storm,” has been previously used as a proxy for predicting SEP effects at Mars [Lillis
et al., 2012].

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5665

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

5.1. Crustal Field Influences

Predicted ionization rates near a
crustal field anomaly and in the
absence of magnetic fields during
the EPAM event are shown in
Figure 10. In the absence of
magnetic fields (black line), the
peak ionization above 70 km is
14,853 cm
3 s
1 at 83 km. There
is a slight increase in ionization
between 105 and 132 km that
peaks at ~2000 cm
3 s
1 at
115 km, which is caused by the
slightly higher fluxes of 100–
300 keV particles measured by
EPAM. Total ionization nearby a
Figure 9. Spectrum of peak spectrum during the 28 October 2003 SEP event crustal field anomaly (red line) is
as measured by Solar Wind Electron, Proton, and Alpha Monitor. 68.3% of ionization in the absence
of magnetic fields. Such a decrease
is expected, as crustal fields attenuate SEP penetration below 500 keV. Peak ionization is 12,552 cm
3 s
1 at
83 km, which is an ~15% reduction compared to the absent fields case. There is no comparable ionization
bump at 115 km nearby the crustal anomaly.
We determined volume rates of ionization in the vicinity of proton injection location (Figure 11). In the
absence of magnetic fields, the maximum ionization rate is 697 cm
3 s
1 per °N per °E, located at the same
geographic location protons were injected at (marked with a star, Figure 11a). Ionization is overall radially
symmetric across the region, spreading in 5° in each direction.
Ionization volume rates near a crustal anomaly are not symmetric at all (Figure 11b). Volume rates are plotted
below contours representing horizontal magnetic fields, colored according to the radial component (blue is
below
40 nT and red is above 40 nT). The maximum ionization rate is 481 cm
3 s
1 per °N per °E at 50°S 180°
E, approximately 2°E from where protons were injected. Ionization over 50 cm
3 s
1 per °N per °E spreads
northwest of the injection location and ionization below 50 cm
3 s
1 per °N per °E largely occurs southeast.
The northwest ionization follows
the Lorentz trajectory of
energetic particles.

5.2. Secondary Electron

Production
ASPEN logs the energy of any pro-
duced secondaries from ionization
and charge stripping. Secondary
energies are aggregated to predict
the spectrum of secondary elec-
trons at each altitude. Similar to
the calculation of ionization rate,
each spectrum per altitude is
integrated over the EPAM fluxes
to predict an altitude profile of
secondary electron production
during an event.

Figure 10. Altitude profiles of ionization rate during the 28 October 2003 As described in section 2.2.3, there
SEP event using a population of protons injected at 50°S and 182°E on the are two approaches to predicting
dayside over nonmagnetized crust (black) and a crustal anomaly (red). secondary electron energy in a

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5666

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

Figure 11. Map of SEP-induced ionization volume rate for SEPs directed at the anomaly located at 50°S and 182°E if
magnetic fields in the simulation are toggled (a) off or (b) on. The star marks where protons are injected. The contours
mark horizontal magnetic fields at an altitude of 400 km where the color indicates the radial component of the magnetic
field where red is +40 nT and blue is
40 nT.

collision: using a random secondary energy from a double differential cross section or using a fixed secondary
energy from extrapolating an existing data set. Both approaches produce near-identical ionization rates
down to an altitude of 95 km. Below an altitude of 95 km, the fixed secondary energy approach predicts
28% less ionization than the Rudd secondary energy. The discrepancy is because the fixed secondary
energy extrapolation predicts substantially higher secondary energies for >1 MeV protons than the Rudd
secondary energy method, causing protons to lose energy quicker and ionize less frequently. Since the
Rudd secondary energy approach is based on an empirically verified double differential cross-section
expression, we consider it more realistic than the fixed secondary energy approach. We will continue to
use this approach in the future.
Altitude profiles of secondary electron spectra were determined in the absence of magnetic fields using fixed
(Figure 12a) and random secondary energy approaches (Figure 12b). The fixed secondary energy model

Figure 12. Spectrum of electrons produced by SEP proton-impact ionization over unmagnetized crust using a (a) fixed
secondary electron energy model and a (b) random secondary electron energy model.

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5667

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

predicts that most secondaries have energies between 10 eV and 200 eV. The random secondary energy
model predicts production of many more secondary electrons with energies distributed between 0.01 eV
and 100 eV. The models have different implication for SEP-produced upward electron fluxes. The production
rate of 110 eV electrons at 150 km is 13 el cm
3 s
1 keV
1 in the fixed energy model and 4 el cm
3 s
1 keV
1
in the random energy model.

6. Conclusion
We have validated and demonstrated the application of a novel proton transport and collisional physics
model. The ASPEN model predicts atmospheric effects by SEP precipitation by including ionization and
heating. It uses an adaptive step length algorithm, allowing accurate predictions while running 20 times
faster than if we were to use the shortest step length. We compared heating rates from the ASPEN model
to published results from previous models, finding that ASPEN predicted significantly greater heating than
the model presented in Leblanc et al. [2002] and that protons deposited their energy up to 20 km lower in
altitude than for models using a conventional continuous slowing-down approximation.
Unlike many other simulations of SEP precipitation into planetary atmospheres, ASPEN accounts for charged
particle motion in spatially varying magnetic fields. We used ASPEN to study the effects of a specific anomaly
located at 50°S and 182°E on atmospheric ionization. Early results suggest that SEP protons deposit a substan-
tial amount of energy in the atmosphere above the crustal field anomalies. However, proximity to crustal
fields can reduce overall atmospheric ionization by ~30%. Atmospheric ionization near a crustal anomaly
spreads over a wider geographic area than atmospheric ionization in the absence of magnetic fields. The
distribution of atmospheric ionization occurs along an axis corresponding to the Lorentz trajectories of
SEP protons.
We also compared ASPEN secondary electron production predictions in negligible magnetic fields using two
different methodologies: randomly selecting a secondary energy using the double differential cross section
or extrapolating previously determined secondary energies to calculate a fixed secondary energy. Overall
ionization predicted by both methods is similar above an altitude of 95 km. We determined production rates
of superthermal electrons (~110 eV) from proton-impact ionization at the limiting altitude where electrons
preferentially thermalize instead of backscattering (150 km) [Lillis et al., 2012]. Electron production rates pre-
dicted using the random secondary energy model are half that from using the fixed secondary energy model.
These electron populations can be modeled using electron transport codes to see if they explain the
increased upward electron fluxes observed by MGS.
We demonstrated that SEP protons cause substantially less ionization if they are incident on a specific loca-
tion in close proximity to a crustal field anomaly. Future work will include an investigation of how crustal
anomalies influence ionization from SEP protons precipitating globally. Precipitating low-energy SEPs
(~10 keV) can be magnetically reflected from between 18 and 50% of the atmospheric area presented by
the dayside southern hemisphere if the crustal fields are on the dayside. We intend to determine and incor-
porate the SEP flux dependence on solar zenith angle in order to study SEP ionization over crustal anomaly
regions, which eventually will be expanded into global maps. The progression of atmospheric ionization dur-
ing a SEP event can also be tracked using data from the SEP instrument on MAVEN, allowing for estimates of
time dependent and total ionization and heating during an event.
Acknowledgments
This research was supported by the Finally, the ASPEN model has broad applications for future use and could be a community-modeling tool. For
MAVEN project (supported by NASA example, the radiation environment at Mars is extremely harsh. The MARIE (MArtian RadIation Experiment)
through the Mars Exploration Program), instrument on board Mars Odyssey was intended to characterize the environment but malfunctioned follow-
NASA PATM grant NNX11AD81G and
NASA MFRP grant NNX13AG52G. C.F. ing a strong SEP event [Luhmann et al., 2007]. The ASPEN open-source modules provide the opportunity for
Dong is currently supported by the independent research development for near-Mars SEP flux calculations, auroral predictions, and surface
NASA Living with a Star Jack Eddy radiation modeling.
Postdoctoral Fellowship Program,
administered by the University
Corporation for Atmospheric Research.
The source code and input files neces-
References
sary to reproduce the ASPEN model Acuña, M. H., et al. (2001), The magnetic field of Mars: Summary results from the aerobraking and mapping orbits, J. Geophys. Res., 106,
results analyzed in this work are avail- 23,403–23,417, doi:10.1029/2000JE001404.
able from the authors upon request Basu, B., J. R. Jasperse, R. M. Robinson, R. R. Vondrak, and D. S. Evans (1987), Linear transport theory of auroral proton precipitation: A
([email protected]). comparison with observations, J. Geophys. Res., 92, 5920–5932, doi:10.1029/JA092iA06p05920.

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5668

 Journal of Geophysical Research: Space Physics 10.1002/2016JA023781

Basu, B., J. R. Jasperse, D. J. Strickland, and R. E. Daniell Jr. (1993), Transport-theoretic model for the electron-proton-hydrogen atom aurora.
1: Theory, J. Geophys. Res., 98, 21,517–21,532, doi:10.1029/93JA01646.
Berger, M. J., J. S. Coursey, M. A. Zucker, and J. Chang (2005), ESTAR, PSTAR, and ASTAR: Computer Programs for Calculating Stopping-Power
and Range Tables for Electrons, Protons, and Helium Ions (version 1.2.3), Natl. Instit. of Stand. and Technol., Gaithersburg, Md. [Available at
http://physics.nist.gov/star.]
Bougher, S. W., S. Engel, R. G. Roble, and B. Foster (2000), Comparative terrestrial planet thermospheres 3. Solar cycle variation of global
structure and winds at solstices, J. Geophys. Res., 105, 17,669–17,692, doi:10.1029/1999JE001232.
Bougher, S. W., J. M. Bell, J. R. Murphy, M. A. Lopez-Valverde, and P. G. Withers (2006), Polar warming in the Mars thermosphere: Seasonal
variations owing to changing insolation and dust distributions, Geophys. Res. Lett., 33, L02203, doi:10.1029/2005GL024059.
Chiu, M. C., et al. (1998), ACE spacecraft, Space Sci. Rev., 86, 257.
Damiani, A., M. Storini, M. L. Santee, and S. Wang (2010), The hydroxyl radical as an indicator of SEP fluxes in the high-latitude terrestrial
atmosphere, Adv. Space Res., 46, 1225–1235, doi:10.1016/j.asr.2010.06.022.
Delory, G. T., J. G. Luhmann, D. Brain, R. J. Lillis, D. L. Mitchell, R. A. Mewaldt, and T. V. Falkenberg (2012), Energetic particles detected by the
Electron Reflectometer instrument on the Mars Global Surveyor, 1999–2006, Space Weather, 10, S06003, doi:10.1029/2012SW000781.
Desorgher, L., E. O. Flückiger, M. Gurtner, M. R. Moser, and R. Bütikofer (2005), ATMOCOSMICS: A GEANT4 code for computing the interaction
of cosmic rays with the Earth’s atmosphere, Int. J. Mod. Phys. A, 20, 6802–6804.
Dong, C. F., S. W. Bougher, Y. J. Ma, G. Toth, A. F. Nagy, and D. Najib (2014), Solar wind interaction with Mars upper atmosphere: Results from
the one-way coupling between the multi-fluid MHD model and the M-TGCM model, Geophys. Res. Lett., 41, 2708–2715, doi:10.1002/
2014GL059515.
Espley, J. R., W. M. Farrell, D. A. Brain, D. D. Morgan, B. Cantor, J. J. Plaut, M. H. Acuña, and G. Picardi (2007), Absorption of MARSIS radar signals:
Solar energetic particles and the daytime ionosphere, Geophys. Res. Lett., 34, L09101, doi:10.1029/2006GL028829.
Futaana, Y., et al. (2008), Mars Express and Venus Express multi-point observation’s of geoeffective solar flare events in December 2006,
Planet. Space Sci., 56(6), 873–880, doi:10.1029/2012SW000781.
Gronoff, G., C. Mertens, J. Lilensten, L. Desorgher, E. Flückiger, and P. Velinov (2011), Ionization processes in the atmosphere of Titan. III.
Ionization by high-Z nuclei cosmic rays, Astron. Astrophys., 529, A143, doi:10.1051/0004-6361/201015675.
Gronoff, G., R. B. Norman, and C. J. Mertens (2015), Computation of cosmic ray ionization and dose at Mars. I: A comparison of HZETRN and
Planetocosmics for proton and alpha particles, Adv. Space Res., 55, 1799–1805.
Henriksen, K. (1978), Variations of proton energy and pitch-angle spectra in the upper atmosphere, J. Atmos. Terr. Phys., 41, 633.
Kallio, E., and S. Barabash (2001), Atmospheric effects of precipitating energetic hydrogen atoms on the Martian atmosphere, J. Geophys. Res.,
106, 165–178, doi:10.1029/2000JA002003.
Kokorowski, M., et al. (2006), Rapid fluctuations of stratospheric electric field following a solar energetic particle event, Geophys. Res. Lett., 33,
L20105, doi:10.1029/2006GL027718.
Leblanc, F., J. G. Luhmann, R. E. Johnson, and E. Chassefiere (2002), Some expected impacts of a solar energetic particle event at Mars,
J. Geophys. Res., 107(A5), 1058, doi:10.1029/2001JA900178.
Lee, C. O., et al. (2017), MAVEN observations of the solar cycle 24 space weather conditions at Mars, J. Geophys. Res. Space Physics, 122,
2768–2794, doi:10.1002/2016JA023495.
Lillis, R. J., D. L. Mitchell, R. P. Lin, and M. H. Acuña (2008), Electron reflectometry in the Martian atmosphere, Icarus, 194, 544–561, doi:10.1016/
j.icarus.2007.09.030.
Lillis, R. J., D. A. Brain, G. T. Delory, D. L. Mitchell, J. G. Luhmann, and R. P. Lin (2012), Evidence for superthermal secondary electrons produced
by SEP ionization in the Martian atmosphere, J. Geophys. Res., 117, E03004, doi:10.1029/2011JE003932.
Lillis, R. J., C. O. Lee, D. Larson, J. G. Luhmann, J. S. Halekas, J. E. Connerney, and B. M. Jakosky (2016), Shadowing and anisotropy of solar
energetic ions at Mars measured by MAVEN during the March 2015 solar storm, J. Geophys. Res. Space Physics, 121, 2818–2829,
doi:10.1002/2015JA022327.
Luhmann, J. G., C. Zeitlin, R. Turner, D. A. Brain, G. Delory, J. G. Lyon, and W. Boynton (2007), Solar energetic particles in near-Mars space,
J. Geophys. Res., 112, E10001, doi:10.1029/2006JE002886.
McKenna-Lawlor, S., V. V. Afonin, E. Kirsch, K. Schwingenschuh, J. A. Slavin, and J. G. Trotignon (1998), An overview of energetic particles (from
55 keV to >30 MeV) recorded in the close Martian environment and their energization in local and external processes, Planet. Space Sci.,
46, 83–102.
Mironova, I. A. and I. G. Usoskin (2014), Possible effect of strong solar energetic particle events on polar stratospheric aerosol: A summary of
observational results, Environ. Res. Lett., 9(1), 015002, doi:10.1088/1748-9326/9/1/015002.
Mironova, I. A., K. L. Aplin, F. Arnold, G. A. Bazilevskaya, R. G. Harrison, A. A. Krivolutsky, K. A. Nicoll, E. V. Rozanov, E. Turunen, and I. G. Usoskin
(2015), Energetic particle influence on the Earth’s atmosphere, Space Sci. Rev., 194, 1, doi:10.1007/s11214-015-0185-4.
Moore, T. E., W. K. Peterson, C. T. Russell, M. O. Chandler, M. R. Collier, H. L. Collin, P. D. Craven, R. Fitzenreiter, B. L. Giles, and C. J. Pollock (1999),
Ionospheric mass ejection in response to a coronal mass ejection, Geophys. Res. Lett., 26, 2339–2342, doi:10.1029/1999GL900456.
Morgan, D. D., D. A. Gurnett, D. L. Kirchner, J. L. Fox, E. Nielsen, and J. J. Plaut (2008), Variation of the martian ionospheric electron density from
Mars Express radar soundings, J. Geophys. Res., 113, A09303, doi:10.1029/2008JA013313.
Norman, R. B., T. C. Slaba, and S. R. Blattnig (2013), An extension of HZETRN for cosmic ray initiated electromagnetic cascades, Adv. Space Res.,
51, 2251–2260.
Noël, S., and G. W. Prölss (1993), Heating and radiation production by neutralized ring current particles, J. Geophys. Res., 98, 17,317–17,325,
doi:10.1029/93JA01500.
Rudd, M. E. (1988), Differential cross sections for secondary electron production by proton impact, Phys. Rev. A., 38(12), 6129–6137.
Schneider, N. M., et al. (2015), Discovery of diffuse aurora on Mars, Science, 350, doi:10.1126/science.aad0313.
Sheel, V., S. A. Haider, P. Withers, K. Kozarev, I. Jun, S. Kang, G. Gronoff, and C. Simon Wedlund (2012), Numerical simulation of the effects of a
solar energetic particle event on the ionosphere of Mars, J. Geophys. Res., 117, A05312, doi:10.1029/2011JA017455.
Usoskin, I. G., L. Desorgher, P. Velinov, M. Storini, E. O. Flückiger, R. Bütikofer, and G. A. Kovaltsov (2009), Ionization of the Earth’s atmosphere
by solar and galactic cosmic rays, Acta Geophys., 57, 88–101, doi:10.2478/s11600-008-0019-9.
Ulusen, D., D. A. Brain, J. G. Luhmann, and D. L. Mitchell (2012), Investigation of Mars’ ionospheric response to solar energetic particle events,
J. Geophys. Res., 117, A12306, doi:10.1029/2012JA017671.

JOLITZ ET AL. SEP PRECIPITATION INTO CRUSTAL FIELDS 5669

View publication stats