Solar control of CO2 + ultraviolet doublet emission on Mars

The CO2+ ultraviolet doublet (UVD) emission near 289 nm is an important feature of dayside airglow emission from planetary upper atmospheres. In this study, we analyzed the brightness profiles of CO2+ UVD emission on Mars by using the extensive observations made by the Imaging Ultraviolet Spectrograph on board the recent Mars Atmosphere and Volatile Evolution spacecraft. Strong solar cycle and solar zenith angle variations in peak emission intensity and altitude were revealed by the data: (1) Both the peak intensity and altitude increase with increasing solar activity, and (2) the peak intensity decreases, whereas the peak altitude increases, with increasing solar zenith angle. These observations can be favorably interpreted by the solar‐driven scenario combined with the fact that photoionization and photoelectron impact ionization are the two most important processes responsible for the production of excited‐state CO2+ and consequently the intensity of CO2+ UVD emission. Despite this, we propose that an extra driver, presumably related to the complicated variation in the background atmosphere, such as the occurrence of global dust storms, is required to fully interpret the observations. In general, our analysis suggests that the CO2+ UVD emission is a useful diagnostic of the variability of the dayside Martian atmosphere under the influences of both internal and external drivers.


Introduction
Airglow emission provides important information on the energy deposition in and chemistry of a planetary upper atmosphere (Slanger et al., 2008, and references therein). For Mars, a typical dayglow spectrum in the ultraviolet (UV) region includes the Cameron band, the fourth positive band, the ultraviolet doublet (UVD), the Fox-Duffendack-Barker band, and several distinctive emission lines from atomic , , and (Barth et al., 1971;Stewart, 1972;Stewart et al., 1972). More recently, additional faint emission features, such as the Vegard-Kaplan band, the γ band, and the first negative band, have been identified (Leblanc et al., 2006;Jain et al., 2015;Stevens et al., 2015Stevens et al., , 2019. Existing analyses suggest that these dayglow emission features are mostly produced by either photon or photoelectron impact excitation (Fox and Dalgarno, 1979;Fox, 1992).
Among the important emission features of a Martian dayglow spectrum, the UVD ( ) emission near 289 nm has captured extensive research interest over the past several decades. Since 1969, the UVD emission on Mars has been measured remotely by the UV spectrometer or spectrograph on board Mariner 6, 7, and 9 (Barth et al., 1969(Barth et al., , 1971Stewart, 1972;Stewart et al., 1972), the Spectroscopy for Investigation of Characteristics of the Atmosphere of Mars (SPICAM) on board the Mars Express (MEx; Leblanc et al., 2006), and the Imaging Ultraviolet Spectrograph (IUVS) on board the Mars Atmosphere and Volatile Evolution (MAVEN; Jain et al., 2015). This feature is mainly caused by the photoionization and photoelectron impact ionization of which produces excited-state (Fox and Dalgarno, 1979;Cox et al., 2010;Gérard et al., 2019). Fluorescent scattering of solar photons also contributes but is usually negligible at altitudes below 180 km (Fox and Dalgarno, 1979). Previous analyses of UVD emission have mainly focused on the observed brightness profiles and their variations. The average peak brightness measured by the MEx SPICAM (Leblanc et al., 2006) is ~30 kR (kilorayleigh) for the solar zenith angle (SZA) range from to . With the aid of the more recent MAVEN IUVS measurements, Jain et al. (2015) reported a peak brightness of 76 kR over a similar SZA range. The difference in peak intensity between the two works is mainly attributable to the solar cycle variation. The former is appropriate for a moderate solar activity condition with a solar radio index at 10.7 cm, denoted as F 10.7 , of 105 in solar flux units (SFU, ), and the latter is appropriate for a higher solar activity condition, with of ~160 SFU. Regarding the peak altitude of UVD emission, the MEx SPICAM observations indicate that it is typically at 120-130 km and increases with increasing SZA (Leblanc et al., 2006). However, the SZA trend in peak altitude has not been confirmed by the MAVEN IUVS observations .
The MAVEN spacecraft has been orbiting Mars since September 2014 , operating with a better trajectory design around the planet in terms of UV observations as compared with the MEx. It has an elliptical trajectory in a 4.5-hr orbit with an inclination of 75°. The apoapsis altitude is ~6,200 km and the periapsis altitude is 150-160 km during the nominal mission phase but could be as low as 120-130 km during isolated "deepdip" campaigns. The MAVEN IUVS instrument has collected thousands of UVD emission profiles in different Mars seasons and under different solar activity conditions, providing a more extensive data set complementary to the earlier Mariner and MEx observations. Furthermore, the Extreme Ultraviolet Monitor (EUVM) on board MAVEN is able to make simultaneous measurements of short-wave solar irradiance received directly at Mars, allowing a better understanding of the solar control of UVD emission.
The present study is aimed at a systematic investigation of UVD emission on Mars by using the MAVEN IUVS measurements. We describe in Section 2 the data set used. In Section 3, we present the observed solar cycle and SZA variations in both peak emission altitude and intensity. This is followed by possible interpretations in Section 4 and concluding remarks in Section 5.

Observations
The IUVS is one of the eight scientific instruments on board the MAVEN spacecraft, measuring the far UV airglow on Mars between 110 and 190 nm at ~0.6 nm resolution and the middle UV airglow between 180 and 340 nm at ~1.2 nm resolution (Mc-Clintock et al., 2015). The IUVS is mounted on an articulated payload platform, which directs the orientation of the instrument slit as it captures the spectra of the Martian atmosphere. The IUVS has three operational modes, limb scan, coronal scan, and disk scan, whose implementation depends on the orbital phase. Here we consider limb observations that cover the tangent point altitude range of 80-225 km. Limb scans are taken near the periapsis, with the slit orientation parallel to the surface. A scan mirror sweeps the slit up and down, allowing the IUVS to perform periapsis observations at different altitudes with ~5 km resolution. A maxim-um number of 12 limb scans are taken during a single orbit. The observed raw data counts are corrected for detector dark currents and then converted to physical brightness by using the sensitivity derived from UV bright stellar observations made during the MAVEN cruise phase. The middle UV and far UV systematic uncertainties estimated from stellar calibrations are 30% and 25%, respectively.
The IUVS instrument team provides three levels of data product on the National Aeronautics and Space Administration Planetary Data System: level 1A (raw data), level 1B (calibrated data), and level 1C (processed data). The processed data provide the brightness profiles for a total number of 28 individual emission lines, which are obtained through multiple linear regression fits of individual spectral components combined with the laboratory spectral data and the reflected solar spectrum background . In this study, we used the level 1C data files, which are tagged "periapse" with version tag V13_R01.
We include in this study 2,189 MAVEN orbits that cover the time period from March 2015, reflecting southern summer conditions in Martian year (MY) 32, to October 2018, also reflecting southern summer conditions but in MY 34. The IUVS limb scan observations are available during each of these orbits, providing a suffi- The profile closest to the median situation is marked by blue, and the second-order polynomial fitting to the observations within 20 km centered at the apparent emission peak is indicated by red. The solar zenith angle for each brightness profile is also indicated in the figure legend for reference.
ciently large sample to analyze the variations in UVD emission on Mars. For each orbit, we computed the respective median brightness profile with the criterion that at least 10 individual measurements have been recorded over the altitude range of 100-170 km and restricted to SZA below 75°. Several representative UVD brightness profiles are presented in Figure 1, obtained during MAVEN orbit number 2,983 appropriate for SZÃ 53°. The profile closest to the median situation is indicated by the blue line. Individual brightness profiles, whenever with tangent points penetrating down to sufficiently low altitudes, show peak emission at 120-130 km. Analogous to Gkouvelis et al. (2018), the peak altitude and intensity were estimated from the second-order polynomial fitting to individual measurements made within ~20 km centered at the apparent emission peak, as indicated by the red line in Figure 1. A similar procedure was applied to the entire data set. The typical uncertainty in peak altitude is ~1 km and the typical uncertainty in peak intensity is less than 5%, both small enough to allow their variations in the Martian upper atmosphere to be retrieved.
CO + 2 Figure 2 shows the distribution of the MAVEN IUVS observations used in this study, in terms of the SZA and integrated solar extreme ultraviolet (EUV) and X-ray flux. The SZA refers to the peak of the median brightness profile obtained for each orbit. For the integrated solar flux as a proxy of solar activity, we used the level 3 solar spectral model constructed from the Flare Irradiance Spectral Model-Mars (version 11) and calibrated it with the MAVEN EUVM band irradiance data (Eparvier et al., 2015;Thiemann et al., 2017). The integration was made over the wavelength range of 0.5-69 nm, where 69 nm corresponds to the minimum photon energy required to produce excited-state .

Variations in Ultraviolet Doublet Emission
For the MAVEN IUVS data set analyzed in this study, the mean peak intensity of UVD emission is 38 kR with a scattering of 34%, where the scattering is defined as the standard deviation in peak intensity divided by the median peak intensity. The maximum peak intensity is 95 kR in MY 32 and the minimum intensity is 12 kR in MY 34. The mean peak altitude is at 120 km with a scattering of ~7%. The highest and lowest peak altitudes are 146 km in MY 33 and 107 km in MY 34, respectively. These values suggest the presence of strong variations in the peak parameters of UVD emission, which we detail below. The SZA variations in the median peak intensity of UVD emission are presented in Figure 3, distinguishing between different solar activity conditions characterized by different ranges of the integrated solar flux over 0.5-69 nm. The vertical and horizontal bars indicate the standard deviations of peak intensity and SZA within each bin. Each bin contains at least 10 individual brightness profiles to ensure that the derived SZA variations are statistically robust. The figure demonstrates a clear trend of decreasing peak intensity with increasing SZA at all solar activity conditions. The observed dependence of the peak intensity on SZA could be reasonably described by a cosine function in the form of where is the subsolar peak intensity. The best-fit cosine functions are indicated by the dashed lines in Figure 3, predicting a subsolar peak intensity of 56 kR, 71 kR, and 102 kR, respectively, for the integrated solar flux below , between and , and above .  A comparison of the subsolar intensities quoted above demonstrates the presence of strong solar cycle variation, which is further displayed in Figure 4 by using the integrated solar flux over 0.5-69 nm as a proxy of solar activity. Different SZA ranges are shown separately: red for 0°-30°, blue for 30°-60°, and green for 60°-75°, respectively. The vertical and horizontal bars represent the standard deviations of peak intensity and solar flux within each bin. Each bin contains at least 18 individual brightness profiles to ensure that the derived solar cycle variations are statistically robust. The figure reveals a remarkable increase in peak intensity with increasing solar activity, which is persistent at all SZAs. Here we use the power law relation to describe the solar cycle variation in peak intensity, hereafter denoted as , which is written as where is the integrated solar flux over 0.5-69 nm, is the peak intensity at a reference solar flux of chosen to be , and is the power index to be constrained by the data. Such a power law relation is chosen to properly reflect the expected limiting behavior of zero peak intensity when the solar radiation is switched off. The best-fit relations are given by the dashed lines in the figure, demonstrating a comparable solar cycle variation with a common power index of ~0.77 independent of the SZA. The best-fit peak intensity for a reference solar flux of is 69 kR at , 48 kR at , and 30 kR at , respectively, suggesting a trend that is compatible with the SZA variation displayed in Figure 3.

Variations in the Ultraviolet Doublet Peak Altitude
This section is devoted to the solar control of the peak altitude of UVD emission. We show in Figure 5 the median peak altitude of UVD emission as a function of the SZA, obtained for different ranges of the integrated solar flux over 0.5-69 nm. The horizontal and vertical bars in the figure show the standard deviations of peak altitude and SZA in each bin. The figure reveals a nearly constant peak altitude at , accompanied by an appreciable increase with increasing SZA toward the terminator. Analogous to the SZA variation in peak electron altitude in the dayside Martian ionosphere (e.g., Morgan et al., 2008;Fox and Weber, 2012;Yao MJ et al., 2019), the SZA variation in the peak emission altitude can be approximately described by a logarithmic secant relation in the form of Z sub H 0.9 mW⋅m −2 0.9 1.3 mW⋅m −2 1.3 mW⋅m −2 H where denotes the subsolar peak altitude in kilometers and is a length scale to be constrained by the data. The best-fit subsolar peak altitudes are 115 km, 120 km, and 122 km, respectively, for the solar flux range of below , between and , and above . The length scale, , is estimated to be 6.5 km, 7.0 km, and 5.5 km, respectively, under these solar cycle conditions, suggesting that the SZA variation in the peak altitude tends to be less pronounced at high solar activity conditions.
We further show the observed solar cycle variation in peak altitude in Figure 6 where the integrated solar flux over 0.5-190 nm is used as a proxy of solar activity. The IUVS observations made at different SZA ranges are indicated separately in the figure, whereas the vertical and horizontal bars represent the standard deviations of peak altitude and solar flux within each bin. Figure 6 reveals a clear trend of increasing peak altitude with increasing sol-  ar flux at all SZAs, which can be empirically modeled with a linear relation in the form of where is the peak altitude in kilometers at a reference solar flux of chosen to be and is the linear slope to be constrained by the data. The best-fit reference peak altitude, , is 101 km, 103 km, and 112 km, respectively, for , , and , suggesting a trend that is consistent with the SZA variation depicted in Figure 5. We caution that here we use the solar flux integrated up to 190 nm (instead of 69 nm as before) as a proxy of solar activity, which is closely related to the physical interpretation of the observed variation (see Section 4). Without showing the details, we shall mention that when using instead of in Figure 6, the correlation between the solar flux and peak altitude is greatly reduced and no unambiguous solar cycle variation could be concluded. Finally, we emphasize that in obtaining the best-fit linear relations, those measurements made with are excluded. This is because under such a condition, the MAVEN IUVS was coincidently observing the Martian upper atmosphere during global dust storms (Fu MH et al., 2020), which would elevate the entire brightness profile, including the emission peak, to higher altitudes (see also Section 4).

Doublet Emission
In Section 3, we reported the solar cycle and SZA variations in both peak intensity and altitude of UVD emission with the aid of the extensive limb scan observations made by the MAVEN IUVS.
The solar cycle and SZA variations in peak intensity are clearly suggested by the data, showing that the peak intensity increases steadily with increasing solar EUV and X-ray flux and also increases systematically with decreasing SZA. In addition, the same variations in peak altitude are revealed by the data, characterized by an increase in peak altitude with both increasing solar activity and increasing SZA. Different empirical functional forms are adopted to describe the observed variations, as given by Equations (1) -(4).
The observations presented here are favorably compared with those presented in the early IUVS investigation of Jain et al. (2015), who reported a similar correlation between the peak intensity of UVD emission and the EUVM irradiance measured in the 17-22 nm band (see their Fig. 5), characterized by a linear correlation coefficient of 0.77. The SZA variation in peak altitude was not obtained by Jain et al. (2015), possibly due to the much smaller IUVS sample used in their study. Using the MEx SPICAM observations, Leblanc et al. (2006), however, were able to obtain a similar SZA variation in peak altitude as reported here.
The observations presented so far could be interpreted by the solar-driven scenario in that the variation in peak intensity is a natural result of enhanced photoionization and photoelectron impact ionization under high solar activity conditions, as these are the two processes mainly responsible for the production of excitedstate in the dayside Martian upper atmosphere (e.g., Fox and Dalgarno, 1979). The same scenario also accounts for the observed decrease in peak intensity toward the terminator following a simultaneous decrease in solar irradiance. The above line of reasoning is analogous to the interpretation of the well-known   (4)) obtained for different SZA ranges. solar cycle and SZA variations in peak electron density in the dayside Martian ionosphere using the idealized Chapman theory (e.g., Fox and Yeager, 2009). The SZA variation in peak emission altitude could be interpreted by the fact that the emission is peaked where the optical depth at relevant wavelengths reaches unity in the Martian upper atmosphere, also analogous to the interpretation of the established SZA variation in peak electron altitude according to the idealized Chapman theory (e.g., Fox and Weber, 2012). Finally, the solar cycle variation in peak altitude is driven by the expansion of the atmosphere with increasing solar activity, similar to the observed solar cycle variation in exobase altitude at Mars (Fu MH et al., 2020).
Even though the solar-driven scenario successfully interprets the general characteristics of the solar cycle and SZA variations presented so far, we caution that such a simple scenario is not able to account for all the details, as indicated by the considerable scattering of individual IUVS observations, even at a fixed SZA and a fixed solar flux (not shown here). Clearly, an extra driver is required, presumably related to the variation in the background atmosphere, which is far more complicated than the simple effect of atmospheric expansion considered above. Variations driven by thermospheric global circulation (e.g., Bougher et al., 2006Bougher et al., , 2009González-Galindo et al., 2018), wave activity (e.g., England et al., 2017;Siddle et al., 2019), and global dust storms (e.g., Strausberg et al., 2005;Kass et al., 2016;Wu ZP et al., 2020) should all leave noticeable signatures in the observed variations in UVD emission. In particular, we speculate that the abnormal elevation of the peak emission altitude at the maximum available solar flux, as displayed in Figure 6, is more likely driven by global dust storms rather than thermospheric expansion, but such a speculation needs to be verified by detailed model calculations, which are beyond the scope of the present study.
Existing studies have indeed demonstrated clearly that the brightness profiles of UVD emission contain important information on the structure and dynamics of the Martian upper atmosphere. For instance, Cox et al. (2010) predicted an increase in the emission peak altitude from northern summer to northern winter, which they interpreted as a consequence of the thermospheric density variation driven by global circulation. On the basis of the observed scale heights of the topside brightness profiles, Leblanc et al. (2006) derived an average temperature of 191 K at 130-170 km in the Martian thermosphere, and Bougher et al. (2017) further reported the seasonal and solar cycle variations in thermospheric temperature. Lo et al. (2015) estimated the density structure from the UVD observations, revealing clear tidal structures between 100 and 190 km. Finally, the Martian thermospheric response to solar flares was recently reported by  based on the short-term variation in the same emission feature. Combining the IUVS observations and those made by other instruments on board MAVEN, such as the Neutral Gas and Ion Mass Spectrometer, would better elucidate the role of atmospheric variability in controlling UVD emission, which we leave for follow-up studies.

Conclusions
Airglow emission provides important information on the structur- al variability of a planetary upper atmosphere under both external and internal drivers (Slanger et al., 2008, and references therein). This study is dedicated to a systematic investigation of the solar cycle and SZA variations in UVD emission at Mars, a distinctive feature in a typical Martian dayglow spectrum that has been extensively studied over the past several decades (e.g., Barth et al., 1969Barth et al., , 1971Stewart, 1972;Stewart et al., 1972;Leblanc et al., 2006;Cox et al., 2010;Jain et al., 2015Gérard et al., 2019). Existing studies have established that this emission feature is mainly produced via photoionization and photoelectron impact ionization of atmospheric (e.g., Fox and Dalgarno, 1979). For the purpose of this study, we analyzed a large number of UVD brightness profiles collected by the MAVEN IUVS instrument when operated in the limb scan mode (McClintock et al., 2015), from which the respective peak emission intensities and altitudes were derived.
The available data set suggests the presence of strong solar cycle and SZA variations in both parameters, which manifest as (1) a significant increase in both peak intensity and altitude with increasing solar activity, and (2) a decrease in peak intensity but an increase in peak altitude with increasing SZA. Our analysis generally confirms previous results obtained either from the MEx SPICAM observations (e.g., Leblanc et al., 2006) or from a smaller set of the MAVEN IUVS observations available at an earlier epoch (e.g., Jain et al., 2015).
The observed solar cycle and SZA variations can be favorably interpreted by the solar-driven scenario in that (1) enhanced solar activity leads to enhanced production of excited-state via photoionization and photoelectron impact ionization, as well as an expansion of the Martian upper atmosphere via solar heating; and (2) regions at a larger SZA feel a lower solar irradiance along a more slanted line of sight, which not only reduces the production of excited-state but also elevates the location of unit optical depth. However, we note from the available IUVS observations that the solar-driven scenario is not able to account for all the details and an extra source of variability is required, presumably related to the complicated variations in the background atmosphere driven by global circulation, wave activity, and global dust storms, among others.