Show Summary Details

Page of

date: 29 October 2020

# Martian Ionospheric Observation and Modelling

• Francisco González-GalindoFrancisco González-GalindoInstituto de Astrofísica de Andalucía - CSIC

### Summary

The Martian ionosphere is a plasma embedded within the neutral upper atmosphere of the planet. Its main source is the ionization of the CO2-dominated Martian mesosphere and thermosphere by energetic EUV solar radiation. The ionosphere of Mars is subject to an important variability induced by changes in its forcing mechanisms (e.g., the UV solar flux) and by variations in the neutral atmosphere (e.g., the presence of global dust storms, atmospheric waves and tides, changes in atmospheric composition, etc.). Its vertical structure is dominated by a maximum in electron concentration at altitude about 120–140 km, coincident with the peak of the ionization rate. Below, there is a secondary peak produced by solar X-rays and photoelectron-impact ionization. A sporadic third layer, possibly of meteoric origin, has been also detected below.

The most abundant ion in the Martian ionosphere is O2+, although O+ can become more abundant in the upper ionospheric layers. While below about 180–200 km the Martian ionosphere is dominated by photochemical processes, above those altitudes the dynamics of the plasma becomes more important. The ionosphere is also an important source of escaping particles via processes such as dissociative recombination of ions or ion pickup. So, characterization of the ionosphere provides or can provide information about such disparate systems and processes as solar radiation reaching the planet, the neutral atmosphere, meteoric influx, atmospheric escape to space, or the interaction of the planet with the solar wind.

It is thus not surprising that the interest about this region dates from the beginning of the space era. From the first measurements provided by the Mariner 4 mission in the 1960s to observations by the Mars Express and MAVEN orbiters in the 2010s, our knowledge of this atmospheric region is the consequence of the accumulation of more than 50 years of discontinuous measurements by different space missions. Numerical simulations by computational models able to simulate the processes that shape the ionosphere have also been commonly employed to obtain information about this region, to provide an interpretation of the observations and to fill their gaps. As a result, at the end of the 2010s the Martian ionosphere was the best known one after that of the Earth. However, there are still areas for which our knowledge is far from being complete. Examples are the details and balance of the mechanisms populating the nightside ionosphere, the origin and variability of the lower ionospheric peak, and the precise mechanisms shaping the topside ionosphere.

### Introduction

Ionospheres are present in all Solar System bodies with an atmosphere, from the massive thick atmospheres of the giant planets to the transient, extremely thin atmospheres of comets. Their ultimate formation mechanism is the ionization of the constituents of the neutral atmosphere by interaction with energetic solar radiation or by impact with energetic particles. Ionospheres are thus strongly linked to the underlying neutral atmosphere and to the energy input to the upper atmosphere of the planet. Studying the ionospheres of terrestrial planets can provide valuable information about their neutral upper atmospheres—regions difficult to sound due to their low density. For example, analysis of the Martian ionosphere has resulted in information about the waves and tides propagating from the lower atmosphere of the planet to its upper layers (Bougher et al., 2001). Comparative studies of the ionospheres of Mars, Venus and Earth in the context of their similarities and differences enable the disentangling of the effects of different processes in different planetary environments. One example is the effect of magnetic fields on the ionosphere. While the Earth possesses a global magnetic field, Mars only presents a weak crustal magnetic field over portions of its surface, and Venus does not display any intrinsic magnetic field. Studying the differences and similarities of the solar wind interaction with the ionosphere in these planets can unveil the role of different physical mechanisms governing this interaction. On Mars, the ionosphere is one of the most important sources of particles escaping the gravitational attraction of the planet, either directly (ion escape) or indirectly (i.e. dissociative recombination of O2+ ions resulting in hot O) (Brain et al., 2017). Although other escape sources do exist, and their relative importance is still matter of debate, clearly the study of the ionosphere is of the maximum importance for the question of the long-term evolution of the Martian climate and atmosphere.

After the terrestrial ionosphere, the Martian is the best known. Observations from spacecraft orbiting Mars are the main source of our understanding of the Martian ionosphere, which has greatly evolved from the first measurements by Mariner 4 in the 60s to the Mars Express and MAVEN measurements of the 2010s. Direct information about the ionosphere can be obtained by observational techniques such as radio occultation, Langmuir probes, retarding potential analyzers, or mass spectrometers. Indirect information about the ionosphere can be obtained from analysis of UV airglow. Computational models of different kinds, from one-dimensional photochemical models to coupled models studying the thermosphere–ionosphere, the exosphere, and the magnetosphere, have been developed to help in the interpretation of observational results and to fill out gaps due to the necessarily incomplete coverage of the observational datasets.

### Main features of the Martian ionosphere

The Martian ionosphere extends between about 60 and 400 km from the surface of the planet, although these values should only be considered as broad averages and are subject to significant variation with different processes and parameters. For example, an ionosphere extending to the planetary surface produced by galactic cosmic rays ionization has been theorized (Cardnell et al., 2016; Molina-Cuberos, Lichtenegger, Schwingenschuh, López-Moreno, & Rodrigo, 2002). Electron-density profiles on the dayside of the planet are characterized by the presence of a broad peak placed at an altitude of about 130 km on average (Figure 1), produced by the ionization of the neutral atmosphere (mostly CO2) by solar photons with wavelengths between about 10 and 90 nm. This layer is similar in nature to the F1 terrestrial layer (see for example Bougher et al., 2017). A secondary layer is found about 25–30 km below the main ionospheric peak, produced by the ionization of CO2 by energetic solar radiation between 1 and 10 nm and by photoelectron impact, similar to the E terrestrial layer. The electron densities at both peaks and their altitudes have been found to vary with different factors, such as the Solar Zenith Angle (SZA hereafter) and solar activity. This variability can be to first-order understood by a simple theoretical description known as Chapman theory. This theory describes the formation, in a plane-parallel atmosphere, of an ionosphere produced by monochromatic solar radiation acting on a single atmospheric species. Under these conditions it can be shown (e.g., Withers, 2009) that the vertical profile of the electron density follows an exponential variation, the peak electron density varying with solar illumination as the square root of the cosine of the SZA, and the peak altitude changing with SZA following the altitude of the layer of unity optical depth. Sporadic layers have been observed at even lower altitudes, as well as above the main peak.

There is some controversy in the nomenclature of these layers. Some authors, by the similarity of the formation mechanisms of the primary and secondary layer with the terrestrial F1 and E layer, respectively, have adopted the same names for the Martian layers. Other authors use the nomenclature M2 and M1 for the primary and secondary layers.

Two regions can be distinguished in the Martian ionosphere: the photochemical region, below about 200 km, where plasma transport is negligible and electron and ion densities are given by the balance between production by photoionization and losses by chemical reactions; and the transport region, where plasma transport processes become dominant. The topside ionosphere is also eventually affected by interaction with the solar wind. As Mars lacks a global magnetic field, the solar wind interacts directly with the ionosphere, producing an induced magnetosphere and a series of plasma regions and boundaries (see for example Brain et al., 2017). This interaction is modified locally by the presence of a crustal magnetic field over localized regions, where “mini-magnetospheres” are formed, protecting some locations from the solar wind and allowing solar wind access to others (Brain et al., 2017).

Here, our main focus will be a description of the evolution of our understanding of the Martian ionosphere achieved through the half-century of space exploration of Mars and through the increasing sophistication of associated computational modeling. We will focus mainly on the photochemical ionosphere, but we will also describe, to a lesser extent the most relevant features produced by solar wind interaction in the transport region of the ionosphere.

### Ionospheric Modeling

Many computational models devoted to the study of the Martian ionosphere have been developed since even before the first observations of the ionosphere were performed. When combined with observations, computational models allow for a deeper insight into the processes at the origin of the observed structures. Models also complement the usually limited temporal and geographical coverage of the observations.

Models can be divided according to the physical processes they simulate and the approaches they use for it. Models can also be 1-D (considering only vertical variations), 2-D (adding the latitudinal or longitudinal variability) and global or 3-D (offering a complete coverage of the planet). While 1-D models offer more flexibility in separating the effects of individual processes, and enable a detailed treatment of the processes involved, global models consistently couple the ionosphere with the variability of the neutral atmosphere originated both in the thermosphere and in the lower atmosphere. Here we will briefly describe some of the more frequently used modeling approaches and mention some of the models used to simulate the Martian ionosphere. For a more detailed and technical description of the different types of models and the equations involved the reader is referred to recent reviews (e.g., Bougher et al., 2008, 2017; Haider, Mahajan, & Kallio, 2011; Haider & Mahajan, 2014).

Photochemical models solve the continuity equation (e.g., Schunk & Nagy, 2009) considering only photochemical productions and losses. They are appropriate for the description of the photochemical region of the ionosphere, below about 180 km. Above this altitude, a description of neutral transport (molecular diffusion, eddy diffusion, global circulation) and plasma transport (ambipolar diffusion, pressure gradients) needs to be included. The interaction of the ionosphere with the solar wind and the effects of magnetic fields may also be simulated using a different family of models, such as MagnetoHydroDynamic (MHD) models or hybrid models. (See Brain et al. (2010, 2017) for a description of the solar wind–ionosphere interaction modeling approaches, a list of recent models and a comparison of their performance.) Some of these plasma interaction models have sufficient vertical resolution to provide a very reasonable description of the photochemical ionosphere too.

One of the photochemical models more extensively used to study the Martian ionosphere is the 1-D model developed by Jean L. Fox and coworkers. Initially developed to simulate the Mariner and Viking observations (Fox & Dalgarno, 1979), the model has significantly evolved over the years. The 2015 version (Fox, 2015) computes the concentrations of 34 ion species and five neutral species (N, C, NO, H2, and H) considering about 300 chemical reactions. Background profiles of 12 major neutral species are taken from global models and observations. In addition to photochemistry, molecular and eddy diffusion, as well as ambipolar diffusion for ions, are included in the model.

Another 1-D model is the Boston University Mars Ionosphere Model. Initially developed as a purely photochemical model (Martinis, Wilson, & Mendillo 2003) it was later extended to include multispecies plasma diffusion (Mendillo et al., 2011). The model calculates, between 80 and 400 km, the ionization produced from a neutral atmosphere generated from mixing ratios obtained from the Mars Climate Database (e.g., Millour et al., 2017). 16 ion species are tracked in the model and 81 chemical reactions are considered (Matta, Withers, & Mendillo, 2013).

Indian researchers have developed a 1-D model able to simulate the ionization produced by the solar EUV and X-ray radiation, the interaction of photoelectrons with the neutrals, solar wind particle precipitation and meteor ablation (Haider, Pandya, & Molina-Cuberos, 2013; Pandya & Haider, 2014). It can simulate the ionosphere between about 50 and 200 km, and has been used to simulate both the dayside and the nightside ionosphere.

Other 1-D photochemical models include the ionospheric model for Mars and Venus developed at Köln University, called IonA (Peter et al., 2014), and the photochemical model developed by Vladimir Krasnopolsky, which among other many applications has been used to study the ionosphere (Krasnopolsky, 2002).

Two 3-D Global Climate Models (GCM) have also been used to simulate the behavior of the Martian ionosphere and its coupling with the neutral lower and upper atmosphere. The Mars Thermospheric GCM (Bougher, Engel, Roble, & Foster, 1999; Bougher, Engel, Hinson, & Murphy, 2004) is a thermospheric GCM coupled to the NASA Ames Mars GCM to take into account of the effects of the lower atmosphere. A dayside photochemical ionosphere model to simulate the five major ions is included in the MTGCM. The coupled models were used, for example, to simulate the longitudinal variations observed in the altitude of the ionospheric peak. Recently, the coupled MTGCM–Ames GCM models formulation has evolved into a single whole-atmosphere model, the Mars Global Ionosphere–Thermosphere Model (MGITM; Bougher et al., 2015a). This new non-hydrostatic global model also accounts for plasma dynamics in the absence of magnetic or electric fields.

The Laboratoire de Météorologie Dynamique Mars Global Climate Model (LMD-MGCM) is a whole-atmosphere GCM (Forget et al., 1999; González-Galindo, Forget, López-Valverde, Angelats i Coll, & Millour, 2009) that includes also a photochemical description of the ionosphere (González-Galindo et al., 2013) and plasma dynamics in the absence of magnetic fields (Chaufray et al., 2014). The photochemical scheme considers 92 chemical reactions between 25 species; the observed day-to-day variability of the UV ionizing solar flux and the dust load in the lower atmosphere during the last eight Martian years are included in the model (González-Galindo, López-Valverde, Forget, García-Comas, Millour, & Montabone, 2015). Selected outputs of the LMD-MGCM, including electron-density profiles, are made available to the community through the Mars Climate Database (Lewis et al., 1999; Millour et al., 2017).

Given the strong couplings and feedbacks between the lower atmosphere, the thermosphere-ionosphere, the exosphere and the magnetosphere of Mars, suites of coupled atmospheric/ionospheric, exospheric, and magnetospheric models have been developed recently. One of them is the HELIOSARES set of models (Leblanc et al., 2017), including the LMD-MGCM, an exospheric Monte Carlo test particle model (Yagi et al., 2012) and the LatHys hybrid magnetospheric model (Modolo et al., 2016). Also the Martian version of the IRAP plasmasphere–ionosphere model has been coupled to the Mars Climate Database derived from the LMD-MGCM (Sánchez-Cano et al., 2018a). Another suite has been developed at the University of Michigan by coupling the MGITM atmospheric model with a Monte Carlo exospheric model (Valeille, Tenishev, Bougher, Combi, & Nagy, 2009) and the hybrid HALFSEF magnetospheric model (Brecht, Ledvina, & Bougher, 2016).

A different family of models, called semi-empirical models, is created by the fit of large ionospheric datasets to empirical equations, generally derived from Chapman theory. These models allow for a simple description of the ionospheric variability with different parameters, such as the solar activity, the SZA, etc. So, Němec, Morgan, Gurnett, Duru, and Truhlík (2011a) produced a semi-empirical model based on Mars Express MARSIS data for the behavior of the ionosphere at and above the main peak, which was later extended (Němec, Morgan, Gurnett, & Andrews, 2016) to include the effects of crustal magnetic fields. Mendillo, Marusiak, Withers, Morgan, and Gurnet (2013a) developed another semi-empirical model for the main ionospheric peak, also based largely on MARSIS data. Another semi-empirical model, called NeMars (Sánchez-Cano, Radicella, Herraiz, Witasse, & Rodríguez‐Caderot, 2013), is able to describe both the primary and secondary ionospheric layers.

Finally, it has to be noted that, in addition to these “generalistic” ionospheric models, a large number of models devoted to study individual processes affecting the ionosphere or specific ionospheric regions has been developed. Examples include models of the ionosphere below 100 km produced by cosmic ray ionization and/or meteor ablation (Molina-Cuberos et al., 2002; Plane et al., 2018), the effects of photoelectrons (e.g., Peterson et al., 2016), or the airglow produced by ionization of atmospheric species (e.g., Jain & Bhardwaj, 2012; Shematovich et al., 2008).

### Past Ionospheric Observations

#### US Mariner and Soviet Mars Missions

The first observation of a planetary ionosphere (apart that of the Earth) was achieved during the Mariner 4 flyby of Mars in 1964 (Kliore et al., 1965) using radio occultation. This technique has been by far the most commonly used to sound the Martian ionosphere. It is based on the modification of the radio signal link between a spacecraft and a ground station on Earth when passing through the Martian atmosphere. Mariner 4 radio occultations provided the first electron-density profile of Mars, showing a peak at an altitude of about 120 km with an electron density of about 9 × 1024 cm-3, and hints of the presence of a lower peak at around 100 km of altitude (Fjeldbo, Fjeldbo, & Eshleman, 1966). Ulterior radio occultations performed by Mariner 6 and 7 in 1969 yielded two additional electron-density profiles (Hogan, Stewart, & Rasool, 1972). The differences in the peak intensity and peak altitude with Mariner 4 observations were quickly interpreted as resulting from variations in e solar activity and in Sun–Mars distance (Whitten & Colin, 1974). While Mariner 4, 6 and 7 performed some radio occultations over the nightside of the planet, no nightside ionosphere was detected.

While no direct information about the composition of the Martian ionosphere was obtained from these early measurements, computational photochemical models constrained by observations of UV atmospheric emissions (in particular, the CO2+ UV doublet and the Cameron bands, both originated ultimately from the interaction of CO2 with the incoming UV solar radiation) performed by Mariner 6 and 7 showed that the most abundant ion was not CO2+ as would be expected from a CO2-dominated upper atmosphere, but likely O2+. The reason is that CO2+ quickly reacts with atomic oxygen to produce O2+, whose loss is mainly dissociative recombination, a relatively slow process (McElroy & McConnell, 1971; Stewart, 1972).

Mariner 9 was, shortly before the soviet missions Mars 2 and Mars 3, the first spacecraft to orbit another planet. Its payload for the study of the upper atmosphere was similar to that of the previous Mariner 6 and 7 missions. During its lifetime, the Mariner 9 radio occultation experiment obtained about 115 electron-density profiles over two periods (November–December 1971 and May–June 1972) (Kliore, Fjeldbo, Seidel, Sykes, & Woiceshy, 1973; Whitten & Colin, 1974). The variation with SZA of both the peak intensity and the peak altitude was analyzed and was found to compare favorably to theoretical expectations based on Chapman theory. Most profiles show a similar vertical structure, with constant slopes above the main ionospheric peak (Kliore, 1992). As the slope of the density profile can be related to temperature, the similarity in the slopes indicate similar temperatures through the measurement period. When approaching the terminator (the transition between the illuminated and the non-illuminated portions of the planetary disk), the vertical structure of the profile is significantly more variable (Kliore, 1992) and the electron-density peak broadens and becomes less well-defined. An interesting result obtained from Mariner 9 observations was the effect on the ionosphere of a global dust storm. Mariner 9 arrived to Mars during a global dust storm, and the first set of radio occultation measurements was obtained during its declining phase. The ionospheric peak was found at significantly larger altitudes during the storm, which was attributed to the warming of the lower atmosphere due to the absorption of solar radiation by the dust and the subsequent atmospheric expansion (Kliore, 1992) (Figure 2). A more recent analysis of the Mariner 9 dust storm dataset (Withers & Pratt, 2013) showed how the peak altitudes steadily decreased during the decay phase of the storm. Recently, this valuable dataset has been recovered, made available to the community, and thoroughly analyzed (Withers, Weiner, & Ferreri, 2015a). Three findings of this recent analysis are: 1) similarly to the altitude of the main peak, the top of the ionosphere extended to significantly higher than usual altitudes during the time of the dust storm, and its altitude decreased by more than 70 km with the decay of the storm; 2) the electron concentration at the secondary peak was found to decrease with SZA, similarly to the main peak; and 3) the separation between the primary and secondary peak was found to be quite constant, and not modified by the global dust storm, which indicates that the atmospheric heating and expansion produced by the storm occurred well below the secondary peak.

The soviet orbiters Mars 2 and Mars 3 operated above Mars from the end of 1971 until May 1972, and Mars 5 obtained data during February 1974. Mars 4, similar to Mars 5, failed orbit insertion. The orbiters were equipped with instrumentation devoted to the study of the solar wind interaction with the atmosphere. A review of the results obtained on this subject can be found in Vaisberg, 1992. Of particular interest for ionospheric studies was the discovery of an ionosphere on the nightside of the planet. While the Mariner missions failed to detect it, Mars 4 and 5 succeed in obtaining two electron-density profiles of the nightside (SZAs 127º and 106º, respectively). Peak electron densities of 4–5 × 103 cm-3 at 110 and 130 km were measured, with hints of a secondary peak (Savich et al., 1976). The necessity for an additional source of ionization on the nightside was derived from these early measurements.

#### First In-situ Observations: Viking Missions

An important step forward in our understanding of the Martian ionosphere was achieved by the first in-situ measurements by the Retarding Potential Analyzers (RPA) of Viking Landers 1 and 2 in 1976. The descent of the Viking landers to the Martian surface enabled for the first time the probing of the composition and the energetics of the Martian ionosphere.

Regarding the composition, profiles of O2+, CO2+ and O+ were obtained (Figure 3). As deduced previously from photochemical modeling constrained by Mariner observations, O2+ was found to be the dominant ion at almost all altitudes probed. The contribution of O+ was found to increase with altitude, reaching a maximum at about 220 km. The ratio of CO2+ to O2+ was found to vary with altitude, from a factor of about 6 to almost 100, and photochemical modeling enabled the derivation of the atomic oxygen density from ion profiles (Chen, Cravens, & Nagy, 1978; Fox & Dalgarno, 1979; Hanson et al., 1977). Modeling constrained by the Viking observations showed that plasma transport processes become dominant in the upper ionosphere (Chen, Cravens, & Nagy, 1978).

For investigation of energetics, the RPA measured, for the first time, ion temperatures. Above the ionospheric peak, the ion temperature departs from the neutral temperature, exceeding 1,000 K above 200 km. A first attempt to understand these high ion temperatures with a model solving the coupled continuity, energy, and momentum equations (Chen, Cravens, & Nagy, 1978) failed to explain the observations if EUV solar radiation was the main heating source; it needed to invoke a topside heating source, probably related to solar wind–ionosphere interaction. Later modeling (Rohrbaugh, Nisbet, Bleuler, & Herman, 1979) found that the energy produced by exothermal ionospheric chemical reactions was sufficient to produce the observed ion temperatures in the presence of a magnetic field reducing thermal conductivity. Although the Viking RPAs were in principle capable of measuring also the temperature of the electrons, the difficulties associated with the analysis of this dataset delayed the derivation of electron temperature profiles for more than 10 years. Hanson and Mantas (1988) derived the presence of three populations of electrons in the ionosphere of Mars with different temperatures, probably including some very energetic (“hot”) electrons from the solar wind. It was also found that the ionospheric plasma pressure was lower than the solar wind pressure, suggesting that a magnetic field must be present in the ionosphere to compensate for the missing pressure.

The Viking orbiters also performed radio occultation measurements of the Martian ionosphere during their lifetimes, collecting hundreds of electron-density profiles. A first analysis of the data obtained during the first year of the mission was published by Lindal et al. (1979), while a comprehensive study of the full dataset in combination with all previous radio occultation measurements was performed by Zhang, Luhmann, Kliore, & Kim (1990a). Using this extended dataset, the decrease of the peak electron density and the increase of the peak altitude with increasing SZA were found to be quite close to those predicted by simple theoretical considerations (Chapman theory). Viking radio occultations on the nightside ionosphere were studied by Zhang, Luhmann, and Kliore (1990b). Of 50 occultations in the nightside, only 19 included a detectable ionospheric peak, with an average density of 5 × 103 cm-3, the other measurements being noisy and the peaks probably below the detection limit of the radio occultation experiment. Nightside peak altitudes are quite high and show a large scatter. Comparison with Venus observations by Pioneer Venus Orbiter (PVO) showed similar SZA dependency for the nightside ionospheres of both Mars and Venus, although peak densities are lower and peak altitudes higher for Mars. By analogy with Venus data, Zhang et al. (1990b) suggested that plasma transport from the dayside to the nightside must be a significant source of nightside ions, but the lack of simultaneous measurements of the solar wind parameters prevented more definitive conclusions.

### Recent Observations

We summarize here the observations of the Martian ionosphere obtained during the period since 1997. We will first provide a short description of the missions and instruments at the origin of the observations and the main features of the obtained datasets. Then, we present the progress in a selection of relevant ionospheric topics achieved by these observations and the application of the computational models described above

#### Mars Global Surveyor

Mars Global Surveyor’s (MGS) arrival at Mars in 1997 marked the end of 20 years without a fully successful Martian mission and can be considered as the starting point of a golden era in Mars exploration. After a long aerobraking period, which provided information about the temperature and density structure in the neutral upper atmosphere (Keating et al., 1998), the final orbit of MGS was a two-hour circular orbit with an altitude of about 380 km, almost polar and sun-synchronous (2 a.m.–2 p.m.) (Albee, Palluconi, & Arvidson, 1998). MGS was a global mapping mission and carried a suite of five instruments to study the interior, surface, and atmosphere of the planet.

Mars Global Surveyor performed radio occultation to sound the ionosphere. More than 5,500 electron-density profiles were obtained on different occultation campaigns distributed during four Mars years (e.g., Fox & Yeager, 2009). MGS increased by more than one order of magnitude the number of profiles obtained. With the exception of about 220 profiles obtained in the high latitudes of the southern hemisphere, all the data were taken in the high latitudes of the Northern hemisphere, and correspond to SZAs larger than 70. The characteristics of MGS orbit allow for a very good longitudinal sampling at fixed latitude, local time, and SZA.

One of the most relevant findings of the mission was the measurement of the magnetic field at Mars by the magnetometer/ electron reflectrometer instrument (MAG/ER). While the existence of a global intrinsic magnetic field was ruled out, the presence of magnetized localized regions on the Martian crust was discovered (Acuña et al., 1998, 2001), mostly in ancient cratered terrains in the Southern hemisphere. MGS MAG/ER also improved the characterization of the interaction of the solar wind with the ionosphere, which was found to combine features of Venus-like and comet-like solar wind interaction (Acuña et al., 2001). In particular, the presence of an induced magnetic field produced by the solar wind–ionosphere interaction was confirmed.

Years after the end of its mission in 2006, the analysis of the MGS datasets was still producing interesting results (e.g., Mayyasi,Withers, & Fallows, 2018)

#### Mars Express

Mars Express was the first mission to another planet launched by the European Space Agency (ESA). It was launched in June 2003 and was inserted into Martian orbit at the end of 2003, and was operating successfully in 2019. Its orbit around Mars is quasi-polar, elliptical with a periapsis altitude of about 250 km and apoapsis at about 10,000 km (Bertaux et al., 2006). Unlike previous missions such as Mars Global Surveyor, Mars Express is not a sun-synchronous satellite, so it can explore the local time-variability of the atmosphere. The long duration of the mission has allowed monitoring of the atmosphere/ionosphere of Mars during more than one solar cycle, improving the characterization of the effects of solar variability. One of the scientific objectives of the mission is the characterization of the atmosphere and the ionosphere, including their interaction with the surface and the interplanetary medium. To achieve their objectives, the mission includes a series of seven scientific instruments. Two of them (MaRS and MARSIS) are capable of sounding the ionosphere directly, while two others (ASPERA-3 and SPICAM) provide indirect information about it.

The Mars Express Radio Science experiment (MaRS) performs radio occultations at two different frequencies (X-band and S-band) to study the neutral lower atmosphere and the ionosphere (Pätzold et al., 2004). The differences between the MGS and the Mars Express orbits make radio occultations performed by both missions quite complementary (Pätzold et al., 2016): MGS radio occultations were confined to high latitudes mostly in the Northern hemisphere, while Mars Express provide a global latitudinal coverage. On the other hand, MGS achieves a much better spatial sampling with a very good longitudinal coverage. MaRS had collected about 900 electron-density profiles over more than 7 Mars years by 2019.

The ASPERA-3 experiment (Analyzer of Space Plasmas and Energetic Atoms) is devoted to study the solar wind–atmosphere interaction. For this purpose it can measure energetic electrons and ions, and Energetic Neutral Atoms (ENAs) originated by charge-exchange between solar wind particles and the neutral exosphere (Barabash & Lundin, 2006). Among many other results, ASPERA-3 has characterized the pressure balance in the boundary between the ionosphere and the solar wind (Dubinin et al., 2008)

The SPICAM (Spectroscopy for Investigation of Characteristics of the Atmosphere of Mars) instrument is devoted to the characterization of the atmospheric structure and composition. It consists on a dual ultraviolet-infrared spectrometer (Bertaux et al., 2006; Montmessin et al., 2017). The UV spectrometer can observe airglow features produced by the photoionization and electron impact ionization of CO2, such as the CO2+ UV doublet (Leblanc, Chaufray, Lilensten, Witasse, & Bertaux, 2006). Unfortunately, after a significant loss of efficiency in late 2011, the SPICAM UV spectrometer stopped working at the end of 2014 (Montmessin et al., 2017)

#### MAVEN

The Mars Atmosphere and Volatile Evolution (MAVEN) mission is devoted to the study of the upper atmosphere of Mars, its interaction with solar radiation and the solar wind, and the characterization of the atmospheric escape, in order to better understand the long-term evolution of the Martian climate (Jakosky et al., 2015). MAVEN is measuring the thermospheric/ ionospheric structure and variability together with their main drivers (EUV solar radiation, solar wind, solar energetic particles) to characterize their controlling processes. MAVEN arrived at Mars in September 2014 and is still operating nominally. Its orbit is elliptical, with a periapsis altitude of 150 km and an apoapsis altitude of more than 6,000 km with a period of 4.5 h. This orbit allows both in-situ sampling of the upper atmosphere during periapsis, and global mapping around apoapsis. The precession of the orbit produces a drift in the latitude and local time of periapsis during the mission. During some specific campaigns (called “deep-dips”) the periapsis altitude is lowered to about 120 km, corresponding approximately to the altitude of the homopause (Bougher et al., 2015b), allowing for in-situ characterization of the lower thermosphere and the ionosphere below the main peak.

MAVEN includes eight scientific instruments. Three of them (NGIMS, LPW and STATIC) characterize the ionosphere in situ. A remote sensing instrument, IUVS, observes atmospheric emissions, some of them originated by ionospheric processes.

NGIMS (Neutral Gas and Ion Mass Spectrometer) is a mass spectrometer able to measure the composition of the neutral atmosphere and the ionosphere below about 500 km. (A complete description of the instrument can be found in Mahaffy et al., 2015). Usually, measurements of the neutral atmosphere and the ionosphere are performed at alternating orbits (Benna et al., 2015a). Temperatures in the upper atmosphere are derived from the scale heights of different species. Focusing on the ionosphere, during the first months of operation NGIMS was able to measure abundance profiles of 22 ion species, including protonated ions in larger than expected abundances, and some isotopes (Benna et al., 2015a).

The LPW (Langmuir Probe and Waves) instrument measures the electron number density and electron temperature profiles using the Langmuir probe technique, as well as waves able to heat ions (Andersson et al., 2015). The range of measurement is 100–106 cm-3 for electron density and 500–50,000 K for electron temperature. Note that, as the electron temperatures in the region of the main ionospheric peak are expected to be similar to the neutral temperatures, and thus lower than 500 K, caution needs to be exercised when interpreting the LPW electron temperature measurements there. The LPW includes also a quasi-independent component, the Extreme Ultraviolet (EUV) instrument. EUV is composed of three radiometers able to measure the solar irradiance at three selected spectral intervals: 0.1–7 nm, 17–22 nm and at Lyman-α‎ (Eparvier, Chamberlin, Woods, & Thiemann, 2015). From these measurements, the full spectral irradiance from 0 to 190 nm can be reconstructed. The presence of the EUV monitor, providing for the first time measurement of the solar forcing acting directly over the upper atmosphere of Mars, obviates the need to extrapolate solar measurements performed on Earth to Martian conditions.

STATIC (SupraThermal And Thermal Ion Compostion) measures the density, temperature and flows of ions (H+, O+, O2+, CO2+) in the ionosphere, the escaping plasma and the ions accelerated by the solar wind interaction (McFadden et al., 2015). The ion temperatures are important for computing the chemical reactions between ions and neutrals (e.g., Bougher et al., 2015a). The high dynamic range of the instrument enables measuring of both cold ionospheric ions and accelerated particles.

IUVS (Imaging Ultraviolet Spectrograph) measures the UV airglow in the 115–340 nm spectral range, with a spectral resolution of between 0.6 and 1.2 nm (McClintock et al., 2015). The instrument is mounted on an Articulated Payload Platform (APP), giving the instrument pointing capabilities independent of the spacecraft. This allows for an almost continuous monitoring of the Martian atmosphere using four different observing modes. Some of the atmospheric emissions detectable by IUVS are originated by ionospheric processes, such as the CO2+ UV doublet, the CO2+ Fox–Duddenback-X system, and CII lines.

### Recent Advances in Ionospheric Research

#### SZA Variability of Ionospheric Peaks

The variability of the observed ionosphere with SZA has been a topic of study since the first ionospheric measurements by the Mariner missions (e.g., Hantsch & Bauer, 1990). The many ionospheric data collected in the last 30 years have enabled better determination of this variability.

Fox and Yeager (2009) studied the SZA variability of the peak density for both the main and the secondary ionospheric peaks as derived from the MGS dataset. They divided the observations into bins of different solar activity according to the F10.7 solar proxy index and fitted an expression of the form:

$Display mathematics$

They found median values for the exponent, a, of 0.46 for the primary peak and 0.48 for the secondary peak. These are slightly lower than the value 0.5 expected from Chapman theory. This deviation is thought to be due to the plane-parallel assumption in Chapman theory. Median values for the subsolar peak density range between 1.5 × 105 and 2 × 105 cm-3 depending on solar activity.

Peter et al. (2014) performed a similar fit to the one above for the density of the main peak observed by MaRS/Mars Express in the SZA range 55–85. They found a value for the exponent a of about 0.44. They also studied the SZA variability of the peak altitude, which was found to increase when approaching the terminator, and of the total electron content, decreasing with increasing SZA. All these variabilities were reproduced by the 1D IonA model (see section on ionospheric modeling above).

As mentioned before, unlike radio occultations, radar sounders can explore the whole range of SZAs. Different studies of the SZA variability in the MARSIS dataset have been published (Gurnett et al., 2008; Morgan et al., 2008). For example, Němec et al. (2011a) used an expression similar to the one above, but instead of the cosine of the SZA they used the Chapman grazing incidence function, which better represents the behavior for SZA>75. In this case they found an exponent value of about 0.55, the differences with previous analyses being explained by the difference between the cosine and the Chapman function. They also studied the SZA dependence of the peak altitude, finding a subsolar peak altitude of about 125 km. The SZA variability of the peak density and peak altitude observed by MARSIS can be seen in Figure 5 and Figure 6, respectively.

The dependence of the TEC with SZA has also been studied using the MARSIS dataset (Lillis et al., 2010; Mendillo et al., 2013b; Safaeinili et al., 2007). An interesting result of these studies is that, for similar values of SZAs, there is an asymmetry in the TEC in the dawn and dusk sides of the planet.

Different computational models have been used to explore the simulated SZA variability of the peak densities, all comparing favorably with observations (e.g., Fallows, Withers, & Matta, 2015; González-Galindo et al., 2013; Peter et al., 2014).

#### Solar Cycle Variability

The increase in the density of ionospheric peaks with solar activity was identified early (Whitten & Colin, 1974) and has since attracted significant attention. Hantsch and Bauer (1990) combined data from the Mariner, Mars, and Viking missions and fitted the variation of the peak electron density with the F10.7 solar proxy index. They found that, from the data available at the time, the solar cycle variability of the ionosphere of Mars was similar to that of Venus.

Many published studies have used the recent Martian datasets to characterize the response of the ionospheric layers to the solar cycle (e.g., Breus et al., 2004; Fox & Yeager, 2009; Girazian & Withers, 2013; Morgan et al., 2008; Němec et al., 2011a; Sánchez-Cano et al., 2016; Withers, Morgan, & Gurnett, 2015b; Zou, Wang, & Nielsen, 2006). In many cases, the variability is expressed by a power law:

$Display mathematics$

where $F$ is the ionizing solar flux, usually represented by a proxy index such as F10.7 or E10.7. The derived values of the exponent for the main peak vary in different reports, with a mean of about 0.35 (Girazian & Withers, 2013). Note that the choice of the representation of the solar flux is important. Withers et al. (2015b) combined data from MGS radio occultations and by MARSIS to cover a full solar cycle. They found that the density in the main peak ceased increasing with increasing F10.7 at about F10.7 = 130. They attributed this not to a real physical effect, but to the limitations of the F10.7 index. This saturation did not happen when using other indexes such as Lyman-alpha emission or the Mg II index.

While in many studies only the main ionospheric peak was studied, in some cases the variability of the secondary ionospheric peak was also investigated (Fallows et al., 2015; Fox & Yeager, 2009). Fox and Yeager (2009) found that the exponent for the lower peak is larger than for the primary peak, due probably to the larger variability in the shortest wavelengths at the origin of the secondary peak. Sánchez-Cano et al. (2016) used several Mars Express ionospheric datasets to study the solar cycle variation of both peaks, as well as that of the topside ionosphere, taking the interaction with the solar wind into consideration. They found that the density of the topside ionosphere was reduced during low solar activity periods, and this reduction was larger when the ionosphere was in a more magnetized state. Dubinin, Fraenz, Andrews, and Morgan (2017) found that dependence on solar activity also operated local electron densities sampled by MARSIS at altitudes between 300 and 1,500 km.

#### Effects of Short-term Solar Variability

Apart from the 11-year solar cycle, the Sun is also variable at shorter time scales, and the effects of these variations over the ionosphere have also been studied. Withers and Mendillo (2005) identified the signatures of the 27-day solar rotation on the peak electron densities from the MGS radio occultation dataset, with a delay on the response of the Martian ionosphere to changes in the Sun as observed from the Earth. A similar finding was made using MARSIS data (Nielsen et al., 2006). Venkateswara Rao, Balan, and Patra (2014), using the MGS radio occultation dataset, found that the solar rotation effects could be felt at all explored altitudes in the ionosphere (<220 km). The solar rotation effects were also simulated with a GCM when including the observed day-to-day variability of the UV solar flux (González-Galindo et al., 2013)

The ionosphere has also been shown to respond to short-term changes in the solar activity, such as solar flares, which increase greatly the energy output from the Sun (both radiation and particles) in periods of minutes. Nielsen et al. (2006), using MARSIS data, found an increase in ionospheric peak density of about 30 % in few minutes, coincident with an increase in X-ray radiation measured at the Earth. Mendillo Withers, Hinson, Rishbeth, and Reinisch (2006) detected large density increases (60–200 %) at and below the secondary peak in the MGS radio occultation dataset, also coincident with detection of solar flares at the Earth. Mahajan, Lodhi, and Singh (2009) found that solar flares produced well-formed lower ionospheric peaks, and some affected simultaneously both the main and the secondary peak. Haider (2012) found that the TEC in the region of the lower ionospheric peak as measured by MGS increased by a factor of 6 during a strong solar storm in May 2005. The ionospheric effects of solar flares has been simulated using 1D photochemical models (Haider et al., 2012; Lollo et al. 2012), finding good agreement with the observations.

During solar events, the flux of high energy solar particles also increase significantly. Haider (2012) found that the Coronal Mass Ejection (CME) associated with a solar flare in May 2005 produced an increase in the TEC of the lower ionosphere a couple of days after the flare. Other solar weather events have been studied (e.g., Duru et al., 2017; Morgan et al., 2014), revealing significant ionospheric effects, in particular on the nightside, and a compression of the ionosphere. However, Ulusen, Brain, Luhmann, and Mitchell (2012) did not find clear evidence for ionospheric density increase between 100 and 200 km altitudes coincident with several SEP effects, which could be attributed to the extra ionization being produced at lower altitudes. Of special interest is the strong solar event on September 2017 which affected both the Earth and Mars (Guo et al., 2018)

#### Neutral Atmosphere–Ionosphere Coupling

Given that the ionosphere is produced from the upper neutral atmosphere, important couplings and feedbacks between them are to be expected. An extreme example is the dramatic increase in the altitude of the ionospheric peak during the global dust storm detected by Mariner 9 (Kliore, 1992; Withers & Pratt, 2013). But more subtle interactions have also been detected.

Bougher, Engel, Hinson, and Forbes (2001), and Bougher, Engel, Hinson, and Murphy (2004), found that the altitude of both the primary and secondary peak varied with the geographical longitude (SZA, latitude and season being similar) with prominent wave-2 and wave-3 patterns (Figure 7). Similar behavior was observed in the neutral densities. This longitudinal inhomogeneity is attributable to non-migrating tides, originated by the interaction of solar illumination with the planet’s topography and propagated to the upper atmosphere. A similar pattern was obtained in simulations with the MTGCM. Mahajan, Singh, Kumar, Raghuvanshi, and Haider (2007) also showed a longitudinal variability in the peak intensity in the MGS dataset.

Zou, Lillis, Wang, and Nielsen (2011) found a seasonal variation in the altitude of the main peak. According to LMD-MGCM predictions of the neutral atmosphere, this increase can be directly related to the seasonal variation of the neutral atmosphere (both neutral atmospheric density and neutral scale height between 20 and 130 km) below the ionospheric peak. Later simulations with the LMD-MGCM including the ionosphere confirmed a significant aphelion-to-perihelion increase both in peak altitude and peak density due to the seasonal variation of the Sun–Mars distance (González-Galindo et al., 2013)

Sánchez-Cano et al. (2018a) analyzed 10 years of TEC data from MARSIS. When looking at the annual behavior of the TEC they found a peak during the Northern spring–summer season not due to solar irradiance, but coincident with a thermospheric density increase, possibly produced by the CO2 in the lower atmosphere. See Figure 8.

It is also important to note that the links between the ionosphere and the neutral atmosphere have been exploited to obtain information about the latter from measurements of the former. For example, using NGIMS measurements of the ionospheric composition and a photochemical model, Fox, Johnson, Ard, Shuman, & Viggiano (2017) have derived the abundance of oxygen in the upper atmosphere.

#### Meteoric Ionosphere

MaRS observations unveiled the presence of an third ionospheric peak at altitudes between 65 and 110 km with average concentration of 8 × 103 cm-3 (Pätzold et al., 2005; see Figure 9). The presence of this peak is sporadic and its altitude appears to be correlated to that of the secondary peak. Computational models accounting for the deposition of material from meteor ablation had predicted the presence of a peak of metallic ions (mainly Mg+ and Fe+) of meteoritic origin at altitudes between 80 and 100 km (e.g., Molina-Cuberos, Witasse, Lebreton, Rodrigo, & López-Moreno, 2003), so the third layer was attributed to meteor ablation. Later, plasma layers at similar altitudes were identified in the MGS dataset (Pandya & Haider, 2012; Withers, Mendillo, Hinson, & Cahoy, 2008), and even in eight of the electron-density profiles obtained by Mariner 7 and 9 (Withers, Christou, & Vaubaillon, 2013), although the Mariner 9 observations of the layer were later found to be most likely non-reliable (Withers et al., 2015a). No significant correlation between the characteristic of this low-altitude plasma layer with atmospheric or solar features (SZA, solar flux) was found.

The interpretation of the low-altitude plasma layer as of meteoric origin has been recently challenged by MAVEN observations of metallic ions. IUVS observed a Mg+ layer at 90 km, but with an average concentration of 250 cm-3, and in no case larger than 1,000 cm-3, still one order of magnitude lower than the electronic density observed by MaRS (Crismani et al., 2017). This suggests that formation mechanisms other than meteor ablation may be the cause of the low-altitude plasma layer. Another surprising observation was the absence of neutral Mg, which models predict should be as abundant as Mg+ (Plane et al., 2018). NGIMS has also detected Mg+ and Fe+ above 130 km (Grebowsky et al., 2017). Contrary to expectations, no gravitational separation was found between the two ions. Another surprising discovery was the presence of isolated layers of metallic ions in the upper ionosphere. Clearly, further modeling work is needed to better understand MAVEN observations of metallic ions.

#### The Nightside Ionosphere

Far less is known about the nightside ionosphere (and particularly the deep nightside, SZA larger than about 110) than about its dayside counterpart. The low plasma densities at the nightside due to the absence of solar radiation are a challenge for observing instruments. Most of our knowledge of the nightside ionosphere comes from Mars Express and, more recently, MAVEN observations.

Both MARSIS and MaRS observations have shown that densities in the nightside ionosphere in regions without crustal fields decrease with increasing SZA until about 120 degrees, suggesting that plasma transport from the dayside may be an important source of nightside ionosphere close to the terminator (Němec, Morgan, Gurnett, & Duru, 2010; Withers et al., 2012a). Similarly, Cui, Galand, Yelle, Wei, and Zhang (2015) analyzed MARSIS TEC data to find that this magnitude decreased with time several thousand seconds after crossing the terminator, indicating dominance of transport in these timescales. Simulations with the LMD-MGCM show the presence of a nightside ionosphere due to transport from dayside and the long chemical lifetime of O2+ and NO+ (Chaufray et al., 2014; González-Galindo et al., 2013), but predicted densities in the deep nightside fall well below the observed values (Figure 10). For higher SZAs, or in regions of strong crustal field, SZA dependence is no longer present, suggesting that precipitation of energetic particles plays a dominant role there. The nightside ionosphere is shown to be quite patchy and variable, and significantly enhanced in regions of vertical crustal magnetic field (Němec, Morgan, Gurnett, & Brain, 2011b; Safaeinili et al., 2007). The reason is the increase in the precipitation of energetic particles there (Shane, Xu, Liemohn, & Mitchell, 2016), while horizontal magnetic fields prevent particle precipitation. This is in agreement with expectations from models of particle precipitation (Fillingim et al., 2007; Lillis, Fillingim, & Brain, 2011). Nightside ionospheric density has also been shown to increase with solar wind dynamic pressure (Diéval, Morgan, Němec, & Gurnett, 2014).

Recently, the composition of the nightside ionosphere has been studied using NGIMS data (Girazian et al., 2017a). NO+ and HCO+ are enhanced with respect to the dayside, and NO+ dominates in the nightside below 130 km. This is in good agreement with predictions by the LMD-MGCM including only photochemistry and transport, but no particle precipitation (González-Galindo et al., 2013). Large variations of the nightside densities were found, both on small temporal and geographical scales, and with seasons. The effects of particle precipitation on the nightside ionospheric composition have also been studied (Girazian et al., 2017b), finding a significant increase in CO2+, O2+ and O+ below 200 km during electron precipitation events, but no effect on NO+.

Separate models have been used until now to study the effects of particle precipitation (Fillingim et al., 2007; Lillis et al., 2011) and day–night transport (Chaufray et al., 2014; González-Galindo et al., 2013) in the nightside ionosphere. Haider et al. (2013) included, in addition of particle precipitation, the ablation of meteors. However, a single model, ideally global and including all relevant processes, is needed. This would greatly improve our knowledge of the relative importance of the different processes populating the nightside ionosphere.

#### Ionospheric Effects of Crustal Fields

The crustal magnetic fields discovered by MGS, and localized mostly in the southern hemisphere and at particular longitudinal ranges, has been shown to affect the Martian ionosphere. So, Ness et al. (2000) showed that the scale height in the topside ionosphere derived from MGS radio occultations was significantly larger over crustal field regions with a vertical magnetic field. Excess electron densities have been found over “mini-magnetospheres” associated to crustal magnetic fields in MGS radio occultations (Krymskii et al., 2003, 2004). Nielsen et al. (2007) observed significant peak electron density increases as measured by MARSIS over crustal regions with vertical magnetic field, not associated to increases in the solar flux or to enhanced particle precipitation. An inflation of the ionosphere over regions of strong crustal magnetic field has also been observed (e.g., Andrews et al., 2015; Diéval et al., 2014; Duru et al., 2006; Gurnett et al., 2005). The TEC and the local electron density measured by MARSIS over regions of crustal field also increases with the intensity of the magnetic field (Dubinin, Fraenz, Andrews, & Morgan, 2016).

In most cases, the increased ionization over crustal field regions with vertical magnetic field has been interpreted as a consequence of an increased electron temperature produced either by penetration of the solar wind (Ness et al., 2000) or by plasma instabilities (Nielsen et al., 2007). Given that the rate of ion–electron recombination, the main loss mechanism of the ionosphere, depends inversely on the electron temperature, larger electron temperatures should produce larger electron densities. High electron temperatures also affect plasma transport above the photochemical region of the ionosphere, again producing larger electron densities (e.g., Flynn et al., 2017).

However, MAVEN observations defy this interpretation. Flynn et al. (2017) analyzed LPW measurements of the electron density and temperature over crustal fields. They found that while the electron density increased over crustal field regions, the electron temperature decreased. Clearly, there is still much to learn about the interaction of the ionosphere with crustal fields and the solar wind.

#### Topside Ionosphere

The behavior of the topside ionosphere is dominated by plasma dynamic processes and the interaction with the impinging solar wind. The topside ionosphere is characterized by a decrease in electron density with increasing altitude. Withers et al. (2012b) analyzed MaRS electron density profiles, and found that only a small proportion of the profiles were characterized by a single scale height (rate of decrease of the density with altitude), while most often a profile is characterized by two or even three different scale heights, indicating either a transition between the physical mechanisms establishing the scale height or a change in some parameter affecting it, such as electron temperature. Duru et al. (2008) analyzed MARSIS measurements of the local electron density to find that the scale height of the topside ionosphere changes from dayside to nightside. They also found important oscillations in the measured electron densities. These oscillations were later analyzed in detail by Gurnett et al. (2010), who attributed them to solar wind interaction. Gopika and Venkateswara Rao (2018) found also strong oscillations in the MGS topside electron-density profiles.

From the analysis of MARSIS remote sensing data, Kopf, Gurnett, Morgan, and Kirchner (2008) found the presence of layers in the topside ionosphere at altitudes between about 180 and 240 km. They found that the occurrence of these layers decreased with increasing SZA, while their density slightly decreased and their altitude slightly increased with SZA. A later analysis using a more complete MARSIS dataset found that the occurrence of the topside layers was lower in the regions of strong crustal field (Kopf, Gurnett, DiBraccio, Morgan, & Halekas, 2017) . Mayyasi et al. (2018) studied the presence of topside layers in the MGS radio science dataset. They found a topside layer in about 60 % of the profiles at an average altitude of 170 km and with average concentration of 9 × 103 cm-3 (an example can be seen in Figure 11). The density and altitude of the layer did not vary with SZA, season or solar irradiance, but did vary with longitude in a similar way to the main and the secondary layers. These topside layers appear to be more frequent in the southern hemisphere. They proposed solar wind electron precipitation as a plausible formation mechanism. It is not clear if the differences between the characteristics of the topside layers observed by MARSIS and MGS are due to different characteristics of the instruments or reflect a real physical difference.

#### Ionospheric Composition and Energetics

After the Viking mission, almost 40 years passed until new measurements of ionospheric composition and energetics were made by different instruments on board the MAVEN mission. NGIMS has measured concentration profiles of more than 20 ions (Benna et al., 2015a), and found a significant dawn–dusk asymmetry. Withers et al. (2015c) compared the ion densities measured by MAVEN/NGIMS with those measured by the Viking missions. They found that the MAVEN measurements of O+ and O2+ densities in the upper ionosphere were significantly larger than the Viking ones. The peak O+ abundance measurement was twice that of the Vikings and its altitude about 60 km higher. They attributed the differences to a combination of seasonal and solar variability. Withers et al. (2015d) compared NGIMS observations of ion density profiles in the upper ionosphere to the predictions of different ionospheric models. The most successful models were those including plasma transport driven by interactions with the solar wind.

The electron temperature measured by LPW was reported by Ergun et al. (2015), confirming some of the features measured by Viking, such as the strong temperature increase from about 180 to 300 km. A functional fit to the measured profiles, of high interest for its implementation in models, was also obtained. Fowler et al. (2015) analyzed the first ever in-situ electron-density and temperature profiles measured in the nightside of the planet by LPW. Explanation of the electron-density profiles in the nightside below 200 km required an additional source of ionization, probably particle precipitation. The electron temperatures above 300 km were larger by a factor of about two than those at dayside.

#### The Encounter with Comet Siding Spring

On October 19, 2014 the comet C/2013 A1 Siding Spring made a remarkably close encounter with Mars, approaching to about 138,000 km, one-third of the Earth–Moon distance. This implies that Mars actually crossed the coma of the comet. Such a close encounter is a very rare event, predicted to occur only once in 100,000 years (Ye & Hui, 2014). We were lucky that at the time of the encounter a total of five spacecraft were operating in Martian orbit, enabling the study of the atmospheric effects of interaction with a comet.

Just one hour after the time of maximum predicted dust flux, the SHARAD radar on board the Mars Reconnaisance Orbiter found a strong increase in the TEC, of up to 1 order of magnitude with respect to normal conditions on the nightside, never seen before (Restano et al., 2015). While SHARAD observations provided no information on the altitude of the increased ionization, MARSIS/MEx observations taken few hours after the closest approach of the comet showed the presence of a transient ionization layer at altitudes between 60 and 100 km from the surface, both on the nightside and the dayside of the planet (Gurnett et al., 2015), probably due to the effects of dust ablation. MARSIS observations were obtained outside the region where dust impacts were predicted, implying that the observed ionization had to be transported from the impact region in about 10 hours. A later reanalysis of the MARSIS observations showed that the transient ionization layer lasted at least 19 hours on the nightside and 12 hours on the dayside (Venkateswara-Rao, Manasa Mohana, Jayaraman, & Rao, 2016). IUVS/MAVEN observed emissions from metallic species (Mg, Mg+, Fe and Fe+), not detected before the passage of the comet (Schneider et al., 2015), confirming predictions made before the encounter (Withers, 2014). The emission peaked at about 115 km, significantly higher than measured by MARSIS. IUVS/MAVEN measurements implied that between 3 and 16 tonnes of dust were deposited on the planet. Later, IUVS measurements were reanalyzed using an improved calibration, increasing the amount of dust deposited to 82 ± 25 tonnes (Crismani et al., 2018). It was also found that the meteor shower lasted less than 3 hours. The dust was deposited in a narrow vertical layer between 105 and 120 km, and rapidly redistributed by horizontal winds and vertical transport (Crismani et al., 2018). After the passage of the comet NGIMS/MAVEN also detected about ten metal ions not previously observed (Benna et al., 2015b). NGIMS observations were made above 185 km, and it was found that after the encounter the abundances of these ions decreased exponentially, but were still detectable after more than two sols.

While most of the studies discussed above focused on the effects of the cometary dust on the ionosphere, it was also predicted that the precipitation of pickup O+ ions originating from photodissociation of H2O molecules from the coma would produce an enhanced ionization peaking at about 110 km (Gronoff et al., 2014). Sánchez-Cano et al. (2018b) used observations by MAVEN and the High Energy Neutron Detector on Mars Oddyssey to confirm the precipitation of pickup O+ ions after the passage of the comet. Another particle shower about one day later was attributed to the impact from the dust tail of the comet.

While the combination of observations from different instruments and spacecraft and computational modeling provided an unprecedentedly detailed view of the planet/comet interaction, further work is needed to understand the altitude difference between MARSIS and IUVS observations, to study the transport of the deposited ions, or to separate the effects of the comet dust from that of particle precipitation.

### Conclusions

Our knowledge of the Martian ionosphere has greatly increased since the first Mariner 4 observations of 1964. Thirteen missions have produced observations of the ionosphere of Mars. In 2019, the ionosphere of Mars was routinely being measured by two different spacecraft, Mars Express, and MAVEN, which have produced a huge amount of data that will take years to be ingested. Before Mars Express our information about the ionosphere of Mars came from about 6,000 electron-density profiles, two ionospheric composition profiles and two ion and electron temperature profiles. After their years in Martian orbit, Mars Express and MAVEN have increased the observational record by several orders of magnitude. The improvement in computational models is also contributing to our understanding of the processes involved in shaping the Martian ionosphere.

However, there are many areas still in need of further study, and significant open questions remain. Some of them are listed here.

While the variability of the electron densities in the dayside ionosphere is relatively well understood, the composition of the ionosphere and its variability with season, latitude, local time, and other possible governing factors, is only beginning to be explored thanks to MAVEN data. Similarly, the variability of electron and ion temperatures still needs full characterization. Photochemical models need to ingest the information about ionospheric energetics and its variability, as electron and ion temperatures determine many ionospheric reaction rates. Most photochemical models use fixed values for these parameters, and the effects of their variability on the predicted ionospheric composition remains to be studied.

The nightside ionosphere is much less known than the dayside one. Systematic exploration with Mars Express and MAVEN has started to unveil some of its features. However, a comprehensive vision of the nightside ionosphere, in particular with computational models including all relevant processes, still needs to be achieved.

The ionosphere below the two well-known ionospheric peaks also needs further study. While it had traditionally been assumed that the sporadic low-altitude plasma layer present at about 80–90 km of altitude was of meteoric origin, MAVEN observations challenged this interpretation. A physical explanation for the presence of this layer consistent with the new observations is still sought.

Our knowledge of the effects of crustal fields over the ionosphere has also been defied by MAVEN observations. While the increased ionization over regions of strong crustal fields was thought to be associated with an increase in the electron temperature, MAVEN observations show, on the contrary, lower electron temperatures over crustal field regions.

We can conclude by remarking that the exploration of the ionosphere of Mars has experienced a golden era. The analysis of data being obtained by Mars Express and MAVEN will keep the Mars ionospheric community busy for years. Answers to most of the current unknowns will probably be obtained, and hopefully new mysteries will arise.