1. 1 SH43A-1153 A Collisional Exospheric Model for Mars: Implications for X-ray and ENA Imaging M. Holmström NASA GSFC Swedish Institute of Space Physics matsh@irf.se AGU Fall Meeting San Francisco December 8, 2005 H, O and CO 2

2. 2 Introduction When the solar wind encounters a non-magnetized planet with an atmosphere, e.g., Mars or Venus, there will be a region of interaction, where solar wind ions collide with neutrals in the planet's exosphere. Two of the processes taking place are: 1) The production of energetic neutral atoms (ENAs) by charge-exchange between a solar wind proton and an exospheric neutral 2) The production of soft X-rays by charge exchange (SWCX) between heavy, highly charged, ions in the solar wind and an exospheric neutral Simulated ENA and X-ray images

3. 3 Images of ENAs and SWCX X-rays can provide global, instantaneous, information on ion-fluxes and neutral densities in the interaction region. It is however not easy to extract this information from the measured line of sight integrals that are convolutions of the ion-fluxes and the neutral densities. We need to introduce models that reduce the complexity of the problem. At Mars, the hydrogen exosphere is enlarged due to the planet's low gravity, and thus provide a large interaction region, extending outward several planet radii. Traditionally, most of the modeling of the outer parts of Mars' exosphere has been using analytical, spherical symmetric, Chamberline profiles. Planetary exospheres are however not spherical symmetric to any good approximation, and asymetries at Mars observed by Mars Express (ENAs) and XMM (SWCX) could be due to assymetries in the exosphere. Here we investigate the assymetries in exospheric densities at Mars due to various factors, and discuss their impact on ENA and SWCX X-ray images.

4. 4 Methods Exospheric asymmetries can be due to many factors, e.g., non- uniform exobase densities and temperatures, photoionization, radiation pressure, charge exchange, recombination and planetary rotation. To account for all these effects numerical simulations are needed. Using Monte Carlo test particle simulations it is possible to account for the above effects (if ion distributions are assumed). Even though neutrals in the exospheres by definition do not collide often, collisions occur. Especially near the exobase the transition is gradual from collision dominated regions at lower heights (with Maxwellian velocity distributions) to essentially collisionless regions at greater heights. We present exospheric simulations that include collisions self consistently using the direct simulation Monte Carlo (DSMC) approach. The code is three dimensional, parallel and uses an adaptive grid, allowing many particles to be included in the simulations, leading to accurate results.

5. 5 The simulation software We have adapted a general three-dimensional hydrodynamic, and MHD, flow solver - the FLASH code from University of Chicago - to include a large number of particles. It is a parallel, portable, open source, code written in Fortran 90 and using MPI for communication. For the problem at hand no fluid is involved in the simulations, and the adaptive grid is only used to track particles belonging to the same cell for the DSMC collisions. However, the code is easily extendable to hybrid (ion particles-electron fluid) simulations, or full particle- in-cell simulations, storing fields on the grid. Two-dimensional slice of the three-dimensional mesh

6. 6 Simulation details Here we consider three species: H, O, and CO 2. H since it is the dominant specie in the upper exosphere, O and CO 2 since they are dominant at the exobase. The DSMC collisions are hard-sphere collisions with velocity dependent crossections. Unless otherwise noted, the simulation parameters used are: Exobase temperature = 200 K. Number densities at exobase, H = 4.2e11 m -3, O = 1.2e13 m -3, CO 2 = 2.3e13 m -3, exobase radius = 3580 km The simulations were executed at the High Performance Computing Center North (HPC2N) in Umeå, Sweden, on ● A 384 processor AMD Opteron 248 cluster (1.7 Tflops peak) with 4 GB memory each, and on ● A 240 processor AMD Athlon MP2000+ cluster (0.8 Tflops peak) with 500 MB each. The code scales well with the number of processors, making Giga-particle simulations within reach

7. 7 Fate of an H atom Charge Exchange Elastic Collisions Photoionization Neutrals can collide with Neutrals, Ions, Electrons and Photons Radiation pressure e - impact ionization 100+ days in the solar wind 100+ days 20 minutes between photon collisions (absorption - reradiation) Not considered With O, CO 2, and H

8. 8 Processor 1 1 2 Particles 3 1 2 Blocks 3 Processor 2 1 2 Particles 3 1 2 Blocks 3 Processor 3 1 2 Particles 3 1 2 Blocks 3 Processor 4 1 2 Particles 3 1 2 Blocks 3 Data Distribution Block cells

9. 9 Redistribute Particles Update Grid Compute Accelerations Sort Particles Collide Particles Move Particles Add/Delete Particles Particle Simulation Cycle FLASH Gravity, Radiation pressure,... High order symplectic time integrator Sources/Sinks Each Particle to the correct Block/Processors Refine or Coarsen Blocks. Redistribute Blocks Particle – Cell map DSMC 5) 6) 4) 1) 2) 3)

10. 10 Verification: Density Profiles Spherical symmetric exosphere

11. 11 Thermal Escape of Hydrogen Charge-exchange and collisions Collisions The Jeans escape for these exobase parameters is 2.06e+26 s -1 Collisions reduce this rate No Collisions

12. 12 satellite ballistic escaping Orbit Classifications [Hodges, JGR, 1994]

13. 13 Velocity Distribution of all exospheric neutral hydrogen for the case without, and with, collisions Satellite orbits No collisions Collisions

14. 14 MSO coordinates x y z v Mars y-axis at 90 deg. longitude Latitude Longitude The x-axis is toward the Sun. The z-axis is perpendicular to the planet's velocity, in the northern ecliptic hemisphere. The y-axis completes the right handed system.

15. 15 Non-Uniform Exobase CO 2 H from 400 km above the exobase and up A simplified non uniform exobase density and temperature that corresponds to summer solstice MSO longitude latitude plots (subsolar point at 0,0) of column densities 30 -30 -90090 0.5 3 8 130 200 240 Exobase temperature Relative exobase density

16. 16 View from the Sun-direction Log(Hydrogen column density) Note that the column density vary by an order of magnitude, for constant planetocentric distances, even away from the planet, in the non-uniform case. This would directly effect SWCX X-ray images Uniform exobaseNon-uniform exobase yy zz

17. 17 Exospheric H densities N Sun Non-uniform exobase [#/m 3 ]

18. 18 Solar Wind ENAs Here we investigate the H-ENAs that are created outside the bow shock by charge exchange between exospheric hydrogen and solar wind protons. Still, this simplified model can give us information on, e.g., ENA fluxes in the shadow of Mars since most of the ENAs produced behind the bow shock (BS) probably does not enter that region. Possible asymmetries in the ENA fluxes behind Mars has been observed by the NPI sensor on Mars Express. Hydrogen atoms outside the BS are converted to ENAs with a rate of 8.4e-8 s -1 (137 days). The new ENA is drawn from a solar wind Maxwell velocity distribution (400 km/s, 1.2e5 K) ENA production region

19. 19 x y x zy z Trajectories of H-ENAs with velocities of 200+ km/s The areas of increased flux seems related to the area of large exobase density and temperature in the Southern hemisphere (page 15) Increased ENA fluxes

20. 20 Conclusions ● Exobase density and temperature, and their spatial distribution, control the extended hydrogen exosphere at Mars ● Asymmetries in exobase parameters propagate to large heights ● The deviation from spherical symmetry in densities and column densities can be at least a factor of ten ● The asymmetries in the hydrogen exosphere will lead to detectable asymmetries in ENA and X-ray fluxes ● Observations of SWCX X-rays (by XMM) and ENAs (in the planet's shadow by NPI on Mars Express) show asymmetries. Further studies are needed to find out if they are caused by asymmetries in the exosphere