A characteristic heavy-ion signature observed in the vicinity of the Martian ionosphere during passages of Corotating Interaction Region (CIR) structures in solar wind is reported. We analyzed data obtained by the IMA/ASPERA-3 onboard the Mars Express (MEX) from September to October 2007. We compared the solar wind velocity at Mars derived from a shifted Maxwellian fitting to the IMA data with time-shifted Advanced Composition Explorer satellite data taken at ∼1 AU to the Martian orbit. Using the derived solar wind velocity, we identified four CIR structures passing through Mars quasiperiodically. Coinciding with the CIR passages, the IMA observed heavy-ion flux enhancement in the vicinity of the Martian ionosphere. The heavy-ion energies reach ≥100 eV and sometimes up to approximately several kiloelectron volts. Observed ion velocity distribution functions show that they are mainly precipitating toward the Martian ionosphere. The flux of the precipitating ions is typically 105–106 (104–105) cm−2 s−1 for the energy range of 50–500 eV (≥500 eV) and it becomes by one order of magnitude higher in one event. While the flux level is consistent with a previous model prediction of sputtering ions, the intermittent occurrence of the heavy-ion precipitation differs from conventional expectation of constant precipitation. These results suggest that the efficiency of the sputtering process in the Martian atmospheric escape is highly variable with dynamic solar wind variations.
 The solar wind can deeply penetrate into the Martian upper atmosphere because Mars does not possess a global intrinsic magnetic field [Acuña et al., 1998]. Outflows of planetary ions from the Martian upper atmosphere by solar-wind-induced processes have been considered to be important for understanding of the evolution of the Martian atmosphere [Jakosky and Phillips, 2001]. On the basis of plasma observations by the Phobos-2 and recently by the Mars Express (MEX) missions, outflows of planetary ions from the Martian upper atmosphere have been reported [e.g., Lundin et al., 1989, 2004; Carlsson et al., 2006; Barabash et al., 2007]. The Phobos-2 observation was carried out at high solar activity, and measurements of escaping ions by Phobos-2 led to an estimated O+ loss rate of 2–3 × 1025 s−1 with energies from 0.5 eV to 24 keV ions [Lundin et al., 1989]. On the other hand, averaged escape rates of planetary ions by using the ASPERA-3 ion mass analyzer (IMA) instrument onboard MEX at low solar activity turn out to be much lower, that is, escape rates of O+, O2+, and CO2+ are estimated at 1.6 × 1023, 1.5 × 1023 s−1, and 8.0 × 1022 s−1, respectively with energies from 30 eV to 30 keV [Barabash et al., 2007]. After June 2007, a new energy table for IMA was uploaded to observe the low-energy (∼10 eV) ions. According to Lundin et al. , they estimated the heavy-ion escape rate of 3.3 × 1024 s−1 with energies from 10 eV to 20 keV. Those studies thus suggested that the estimated total escape fluxes from the Martian upper atmosphere differ substantially between the solar minimum and maximum periods. There exist several candidates for the solar-wind-induced escape processes, for example, ion pick-up, sputtering, and the ionospheric outflow [e.g., Chassefière and Leblanc, 2004]. The escape process responsible for the difference between the solar wind conditions is poorly understood.
 Computer simulations can be used as a complement method to in situ observations to understand the interaction between the solar wind and the Martian upper atmosphere. A number of numerical simulation models, including MHD, multifluid, hybrid approaches, and combinations of MHD and test particle simulation [e.g., Ma et al., 2004; Terada et al., 2009; Chaufray et al., 2007; Fang et al., 2008], have been used to estimate escape fluxes for several solar wind induced processes such as ion pickup, sputtering, and viscous interaction under several solar wind conditions. For example, Brain et al.  reviewed numerical models of the global plasma environment on Mars and compared the results from different Martian global plasma interaction models under the same solar wind condition. However, most of the previous modeling studies investigated typical solar wind conditions on Mars, and the response of each escape process to the dynamic solar wind change such as the interplanetary shock arrival is still far from being understood.
 Corotating interaction regions (CIRs) are formed at the interface of slow- and high-speed solar wind streams. When a CIR passes through a magnetized planet like Earth, it is observationally known to cause geoeffective activities, such as magnetic storms, as a consequence of the interaction with the Earth's magnetosphere [e.g., Tsurutani et al., 2006; Richardson et al., 2006]. Coronal mass ejections (CMEs) are also remarkable interplanetary disturbances in the solar wind, and it is known to affect critically to the geospace environment. If these interplanetary disturbances hit an unmagnetized planet like Mars or Venus, the solar wind can directly interact with its upper atmosphere due to the lack of a global magnetic field. Hence it is potentially expected that interplanetary disturbances such as CIRs, CMEs, and associated high-energy particles may result in more drastic dynamic change in terms of the Martian atmospheric erosion than that of Earth [Futaana et al., 2008]. In addition, Mars has a higher chance of encountering CIR shock structures than Venus, because the CIRs are developed well beyond 1 AU from the Sun [e.g., Gosling and Pizzo, 1999].
 The plasma interaction between the unmagnetized planets and interplanetary disturbances in the solar wind has been an important issue to understand atmospheric escape from Mars. Dubinin et al.  pointed out that when a CIR arrived on Mars, a scavenging of the ionosphere occurred, and the estimated outflow rate was increased by a factor of ∼10. Luhmann et al.  showed the planetary ion escape enhancements on Venus related to the several CMEs or ICMEs. The cause of the escape enhancement accompanied by ICMEs, however, could be complicated as inferred from the recent Venus Express (VEX) ASPERA-4 observational data [Luhmann et al., 2008]. According to a latest statistical study by Edberg et al.  using MEX ASPERA-3 IMA data during the solar minimum, the outflow rate of heavy ions from the Martian upper atmosphere during CIR passage, which is accompanied by pressure pulses in the solar wind, becomes about 2.5 times larger than usual. These studies focused on the change of the outflow flux, and the physical process responsible for the heavy-ion escape enhancement by CIRs is yet an open question.
 In this paper, we report on heavy-ion signatures observed in the vicinity of the Martian ionosphere during passages of CIRs on the basis of analysis of MEX ASPERA-3 IMA data. In section 2, we introduce MEX ASPERA-3 IMA instrumentation and a derivation method of the solar wind velocity from IMA data. In section 3, we show the IMA observations of heavy-ion enhancement in the vicinity of the Martian ionosphere during a CIR passage on 19 September 2007. In section 4 we report on correlation between the heavy-ion enhancement signatures and CIR structure passages in the solar wind using MEX observations for about 1 month when the angle between Mars and Earth as viewed from the Sun is small (∼42°). Finally in section 5, we briefly summarize the observation results and discuss a possible scenario to explain the observations.
2. Instrumentation/Data Analysis
 Mars Express (MEX) was launched in 2003 and went into a quasipolar orbit around Mars with a pericenter (apocenter) altitude of about 275 km (∼10,000 km) [Chicarro et al., 2004]. The Analyzer of Space Plasma and Energetic Atoms (ASPERA-3) onboard MEX has carried out the plasma observations since then and provided the in situ plasma data around Mars during a solar minimum period. The highly ecliptic orbit of MEX allows us to observe various plasma domains around Mars (i.e., the solar wind, magnetosheath, the magnetic pile-up region down to the vicinity of the Martian ionosphere). The ASPERA-3 instrument is a comprehensive plasma package capable of measuring ENAs, electrons, and ions [Barabash et al., 2006]. ASPERA-3 consists of two units, the main unit (MU) and the ion mass analyzer (IMA). The MU comprises three sensors, neutral particle imager (NPI), neutral particle detector (NPD), electron spectrometer (ELS). In this study we use data from IMA. The IMA instrument determines the composition, energy, and angular distribution of ions in the energy range from ∼10 eV to 30 keV per charge. The IMA instrument can detect ion species with a mass per charge ratio of up to ∼80, and thus it can distinguish ion species such as He2+ for the solar wind and O+, O2+, and CO2+ for the planetary ions. The time resolution to complete a 3D full spectrum scan is 192 s, and it has a field of view (FOV) of 90° × 360°. Electrostatic sweeping performs elevation (±45°) coverage. In addition, the measurements of the low-energy range (≤50 eV) are carried out without the elevation scanning, and as a result, an FOV is 4° × 360° in this low-energy range (see Barabash et al.  for details of ASPERA-3's general performance).
 We analyze all MEX ASPERA-3 IMA data from 17 September to 11 October 2007, and we also use the solar wind key parameters obtained from the Advanced Composition Explorer (ACE) satellite considering the travel time of the solar wind from Earth to the Martian orbit. During this period, the angle range between Mars and Earth as viewed from the Sun was about from 46.5° to 36.5° and expected to be immersed approximately in the same expanding solar wind plasma on Mars.
 In order to calculate the solar wind velocity from the IMA data, we carried out the shifted Maxwellian fitting to each IMA velocity distribution obtained in one full 3D scan (i.e., every 192 s), as described below. First, assuming that the main ion component observed in the solar wind region consists of protons and has one-dimensional shifted Maxwellian distribution for the magnitude of velocity, the thermal velocity is much less than the bulk velocity. A shifted Maxwellian distribution used as the fitting function can be written as follows:
where v represents the magnitude of velocity in the phase space. Both A1 and A2 are coefficients related to solar wind density and temperature and vi is the solar wind velocity to be obtained. Then using the equation of (21) in subsection 3.2 of Fränz et al. , the observed ion distribution function to be fitted can be constructed as:
where C, G, and τ represent the detector counts integrated over mass channels as well as azimuthal and elevation channels, the geometric factor of the detector, and the acquisition time, respectively. ΔE(=En+1 − En) indicates the energy width of the n-th energy channel. In order to focus on the main component in the distribution function, the data with ≤2 counts are not used. The nonlinear least-squares method and the Newton-Raphson method are used in the fitting.
3. The 19 September 2007 Event
 In this section, we report on a typical CIR passage event observed on 19 September 2007. Figure 1a shows a typical example of IMA observation during a quiet solar wind condition during the period of interest in this study. As shown in the energy-time spectrogram, the MEX spacecraft first passed through the solar wind region and subsequently crossed the Martian bow shock (BS) at ∼03:32 universal time (UT) to the magnetosheath and then the magnetic pileup boundary (MPB) at ∼04:35 UT in the nightside. In this paper, MPB is defined as where the count rate of solar wind plasma decreases sharply [Nagy et al., 2004] because of the absence of the magnetometer onboard MEX, although MPB is generally defined from the magnetic field measurement. After the spacecraft traverses the vicinity of the Martian ionosphere corresponding to the pericenter (the magenta vertical line), the spacecraft again traverses MPB, the dayside magnetosheath, and BS again in the outbound path.
 In the following orbit 4767 shown in Figure 1b, we can identify an additional signature in the energy-time spectrogram ∼10:55 UT on 19 September 2007, which shows the abrupt discontinuous increases of both the average ion energy and temperature. As shown in the closeup figure around the signature (Figure 2), the ion velocity calculated by the shifted Maxwellian fitting described in section 2 (bottom panel) shows a shocklike increase. As reported in section 4, using three kinds of the solar wind data at the Martian orbit, it turns out that the discontinuity is observed when MEX encountered a CIR structure of the solar wind. Due to the CIR structure passage, the ion flux observed by IMA increased in the following orbit 4768 shown in Figure 1c. Another remarkable feature in Figure 1b is that the ion flux enhancement attained to a few keV is detected at ∼11:53 UT, when MEX passed the vicinity of the Martian ionosphere just after the CIR structure arrival. The ion flux enhancement was not detected in the previous orbit 4766 (Figure 1a).
Figure 3 displays the MEX orbit 4767 plotted in the cylindrical XMSO, coordinates. The MSO coordinates represent the Mars-centered Solar Orbital coordinates, in which the XMSO axis points toward the Sun, the ZMSO axis is perpendicular to the planetary revolution velocity and points to the northern ecliptic hemisphere, and the YMSO completes a right-hand system. The pericenter of this orbit is located at the altitude of ∼300 km near the southern terminator region. The comparison between the model of the MPB location [Edberg et al., 2008] (dashed lines) and the actual location of MPB (one of arrows) suggests that the induced magnetosphere around the Mars was significantly compressed after the CIR passage.
 As clearly shown in the mass-energy matrix obtained at the time when heavy-ion flux was found in orbit 4767 (Figure 4), the ion enhancement signature up to a few keV observed near the pericenter is mostly composed of the planetary heavy ions. The ions counted in the high-energy range above 1 keV mainly consist of O+, while the ions in the low-energy range (<100 eV) include the group of molecular ions such as O2+ and CO2+ in addition to O+. Since the distributions of O+, O2+, and CO2+ components in the IMA mass spectrum are expected partially overlap to each other, it is thus difficult to distinguish these planetary heavy ions clearly.
Figure 5 shows the cut of the three-dimensional distribution function of the heavy ions in the MEX spacecraft (SC) coordinates [Barabash et al., 2006], which is obtained by integration over all mass channels corresponding to ≥8 amu, for the same heavy-ion enhancement event. Both Figures 5a and 5b depict the SC coordinates, and the notations of both “Mars” and “Sun” show the looking directions of Sun and Mars, respectively. It is clear from the figure that the ions observed in the energy range of ≥100 eV are predominantly precipitating to the Martian upper atmosphere (from Figure 5a) with some antisunward velocity component (from Figure 5b). On the other hand, the low-energy component (a few tens of electronvolts) is mainly flowing upward from Mars. Therefore, the observation revealed that heavy-ion fluxes precipitating to the Martian upper atmosphere are enhanced during the CIR passage. However, a part of the IMA field-of-view, the elevation angle below ∼−2° and azimuth sectors mainly from 9 to 15 or the elevation above ∼8° and the azimuth from 11 to 13 is significantly obstructed by the spacecraft body or a solar panel. Therefore, ion fluxes observed in these directions could be low due to artificial reasons. In this specific case near the pericenter, the obstructed directions correspond to mostly the upward half of the distribution function. Thus we may miss the upward ions, if any, but the detection of the downward precipitation ions are not affected by the field-of-view obstruction.
4. Solar Wind Conditions and Other Events
 In this section, we investigate the relationship between the heavy-ion enhancement signature reported in section 3 and CIR passages in the solar wind using MEX data from September to October 2007. Figure 6a shows the comparison of the solar wind velocity derived from two different methods: A shifted Maxwellian fitting of MEX ASPERA-3 IMA data described in section 2 (solid circles) and time-shifted ACE satellite data from Earth to the Martian orbit (solid line). Here the time shift of the ACE satellite data (111.056 h) is determined from a combination of the traveling time estimation between ACE and MEX and adjustment by a few hours from the timing of the CIR-shock arrival between ACE and MEX data. This method of the constant time shifting might cause an overestimation during the latter of time period, because Mars gradually gets closer to Earth in this time period by ∼11 h or so [Vennerstrom et al., 2003]. However, the time deviation is not large compared to the MEX's orbital period (∼6.7 h). The solar wind density (Figure 6b) and magnetic field strength (6c) observed by ACE at ∼1 AU are also time shifted without compensating about the spatial development from Earth to Mars to infer the solar wind conditions of the Martian orbit.
 As shown in Figure 6a, the solar wind velocities obtained from MEX and ACE data have similar time variation, and both spacecraft observed the CIR-like velocity variation (i.e., the change from slow to fast solar wind flows) four times during the period of interest. Thus we concluded that four CIR structures passed by Mars quasiperiodically.
Figure 6d shows the flux of precipitating heavy ions observed by MEX near the pericenter of each orbit. In this paper, we investigated the observation only around the pericenter so as to avoid the solar wind proton contamination [Fränz et al., 2006] and effects of the spacecraft attitude change. However, the heavy-ion precipitations are sometimes observed continuously from the magnetic pile-up region. The precipitating flux is derived by the momentum calculation described in section 5 of Fränz et al.  after single-count data are excluded. In order to exclude the noise contamination, only the data with more than 3 counts at least at the peak of the ion distribution are shown. The obstruction areas of the IMA field-of view are located mostly in the upward half of the field-of-view near the pericenter as mentioned in section 3. Thus the effect of the obstruction is negligible in calculation of the downward (precipitation) flux. The open circles correspond to the high-energy (≥500 eV) ions, and the solid circles correspond to the low-energy (50 ≤ E < 500 eV) ions. The division of the energy range at 500 eV is adopted to compare the result with a previous model calculation by Chaufray et al. . The value of 50 eV corresponds to the lower limit of the IMA observations with the elevation scanning. It should be noted that the level of background noise counts is not always uniform, and therefore, the low precipitation flux (about ≤105 cm−2 s−1) can sometimes be affected by noise counts.
 While the MEX had reasonably good data coverage in the vicinity of the ionosphere (at the pericenters) as shown in Figure 6e, the ion precipitation is detected only in the limited time intervals (Figure 6d). A remarkable feature here is that the most of the precipitating heavy-ion detections coincide with passage of the four CIR structures identified from the solar wind data. As shown in Figures 6b and 6c, the precipitation flux of ≥105 cm−2 s−1 events tend to be observed during the period of the high magnetic field strength rather than during a high-density period. The 19 September 2007 event reported in section 3 corresponds to the left-most vertical dashed line marked as orbit 4767. It should be noted that the ion flux in the low-energy range (10 ≤ E < 50 eV) in which the ASPERA-3 IMA operates with a nonelevation scanning mode (not shown) indicates that the upward flux is more frequently observed near the pericenter locations and, its correlation with the solar wind conditions is less clear.
Figure 7 shows energy-time spectrograms of other precipitating heavy-ion events indicated by the rest of vertical dashed lines shown in Figure 6. In each panel, a heavy-ion enhancement signature can be identified near the pericenter location (vertical magenta line). While the CIR shocklike structure was not observed in these events, the high temperature in the solar wind region compared to the quiet period suggests that CIRs already arrived at Mars before these events. During the period shown in Figure 6, MEX's pericenter is located in Southern Hemisphere, and the four precipitation events shown in Figure 1b and 7 are observed near the magnetic anomaly region. However, a statistical survey for longer period including Northern Hemisphere orbits indicates that the magnetic anomaly is not a necessary condition for the heavy-ion precipitation during CIRs.
5. Summary and Discussions
 In this paper, we reported a characteristic heavy-ion enhancement signature observed in the vicinity of the Martian ionosphere during passages of CIR structures in the solar wind. The ion distribution function of heavy ions during the events shows that the heavy-ion flux is predominantly downward in the energy range of ≥100 eV (i.e., precipitating to the Martian upper atmosphere). These precipitating ions are not always observed but they are often observed only during a CIR structure passage through Mars. The flux of the precipitating ions is typically 105–106 (104–105) cm−2 s−1 for the energy range of 50–500 eV (≥500 eV), and it became one order of magnitude higher during the 19 September 2007 event.
 In the four events reported in this paper, the heavy-ion precipitation events in which we can get relatively good time adjustment for the solar wind based on the direct observation of the CIR shocklike structure by MEX ASPERA-3/IMA, there is a tendency that high heavy-ion precipitation fluxes (≥105 cm−2 s−1) are observed during the high magnetic field period rather than the high-density period of the solar wind as shown in Figure 6. A possible scenario to explain the intermittent appearance of the precipitating planetary heavy ions might be as follows: The Martian upper atmospheric ions are produced in the dayside region by photoionization, impact with solar wind electrons or charge exchange reaction with solar wind ions [e.g., Shizgal and Arkos, 1996]. Once ionized, these planetary ions can be picked up in the solar wind and rapidly swept away from Mars mainly in the solar wind electric field direction, because the gyro radii of the picked-up heavy ions are typically much larger than those of the Martian size (the radius of ≃ 3379 km). However, when the strength of the interplanetary magnetic field is increased due to the compressed interplanetary magnetic field structure embedded in the CIR, the gyro radii of picked up planetary heavy ions become smaller and sometimes are comparable to the Martian size. During the period, picked-up heavy ions have more chance to precipitate into the Martian upper atmosphere.
Table 1 shows a summary of the typical gyro radius estimation for picked-up O+ ions before and after CIR passages. Each gyro radius is calculated by using the solar wind velocity derived from the MEX data and the interplanetary magnetic field strength B inferred from ACE data assuming that the falling factor is 1/r2 from Earth to the Martian orbit. Here we took into account the decrease of B from Earth to the Martian orbit; however, we cannot know the accurate falling factor, considering the time development of the CIR from terrestrial to the Martian orbit. Therefore, this may have caused an overestimation of the gyro radii. For example, if we do not assume that the falling factor is 1/r2 but constant over r, the gyro radii becomes about 2.3 times smaller than the listed values in Table 1. These values are closer to the Martian values. For calculation of gyro radius before a CIR, averages of the solar wind velocity and magnetic field strength are used. As for the values after a CIR passage, the peak value of B and simultaneous solar wind velocity in each CIR structure are used. The result indicates that the gyro radii of picked up O+ ions became smaller after the CIR arrival and became comparable to the Martian radius or diameter. In addition, the gyro radius estimated from the maximum energy of the precipitating heavy ions is smaller than the Martian radius for all events as shown in the right hand column of Table 1. This variation of the gyro radius of picked up ions before and after CIR passage is consistent with the above scenario.
Table 1. A Summary of Estimated Gyro Radius of Picked-up O+ Ions Before and After CIR Passages
Before CIR Shock
After CIR Shock
Gyro Radius (km) (Precipitate Ions)
Magnetic Field (nT)
Gyro Radius (km)
Magnetic Field (nT)
Gyro Radius (km)
 As mentioned above in section 1, Mars does not have a global intrinsic magnetic field. However, it has been known that Mars has the localized crustal magnetic field mainly in the Southern Hemisphere by Mars Global Surveyor magnetometer observation [e.g., Acuña et al., 1998; Connerney et al., 2005]. We conducted a statistical survey for MEX data from July 2007 to March 2008 in order to investigate the dependence between heavy-ion precipitations and the Martian magnetic anomaly. As a result of the statistical survey, we confirmed those events are distributed rather randomly regardless of the crustal magnetic anomaly.
 In the statistical survey, we adopted the standard time adjustment method [Vennerstrom et al., 2003; Edberg et al., 2010] for the solar wind data between ACE and MEX, in which the time of ACE data is shifted by considering the radial distance between the two planets, as well as on the difference in heliospheric longitude [Vennerstrom et al., 2003]. It turns out that neither the magnetic field strength nor the density (dynamic pressure) enhancements in the solar wind have clear correlation with the heavy-ion precipitation flux to Mars. It should be noted that in the statistical survey, accuracy of the time adjustment between MEX and ACE is not enough to investigate which solar wind structure, such as high Pdyn and |B|, inside a CIR is responsible for the heavy-ion precipitation enhancements. Another possibility to explain the poor correlation can be the magnetic field direction in the solar wind, since the direction of the convective electric field in the solar wind is known to cause the asymmetry for the pickup ions [Barabash et al., 2007]. Theoretically, either enhanced solar wind density, magnetic field strength, or both can cause an enhancement of the heavy-ion precipitation. For example, increase of Pdyn will raise the strength of the magnetic field pileup near the planet, and it may enhance the heavy-ion precipitation. The mechanism responsible for the ion precipitation enhancement during CIRs is thus an open question.
 The previous studies based on the MEX observations show that the enhancement of the antisunward escaping flux of heavy ions coincides with the CIRs arrival [e.g., Dubinin et al., 2009; Edberg et al., 2010]. In this paper, we focus only on the observations in the vicinity of the terminator region due to the orbit geometry. It should be noted that the heavy-ion precipitations are sometimes observed continuously from the magnetic pileup region. In the event reported by Dubinin et al.  around 10 February 2008, the MEX pericenter was located in the dayside (∼15 local time). At this location, we confirmed that the ion distribution function shows a drastic antisunward flow. This suggests that the ion distribution function during CIR passages depends on the observation location. A statistical study by Edberg et al.  showed that the tailward flux of O+ from the Martian upper atmosphere during CIR passages is about 2.5 times larger than usual on the basis of MEX ASPERA-3 observations. As shown in Figure 5, the tailward direction is often included in the precipitating sector in the ion distribution function in the vicinity of the pericenter for the period we investigated in this paper. Thus our results are consistent with those of the tailward O+ flux enhancement during CIR passages. On the one hand, our results also suggest that it is important to investigate three-dimensional velocity distribution function data to understand the meaning of the tailward flux enhancement.
 Finally, we compare the observed heavy-ion precipitation flux in the CIR events to a hybrid simulation result by Chaufray et al. . From Figure 5(a) and 5(c) of Chaufray et al.  that shows the precipitating O+ flux, which is thought to be responsible for the sputtering escape, the expected precipitation flux at the pericenter locations where MEX observed the heavy-ion enhancement is ∼105–106 (∼104–105) cm−2 s−1 in the energy range of 50–500 eV (≥500 eV). The simulation results agree well with the observed flux reported in Figure 6d on average. On the other hand, the important discrepancy between the model prediction and the observation is the occurrence frequency of these events. Observations reveal that this level of heavy-ion precipitation flux is operating only during the CIR structure passages, and the precipitation flux becomes one order of magnitude higher in one of four events reported, while the simulation results indicate that heavy-ion precipitation always exist even under typical moderate solar wind conditions. These results suggest that the efficiency of the sputtering process in the Martian atmospheric escape is highly variable with dynamic solar wind variations.
 The study also indicate that simultaneous observations of the solar wind and heavy-ion escape are necessary to understand detailed response of the atmospheric escape to the solar wind variation and suggests importance of the solar wind monitor for future Mars missions.
 This work is supported by a Type-II Research Grant of the Institute for Advanced Research, Nagoya University. We are grateful to all Mars Express ASPERA-3 science members for their collaboration. We also thank the ACE MAG and SWEPAM instrument teams and the ACE Science Center for providing the ACE data.
 Masaki Fujimoto thanks the reviewers for their assistance in evaluating this paper.