Origins of Europa Na cloud and torus

Transcription

1 Icarus 178 (2005) Origins of Europa Na cloud and torus F. Leblanc a,,a.e.potter b, R.M. Killen c, R.E. Johnson d a Service d Aéronomie du CNRS, Réduit de Verrières, BP3, Verrières Le Buisson cedex, France b National Solar Observatory, Tucson, AZ, USA c University of Maryland, College Park, MD, USA d University of Virginia, Charlottesville, VA, USA Received 15 September 2004; revised 4 March 2005 Available online 22 August 2005 Abstract Following the discovery observations by Brown and Hill [Brown, M.E., Hill, R.E., Discovery of an extended sodium atmosphere around Europa. Nature 380, ], only few new observations of Europa Na cloud and torus have been reported [Brown, M.E., Potassium in Europa s atmosphere. Icarus 151, ; Brown, M.E., The structure and variability of Europa s sodium atmosphere. Icarus. Submitted for publication]. Previous works [Johnson, R.E., Leblanc, F., Yakshinskiy, B.V., Madey, T.E., Energy distributions for desorption of sodium and potassium from ice: The Na/K ratio at Europa. Icarus 156, ; Leblanc, F., Johnson, R.E., Brown, M., Europa s sodium atmosphere: An ocean source? Icarus 159, ] have reproduced some of these observations and concluded that the origin of Europa s Na cloud is primarily sputtering of Europa s surface by jovian magnetospheric energetic particles. Leblanc, Johnson, and Brown [Leblanc, F., Johnson, R.E., Brown, M., Europa s sodium atmosphere: An ocean source? Icarus 159, ] suggested there was a correlation of the rapid variation of the total content of Europa s Na cloud with the variation of Europa s jovian centrifugal latitude. The contribution to the observed signal due to Na atoms ejected from Io and energetic Na ejected from Europa by atmospheric sputtering were also considered. In this work, we present improved results using a more accurate description of the variation of the magnetospheric incident flux, electron impact ionization and Io background with respect to Europa s and Io s positions in Jupiter s magnetosphere. We also describe the energetic component due to atmospheric sputtering. This model of the sodium cloud is constrained by new observations of Europa s sodium cloud close to Europa s surface by Potter and co-workers also presented here. Their high spectral resolution provides unprecedented detail of the velocity distribution of the Na atoms at different positions around Europa and for various Europa positions around Jupiter. This analysis confirms a number of earlier conclusions and highlights observed variations of Europa Na cloud with respect to Europa s local time within the jovian magnetosphere. These large observed variations require that the ballistic transport across Europa s surface must be taken into account Elsevier Inc. All rights reserved. Keywords: Europa; Satellites of Jupiter; Atmospheres; Magnetosphere; Io 1. Introduction Jupiter s moon Europa continues to be one of the most interesting objects in the Solar System because of its geologically young surface (e.g., Pappalardo et al., 1999), a possible salty subsurface ocean (e.g., Kivelson et al., 2000), * Corresponding author. Fax: +33 (0) address: francois.leblanc@aerov.jussieu.fr (F. Leblanc). and the existence of a tenuous atmosphere (e.g., Hall et al., 1995). In addition, Europa s surface and atmosphere are in a rough chemical steady state with the external radiation field (Carlson et al., 1999; Cooper et al., 2001; Paranicas et al., 2002). Therefore, studying the origin and evolution of its thin atmosphere can lead to an understanding of the surface composition. Since Europa s surface is young, this understanding could, in principal, lead to information on the subsurface material (Johnson et al., 1998) /$ see front matter 2005 Elsevier Inc. All rights reserved. doi: /j.icarus

2 368 F. Leblanc et al. / Icarus 178 (2005) As predicted from laboratory measurements (Johnson, 1990), the decomposition of water ice due to the chargedparticle irradiation of the surface produces an O 2 atmosphere with a surface density 10 8 cm 3 and a column density cm 2 (Johnson et al., 1982; Hall et al., 1995; Ip et al., 1998; Saur et al., 1998). A much thinner atmosphere of Na has also been measured (Brown and Hill, 1996), with a column density of sodium cm 2 and an even smaller potassium atmosphere (Brown, 2001). The observation of a potassium to sodium ratio different from that at Io (Brown, 2001) and recent analysis of the sodium observations (Johnson, 2000; Johnson et al., 2002; Leblanc et al., 2002) suggest that the observed sodium is not due to the implantation of sodium ions from the jovian plasma torus. Therefore, the observed sodium may be endogenic. With the upward revision of the Brown and Hill (1996) sodium column density by Brown (2004) due to a factor 2 error in the distance scale in Brown and Hill s (1996) paper, both the surface density and the escape flux analyzed by Johnson (2000) should be increased. This strengthens the argument for an endogenic source as shown in Leblanc et al. (2002). A 3D Monte Carlo model of the Europa Na cloud was developed in Leblanc et al. (2002) and compared to the observations later published by Brown (2004). These authors concluded that Europa permanently loses a significant quantity of Na. They also suggested that the Na cloud should vary significantly with respect to the distance of Europa from the Io centrifugal equator and that Io should also contribute to the Na emission at Europa. The energy distribution of the Na atoms ejected from Europa was considered and the possible presence of an energetic tail was suggested in order to reproduce the Brown observations. This work provides some new insights into the formation of the Europa Na cloud. With the help of new observations by Potter and co-workers, the model of Leblanc et al. (2002) has been improved. These observations give, for the first time, details on the velocity distribution of the Na atoms at different positions around Europa. They also provide information on the Na distribution significantly closer to Europa than previous observations. The improvements suggested by the new observations are also analyzed in the context of the Brown measurements already discussed in Leblanc et al. (2002). We first describe the new set of observations of Potter and co-workers. The geometry of observations is given for each date and some basic elements of discussion are extracted from these observations. In the second part, we describe the several improvements that are suggested by these observations which have been implemented in our model. In a third part, we compare simulations with Brown (2004) and Potter and co-workers observations respectively, and conclude, in a fourth part, on some new characteristics of Europa s Na cloud. The last section summarizes the principal and newly discovered characteristics of Europa Na cloud highlighted in this paper. 2. Observations Table 1 summarizes the geometry of the observations. The observations of Brown (2004) have been already discussed in Leblanc et al. (2002). They were obtained on 12/28/1999 at the Keck Telescope with a spectral resolving power (λ/ λ with λ the wavelength) of 60,000 with the facility echelle spectrograph HIRES. The new measurements presented here were obtained by A.E. Potter and co-workers in Europa sodium emission was observed using the McMath-Pierce Solar Telescope, which is a 3-mirror f/54 heliostat design with a 1.61 meter aperture, located at the Kitt Peak National Observatory, Tucson, Arizona. The spectra were recorded using a 4.5 meter Czery Turner spectrograph to yield spectra with a resolving power (λ/ λ) of about 150,000. Exposures of Europa were typically a half-hour, with the position of the satellite controlled by an auto-guider. Two sets of observations were usually combined to improve the signal-tonoise ratio, namely for 11/28/2000 two sets of 600 s and for 11/29 30/2000 two sets of 1800 s. Seeing (Full Width Half Maximum, FWHM, of a point source, such as a star, seen through the atmosphere due to atmospheric turbulence) was usually about 1.5 arc second over the period of observation, but cumulative guiding errors made the effective seeing approximately 2.0 or more arc seconds. The FWHM of the thorium emission lines in thorium-argon lamp calibration spectra are approximately 1.5 pixels, or 30 milli- Angstroms, corresponding to an uncertainty in velocity of the order of 0.8 km/s. The spectra were obtained using an image slicer with an aperture of 10 by 10. Images of the emission intensity in D2 are then extracted from 1 1 pixel chosen at different positions in each image. The Potter and co-workers observations were calibrated using the V magnitude of Europa (stellar magnitude in the V-Band centered at 4280 Å). The emission intensity was extracted from the observed combination of emission signal and reflection signal by fitting a Voigt profile to the Fraunhofer line at the peak of the Europa reflection (R.C. Woodward, private communication). Such a profile results from the convolution of the thermal Doppler broadening (Gaussian distribution) and the impact pressure broadening of a spectral line (Lorentz broadening). Because Europa is so reflective, there is no evidence of a sodium emission signal (the Fraunhofer line is perfectly symmetric). This fit of the Fraunhofer line is then scaled to another location where the atmospheric emission signal is evident and is used to subtract out the background signal. The signal-to-noise in the November 28 and 30 Europa data varied from about 3 to 8 and for the November 29 Europa data from about 5 to 20. Fig. 1 provides the geometry for each observation. It illustrates which part of the sodium cloud of Europa has been observed each time. We note that the geometry of Potter and co-workers measurements on 11/30/2000 (Fig. 1d) is close to that of Brown s measurement on 12/18/2000 (Fig. 1a) and corresponds to the observation of sodium above Europa s

3 Europa s Na cloud and torus 369 Table 1 Set of observations used in this work. 12/28/1999 observation has been obtained by Brown and 11/ /2000 observations by Potter and co-workers E = East, W = West, N = North, and S = South directions (see Fig. 1 and caption for the definitions of these directions) Date Time (UT) System III CML ( ) System III Europa ( ) Jovian local time angle of Europa ( ) Europa centrifugal latitude ( ) Jovian local time angle of Io ( ) Position of the slit/pixel and orientation 12/28/1999 4: Centered NS 5: : Centered EW 5: : North EW 6: : South EW 7: : North EW 7: : South EW 8: /28/2000 3: ,3 E, W, 4: N, S from center 11/29/2000 6: E, W, N, S 7: /30/2000 2: : and 3,4 E, W, N from center In the case of Brown, no data were obtained closer than 6 R E to Europa whereas in the case of Potter and co-workers, the closest observations are obtained at 3.5 R E from Europa center. Slit size: where 1 R E = 0.47 for Brown data. Pixel size: 1 1 with 1 R E = for Potter and co-workers data. Local time angle is defined in degrees starting from midnight Jupiter local time. Fig. 1. Europa s and Io s positions in their orbital plane. Panel (a): Brown observations on 12/28/1999 between 4.28 and 8.15 UT. Panels (b), (c), and (d): Potter and co-workers observations on 11/28/2000 at 3.44 UT, 11/29/2000 at 7:24 UT and 11/30/2000 at 3:14 UT, respectively. East and West directions (referred to East and West directions of the observer) used in the paper are indicated on the figure. North pole is towards the reader. The thick lines drawn in each panel display the lines of sight of each observation. The Sun and the Earth are placed in the negative x 0 direction. In this figure, the z 0 axis is away from the reader and corresponds to the South pole direction whereas the z axis is towards the reader in the North pole direction. leading hemisphere, since the trailing hemisphere is almost totally hidden by Europa s disk. The 11/28/2000 geometry corresponds to an observation of sodium over Europa s trailing hemisphere (Fig. 1b) whereas the 11/29/2000 geometry corresponds to an observation of the leading anti-jovian quadrant (Fig. 1c). Therefore, the new observations will allow us to explore the leading/trailing hemispheric asymmetry of the production of Europa s atmosphere (Johnson, 2000; Shematovich et al., 2004). We will first discuss the recent observations of Potter and co-workers obtained during three consecutive nights with the same instrument. Table 2 provides the total emission intensity observed at different positions (pixels) around Europa within the image slicer used during the three consecutive nights of observations (last column). As shown in Table 2, East/West differences of the emission intensity are well correlated to trailing/leading hemispheric differences (see 5th column) within 2 to 3 of Europa. It confirms that Europa s Na cloud is dominated by the production of atmospheric Na from the trailing hemisphere. This is consistent with the larger flux of energetic ions onto the trailing hemisphere as described in Popieszalska and Johnson (1989) causing enhanced ejection of material from the trailing hemisphere. This is also consistent with the higher concentrations of non-ice materials on this hemisphere, possibly including Na containing salts (McCord et al., 1999; Johnson et al., 2004). However, the differ-

4 370 F. Leblanc et al. / Icarus 178 (2005) Table 2 Potter and co-workers observations: pixel position in the observer frame (x 0,y 0,z 0 ) and in the (x,y,z) frame and total Na emission measured in each pixel (1 1 extracted from image slicer used by Potter and co-workers) Date 11/28/2000 Fig. 1b 11/29/2000 Fig. 1c 11/30/2000 Fig. 1d Pixel position (x 0,y 0,z 0 ) (x,y,z) coordinates of the point of the line of sight closest to Europa Remarks Measurement (kr) 2 North (0 R E,0R E,3.8R E ) 17.7 R E North of centrifugal equator 1.8 ± South (0 R E,0R E, 3.8 R E ) 11.1 R E North of centrifugal equator 1.9 ± East (2.8 R E,2.8R E,0R E ) Sub-jovian leading hemisphere partly hidden by Europa 1.1 ± West ( 2.8 R E, 2.8 R E,0R E ) Anti-jovian trailing hemisphere on the observer s side 2.1 ± North (0 R E,0R E,5.5R E ) 20.4 R E North of centrifugal equator 1.1 ± South (0 R E,0R E, 5.5 R E ) 9.4R E North of centrifugal equator 0.9 ± East (4.1 R E,4.1R E,0R E ) Sub-jovian leading hemisphere partly hidden by Europa 1.1 ± West ( 4.1 R E, 4.1 R E,0R E ) Anti-jovian trailing hemisphere on the observer s side and away from Jupiter 1.1 ± North (0 R E,0R E,3.8R E ) 24.5 R E South of centrifugal equator 5.2 ± South (0 R E,0R E, 3.8 R E ) 32.1 R E South of centrifugal equator 4.2 ± East (3 R E,1.5R E,0R E ) Anti-jovian quadrant of the leading hemisphere 4.9 ± West ( 3 R E,1.5R E,0R E ) Sub-jovian quadrant of the trailing hemisphere 5.2 ± North (0 R E,0R E,5.5R E ) 22.8 R E South of centrifugal equator 2.4 ± East (5 R E,1.5R E,0R E ) Anti-jovian quadrant of the leading hemisphere 3.4 ± West ( 5 R E,1.5R E,0R E ) Sub-jovian quadrant of the trailing hemisphere 4.4 ± North (0 R E,0R E,7.6R E ) 20.7 R E South of centrifugal equator 2.3 ± East (7 R E,1.5R E,0R E ) Anti-jovian quadrant of the leading hemisphere 1.5 ± West ( 7 R E,1.5R E,0R E ) Sub-jovian quadrant of the trailing hemisphere 1.5 ± North (0 R E,0R E,3.8R E ) 40.9 R E South of centrifugal equator 0.9 ± South (0 R E,0R E, 3.8 R E ) 48.5 R E South of centrifugal equator 1.2 ± East ( 2.8 R E,2.8R E,0R E ) Anti-jovian trailing hemisphere partly hidden by Europa 1.4 ± West (2.5 R E, 3 R E,0R E ) Sub-jovian leading hemisphere on the observer s side 1.1 ± North (0 R E,0R E,5.5R E ) 39.2 R E South of centrifugal equator 0.5 ± East ( 4.8 R E,4.8R E,0R E ) Anti-jovian trailing hemisphere partly hidden by Europa 0.6 ± West (3 R E, 5.2 R E,0R E ) Sub-jovian leading hemisphere on the observer s side 0.5 ± North (0 R E,0R E,7.6R E ) 37.1 R E South of centrifugal equator 0.2 ± East ( 6.5 R E,6.5R E,0R E ) Anti-jovian trailing hemisphere partly hidden by Europa 0.2 ± West (5 R E, 7 R E,0R E ) Sub-jovian leading hemisphere on the observer s side 0.4 ± 0.1 The xyz coordinates correspond to a leading/trailing reference frame centered on Europa s disk where the x direction is along the trailing/leading direction (the x and y axis rotate with Europa around Jupiter). Positive x corresponds to leading phase whereas trailing phase corresponds to negative values. The xy directions are provided in Fig. 1 for each date. The last column provides the total emission intensity. The respective position of the pixel is described in columns 2 to 4. ence in the intensity of the emission between the East and West sides of Europa changes with the position of Europa around Jupiter. When the dayside corresponds to the trailing hemisphere, the ratio between East and West is the largest (factor 2 on 11/28/2000; Fig. 1b). When the dayside does not correspond to the trailing hemisphere (11/29/2000; Fig. 1c), or is opposite to the trailing hemisphere (11/30/2000; Fig. 1d), the ratio between East/West emissions is significantly smaller. Therefore, the new observations show that the content of the sodium cloud depends also on the solar flux (thermal and photon stimulated desorptions and photo-ionization). There are also North/South asymmetries correlated with the centrifugal latitude of Europa for 11/28/2000 and 11/29/2000 for the 2 observation but not for 11/30/2000. For 11/28/2000 and 11/29/2000, the largest North/South asymmetry is observed in the hemisphere of Europa which is closest to Io centrifugal equator. On 11/30/2000, the asymmetry is inverted but much smaller and the total emission is also smaller than during the two previous nights of observation. If the sodium cloud expands spherically as a purely escaping atmosphere and the ionization and collisions with the background atmosphere are negligible, the intensity of the line-of-sight emission should vary following a 1/r law, where r is the distance to Europa s center. That is, from 2 to 3 and from 3 and 4 from Europa center in Potter and co-workers observations, the total emission should decrease by a factor 1.5 and 1.33, respectively. From Table 2, the ratio between 2 and 3 emission intensities is between 1.6 and 2.1 on 11/28/2000 (except for the East axis), between 1.2 and 2.1 on 11/29/2000 and between 1.7 and 2.3 on 11/30/2000. The ratio between 3 and 4 is equal to values between 1.04 and 2.9 on 11/29/2000 and between 1.2 and 2.6 on 11/30/2000. Therefore, the average observed decrease of the emission intensity is larger than for a purely escaping atmosphere, implying a Na cloud composed mainly by atoms gravitationally bound on ballistic trajectories. Therefore most of the atoms released from Europa must have lower energies than the escape energy of a Na atom that is 0.5 ev at Europa s surface and even less than 0.4 ev, the en-

5 Europa s Na cloud and torus 371 ergy needed to reach 8 R E (4 for the Potter and co-workers observations). This is in good agreement with laboratory measurements of the energy distribution of sodium ejected from an icy surface (Johnson et al., 2002). It is also seen that the signal on 11/29/2000 (Fig. 1b) was two times larger than on 11/28/2000 (Fig. 1c) and four times larger than on 11/30/2000 (Fig. 1d). These could be explained by a more intense trailing hemisphere source and the differences in line of sight, illustrated by Fig. 1. The 11/29/2000 observation is the most favorable for observing the densest regions of the Na cloud if the main source is sputtering of the surface by energetic jovian particles centered on the trailing hemisphere. Similarly intensity on 11/30/2000 is the smallest since the trailing hemisphere of Europa was partly hidden by the satellite. On 11/28/2000, Europa s trailing hemisphere is within the field of view of the observer but, unlike on 11/29/2000, Europa s bright disk hides part of the cloud originating from the trailing hemisphere. It appears, therefore, that the principal variations within each set of observations and from one observation to another are due to the geometry of the observations. This implies a strong aspherical structure of the Na cloud with the dominant source centered on the trailing hemisphere. The length of each observation (60 min) does not allow us to constrain variations associated with the change of the Europa centrifugal latitude. The influence of the centrifugal equator position on the Na cloud was discussed in Leblanc et al. (2002) and will be only reanalyzed below in the frame of Brown (2004) observations. 3. Model The model used here is an improved version of that described in Leblanc et al. (2002). It is a 3D Monte Carlo test particle approach. That is, test particles are ejected from Europa s surface following an energy distribution constrained by laboratory measurements (Wiens et al., 1997; Johnson et al., 2002) and by comparison to observations. The surface distribution of the ejecta is dominated by a leading trailing asymmetry based on a model of the bombardment of Europa s surface by energetic jovian particles (Popieszalska and Johnson, 1989; Cooper et al., 2001; Paranicas et al., 2002). The trajectory of each test particle is calculated in gravitational fields of both Jupiter and Europa accounting for the electron impact ionization and photo-ionizations rates as well as the solar radiation pressure. The ejecta flux is assumed to vary with Europa s centrifugal latitude with a period of 10 hours, from Na/cm 2 /s when Europa is furthest from the centrifugal equator (centrifugal latitude equal to +/ 7 )to Na/cm 2 /s when Europa trajectory cuts the jovian centrifugal equator. This variation is due to the fact that the flux of energetic particles impacting onto Europa s surface changes with its centrifugal latitude. The range of variation is estimated roughly from the variation of the electron density computed by Smyth and Combi (1988) along the trajectory of Europa within the jovian magnetosphere. Between +/ 1 Jupiter radius above and below the centrifugal equator, the electron density varies from 30 to 70 electrons per cm 3. Larger variations are possible, up to a factor 8 according to a reanalysis of Voyager observations by Moncuquet et al. (2002). Moreover at Europa s orbit, since ions are probably more anisotropic than electrons, even larger latitudinal differences for the ion could be expected. The spatial distribution of the sodium source also changes with Europa s centrifugal latitude. It varies from a preferentially Northern hemisphere ejection when Europa is South of the jovian centrifugal equator to a preferentially Southern hemisphere ejection when Europa is North of the centrifugal equator. Such a North/South asymmetry is suggested by the Potter and co-workers observations and will be discussed below. The sodium ionization rate is driven by electron impact and, hence, also varies, with the centrifugal latitude and radial distance in the Jovian magnetosphere (Smyth and Combi, 1988). Sodium from Io can represent a significant part of the total Na emission at Europa (Leblanc et al., 2002). This background is estimated here from a separate simulation of Io s Na cloud. Following Smyth and Combi (1988, 1997) Na atoms are ejected from Io with a velocity distribution typical for atmospheric sputtering: f(v)= (1/v b ) (v/v b ) 3 ( vb/( 2 v 2 + vb 2 )) α ( 1 (( v 2 + vb) 2 ) /v 2 1/2 ) M, v b, v M, and α are defined in Smyth and Combi (1988, Appendix D) and are respectively related to the most probable speed of the flux distribution, to the maximum speed of the flux distribution and to the dispersion of the distribution. We use α = 7/3, v M = 57 km/s and v peak = 0.5km/s (Smyth and Combi, 1997) and a flux between Na/s and Na/s depending on the centrifugal latitude of Io (Wilson et al., 2002; Smyth and Combi, 1997). We do not consider molecular ion dissociation or charge exchange sources (Wilson et al., 2002) which produce fast sodium beyond the range of Doppler shift considered in this study (typically less than 0.5 Å centered on Europa Doppler shift). In addition, the Potter and co-workers observation have too small a signal/noise ratio beyond this range. This simulation provides a very rough estimate of the role of Io in the observations at Europa (e.g., Burger, 2003; Burger and Johnson, 2004). Column density and emission intensities are then calculated for the Europa observation geometries displayed Fig. 1. The simulations in Leblanc et al. (2002) were further improved in the following way: The phase angle of Europa is accurately taken into account for the column density integration, and the velocity of Europa with respect to the Earth is accurately

6 372 F. Leblanc et al. / Icarus 178 (2005) taken into account. We neglect the obliquity of Jupiter with respect to the ecliptic plane of the Earth. The simulation of Brown observations now exactly reproduces the conditions of each measurement (see Table 1). In particular, the variation of the centrifugal equatorduringthe4hofobservationisaccounted for as discussed in Leblanc et al. (2002). The spectral resolution of each observation is taken into account when integrating the total Na column density. This is done in order to correctly integrate Io s contribution at Europa s orbit. In order to investigate the role of Europa s O 2 atmosphere (Hall et al., 1995), we used the recent atmospheric model of Shematovich et al. (2004) to calculate the energy lost by the sodium atoms passing through the atmosphere. The O 2 atmosphere is likely not globally uniform (Shematovich et al., 2004; McGrath et al., 2004). Here we assumed a hemispherical atmosphere as the highest sputtering rates are likely on the trailing hemisphere (Popieszalska and Johnson, 1989). We use the density vs altitude calculated by Shematovich et al. (2004). However as discussed in the text, the results presented in Section 4 when atmospheric collisions are included, are for an atmosphere based on density profiles reduced by a factor 3. Collisions between the ejecta and the atmospheric particles are simulated with the same approach as in Leblanc et al. (2002). Finally, we also consider the role of the magnetospheric heavy ion (O n+ and S n+,seecooper et al., 2001) sputtering of Europa s atmosphere. Leblanc et al. (2002) suggested such a process is needed in order to explain the apparent energetic tail of the energy distribution of the Na. This contribution is estimated by introducing a flux of magnetospheric ions with an energy distribution deduced from Galileo observations (Cooper et al., 2001) and extrapolated linearly to low energy. The number of collisions between the impacting ions and sodium atoms in Europa s atmosphere is calculated and the collisions are then simulated with the same approximation used to treat the collisions between the Na atoms and the atmospheric molecules (Leblanc et al., 2002). When including the plasma ion flux onto the atmosphere we, therefore, no longer need to introduce an artificial energetic tail in the surface ejecta distribution (Leblanc et al., 2002). 4. Results 4.1. Brown s measurements on 12/28/1999 The simulated Na cloud in Fig. 2 appears thicker away from Jupiter because negative y 0 values correspond to lines of sight which cross Europa s sodium tail over the trailing hemisphere (see Fig. 1 for the position of the lines of sight with respect to y 0 ). In Fig. 2aofLeblanc et al. (2002),theNa cloud shape in the orbital plane shows the extended banana shape of the cloud first suggested by Smyth and Combi (1988) for Io and recently discussed for Europa (Burger and Johnson, 2004) for Io s Na cloud. The simulated Na cloud in Fig. 2 is less extended along the z 0 direction at 8:15 UT (Fig. 2, right panel) than at 4:28 UT (Fig. 2, left panel) because the ejection rate from Europa s surface is assumed to vary with respect to Jupiter s centrifugal equator implying a larger ejection rate before 4:28 UT than after. Fig. 2 illustrates the quick change of the Na cloud induced by Europa s motion inside the jovian magnetosphere. In this simulation, we also assume a slight North/South asymmetry of the ejecta flux which can be shifted with respect to Europa s equator towards the pole by up to 10 depending on Europa s centrifugal latitude. In Fig. 3, we plotted the results of the simulation for the conditions shown in Fig. 2 along with the observations of Brown (2004) described in Leblanc et al. (2002). Aninversion of the East vs the West directions of Brown s observations was made in Figs. 3 8 in Leblanc et al. (2002). The orientation is corrected in Fig. 3 of this paper. The agreement which is obtained between simulation and observations that are closer than 20 R E to Europa (Fig. 3) confirms that the ejected flux from Europa seen above 6 R E could Fig. 2. D1 + D2 emission (log 10 of krayleigh, kr) obtained by the 3D Monte Carlo model described in Section 3 for Brown measurement conditions. Left panel: at 4:28 UT, 12/28/1999. Right panel: at 8:15 UT, 12/28/1999. Also shown are the slit positions and sizes with respect to Europa used in Brown s observations. From top to bottom, the horizontal slits are named East/West at 20 R E North, East/West at 10 R E North, East/West, East/West at 10 R E South and East/West at 20 R E South, respectively. The vertical slit is named North/South. The dashed dark line represents the position of the jovian centrifugal equator at these two moments.

7 Europa s Na cloud and torus 373 Fig. 3. Comparison between Brown data (circles with error bars) with our simulations (a) without the contribution due to Io and with an energetic tail (gray solid lines with triangles), and (b) without an energetic tail (gray solid lines with squares), and (c) with the calculated contribution from Io as described below (dot dashed line). No Europa atmosphere was included in this simulation. Fig. 4. Na density in Europa s orbital plane (log10 of Na/cm3 ) produced by the ejection of Na atoms at Io (only sputtering induced by Io s torus particle bombardment of Io s atmosphere). The Na considered here are the ones for which velocity is within the range of detection of an observation centered on Europa Doppler shift. The positions of Europa and Io are indicated by crosses (dark and white crosses are for Europa and Io, respectively). Left panel: 4:28 UT, 12/28/1999. Right panel: 8:15 UT, 12/28/1999. vary significantly with respect to the Europa centrifugal latitude. An even larger impacting flux variation, as suggested by Moncuquet et al. (2002) would induce a less satisfactory agreement within this region than that shown in Fig. 4 but would remain within the error bars for this observa- tion. Fig. 3 shows also that the energetic tail of the velocity distribution of the ejecta does not improve the comparison. The simulations underestimate the emission intensity above 20 RE from Europa s center outside Europa s orbital plane. This discrepancy will be discussed in Section 6.

8 374 F. Leblanc et al. / Icarus 178 (2005) In order to investigate the role of Io as a source of part of the Na observed at Europa, we simulate its contribution as described in Section 3, taking into account the variation of the relative position of the two satellites within Jupiter s magnetosphere. The methods used to calculate trajectories, to reconstruct the density in Jupiter magnetosphere and to calculate the column density are the same as for the Europa cloud. Fig. 4 displays the shape of the Io cloud (in Na/cm 3 ) at the beginning and at the end of Brown s observations in the orbital plane of the two satellites. In Fig. 4 we show only the density of the Na atoms that are within the range of detection of an observation centered on Europa Doppler shift (typically Europa/Earth relative velocity +/ 5km/s). Europa is seen to have been in the tail of Io s Na torus during Brown s observation. The peak of density in Fig. 4 is that population of Na atoms released from Io when or just before Io passed Europa. These have been slowed by Europa s gravity, but the mean velocity remains slightly larger than Europa s orbital velocity. Using these results the dashed lines in Fig. 3 are the sum of the simulation of Europa s Na cloud and of the Na cloud associated with Io. The Io background correction is seen to be small but nonnegligible. However, it cannot explain the discrepancy between simulation and observations in Fig. 3 in the furthest regions from Europa Potter and co-workers measurements Observation on 11/28/2000 (Fig. 1b) Fig. 5 displays the Na cloud obtained for Europa s position on the 11/28/2000 (Fig. 1b). On this date, the Na cloud displays a slight North/South asymmetry (Table 2) with somewhat larger emission intensities in the Southern. Because Europa was North of the centrifugal equator at the time of this observation (Table 1), this asymmetry can be accounted for if the bombarded surface, and, therefore, the spatial distribution of sodium, is slightly shifted towards the south pole. There is also an East/West asymmetry which is due to the shape of the cloud at Europa. This cloud is slightly extended towards the East direction that is, towards Jupiter (negative y 0 direction, see Fig. 1). Fig. 6 displays the velocity distributions observed by Potter and co-workers (circles) within the pixels represented by squares in Fig. 5, and compares them to the result of our model. The simulated distribution takes into account the Doppler shifts of each moving atom and its effect on the emission intensity of the emission for the spectral resolution of Potter and co-workers observations. The distribution of velocities from the simulations is in reasonable agreement with observations at most positions. The observed energetic velocities are underestimated by the simulation. Velocity distributions observed on East/West sides of Europa (panels a, b, e, and f) are roughly symmetric with respect to zero and are less peaked than in the simulation. In contrast, both observations at 2 and 3 North/South (panels c, d, g, and h) have a larger extended distribution in Fig. 5. D2 emission (log 10 of kr) shown in this figure has been obtained by the 3D Monte Carlo model described in Section 3 for Potter and co-workers measurement conditions on the 11/28/2000. Also shown are the pixel positions and sizes with respect to Europa extracted from the image slicer of Potter and co-workers observation (dark squares). At the time of the observations, the closest point of the jovian centrifugal equator was at 15 Europa radii South from Europa center. From top to bottom along y 0 = 0, the dark squares represent the pixels named 3 North, 2 North, 2 South, and 3 South. From left to right along z 0 = 0, the squares represent the positions at 3 East, 2 East, 2 West, and 3 West. the positive velocity range than in the negative values. Since our simulation does not show such an asymmetry in the velocity distribution, the source of Na atoms ejected from the leading side of Europa at high latitude is probably larger than described in our model (producing an important contribution of particles with positive velocities out of the orbital plane). The energetic tail of the distribution suggested in Leblanc et al. (2002) slightly improves the agreement between simulation and observation (see as an example Fig. 6 panels e and f). The effect of collisions in the atmosphere does not improve the agreement. The role of the magnetospheric jovian ions in production of energetic Na particles is not efficient even if the flux of magnetospheric jovian ion is increased in an unrealistic way (i.e. the flux of energetic jovian particles at Europa calculated as described in Section 3 is increased by up to three orders of magnitude). At the time of these observations, Io and Europa were in opposite hemisphere with respect to Jupiter, as shown in Fig. 7, implying a very small contribution of the Io Na atoms at Europa. Moreover, the particles ejected from Io which reach the Europa environment have too large a velocity to be seen within the considered range of Doppler shift. As a consequence Io background is fully negligible at this date. Table 3 provides the emission intensities in kr (krayleigh) measured by Potter and co-workers for those extracted pixels corresponding to the velocity distributions displayed Fig. 6. Table 3 also provides results obtained with our model. The closest agreement between simulation and observation is obtained for a Na cloud which does not interact with an atmosphere and with a small North/South asymmetry of

9 Europa s Na cloud and torus 375 Fig. 6. Velocity distribution of the Na atoms for 11/28/2000 observation. The simulated distribution is normalized such that the total associated emission is equal to the total emission of the observation. (a) 2 East, (b) 2 West, (c) 2 South, (d) 2 North, (e) 3 East, (f) 3 West, (g) 3 South, (h) 3 North (Fig. 5). Circle symbols: measurements. Line with triangles: simulation without atmosphere and without ion contribution (using the energetic distribution without an energy tail, see Leblanc et al., 2002). Line with square: identical but with an energetic tail (Leblanc et al., 2002). Line with crosses: simulation with an atmosphere and with magnetic ion sputtering. Negative velocities are for particles moving towards the observer with respect to Europa (negative Doppler shift). the ejecta with respect to centrifugal latitude (6th and 7th columns of Table 3). This corresponds to differences of less than 30% for all the different positions around Europa except for the 2 East position as will be discussed further. The results displayed in the third and fourth columns show that even with a reduced atmospheric density with respect to Shematovich et al. (2004), the decrease of the emission intensity from 2 to 3 is too large. This result is independent of the energy distribution of the Na ejected from Europa (see 3rd and 4th columns in Table 3). When collisions between the Na ejecta and the O 2 atmosphere are taken into account, the density of Europa Na cloud decreases faster with distance to Europa than without collision. The O 2 atmosphere was described in a rather simple way (introducing only a trailing/leading hemispherical asymmetry associated to the jovian energetic particle bombardment of the surface). Such description is probably too simple in light of the most recent observations by HST (McGrath et al., 2004). However,

10 376 F. Leblanc et al. / Icarus 178 (2005) Table 3 Total D2 emissions in kr in each pixel extracted from the image slicer on 11/28/2000 (last column) and results of the simulation (3rd to 7th columns) Date Position of the pixel Simulation with atmosphere (kr) Simulation without atmosphere (kr) Observation With energetic tail Without energetic tail With energetic tail Without energetic tail (kr) With a strong North/ South asymmetry With a small North/South asymmetry 11/28/00 2 East ± West ± South ± North ± East ± West ± South ± North ± 0.3 Position of the pixels in the second column refers to Fig. 5. Energetic tail case corresponds to simulation using an energy distribution of the Na ejecta with a significant increase of the energetic part of the distribution (see Leblanc et al., 2002, for more details). Simulation with atmosphere corresponds to simulation where collisions between Na ejected from the surface and O 2 atmosphere are included. North/South asymmetry is for a North/South asymmetry of the spatial distribution of the ejecta varying with respect to Europa centrifugal latitude. Strong North/South asymmetry is for a 45 shift of the peak of ejection towards North or South poles depending of Europa centrifugal latitude. Small North/South asymmetry is for only 10 shift. Fig. 7. Na density in Europa s orbital plane (log 10 of Na/cm 3 ) produced by the ejection of Na atoms at Io (only sputtering induced by Io s torus particle bombardment of Io s atmosphere and only the Na atoms whose velocity corresponds to the observer range of detection centered on Europa +/ 5 km/s). The positions of Europa and Io are indicated by the dark and white crosses respectively and correspond to their position at the beginning of Potter s 11/28/2000 measurements (3:48 UT). x 0, y 0 axes are shown in Fig. 1. The North pole direction (the rotation vector direction of Io and Europa) is towards the reader. Jupiter is in the center of the figure. based on the results in Table 3, no significant interaction is likely between the Na ejected from the surface of Europa and the O 2 atmosphere, suggesting a much more tenuous average atmosphere. As seen in Table 3, the North/South asymmetry is too large when assuming the maximum asymmetry in the ejecta flux as discussed. The observed East/West asymmetry (8th column in Table 3) is difficult to explain; in particular the lack of decrease of the emission intensity from 2 to 3 along the East direction cannot be explained by this simulation. The simulation is slightly better at 3 from Europa when an energy distribution with an energetic tail is used. Fig. 8. Same legend as in Fig. 5 but for 11/29/2000. Dark squares correspond to pixels from top to bottom along y 0 = 0: 4 North, 3 North, 2 North, 2 South, 3 South and from left to right along z 0 = 0, 4 East, 3 East, 2 East, 2 West, 3 West, and 4 West. Europa center is 28.3 R E South of the jovian centrifugal equator Observation on 11/29/2000 (Fig. 1c) In Fig. 8, the North/South asymmetry of the Na cloud is due to the assumption that the spatial distribution of the ejecta rate is dependent on the distance to the jovian centrifugal equator as discussed. At this date, Europa was South of the jovian equator (Table 1). The East/West asymmetry is due to the geometry of the observations, the leading side of Europa is better seen from the East side (negative y 0 ) whereas the trailing side is better seen from the West side of Europa (positive y 0,see Fig. 1). Fig. 9 displays the velocity distributions as observed during the night of 11/29/2000 at the pixel positions shown in Fig. 8 (dark squares). Here we see Europa s leading side. The

11 Europa s Na cloud and torus 377 Fig. 9. Velocity distribution of the Na atoms for 11/29/2000 Potter and co-workers observation. The simulated distribution has been normalized to the total emission of the observation. (a) 2 East, (b) 2 West, (c) 2 South, (d) 2 North, (e) 3 East, (f) 3 West, (g) 3 North, (h) 4 East, (i) 4 West, (j) 4 North (Fig. 8). Same legend as in Fig. 6. agreement between the observations (circle symbols) and the simulations (triangle, square, and cross lines) is reasonably good in all the pixels. The velocity distribution along the East direction, away from Jupiter (panel a, e, and h) is usually slightly asymmetric, with a broader distribution at positive velocities. This is less evident for the other directions.

12 378 F. Leblanc et al. / Icarus 178 (2005) The positive and negative velocities correspond to atoms moving away from the observer and toward the observer respectively; that is, atoms moving slower or faster than Europa. Our simulation reproduces the symmetry/asymmetry of the velocity distribution in most of the pixels (a Na cloud mainly formed from the ejection of Na atoms from the trailing hemisphere). The role of the atmosphere or of collisions with jovian energetic ions within the atmosphere (lines with crosses) suggested by Leblanc et al. (2002) is not confirmed by the result of the simulation. On another hand, the energetic tail (lines with squares in Fig. 9) that was suggested as produced by collision between jovian energetic ions and Europa s atmosphere and calculated in order to get a best fit to Brown s observations (2004) in Leblanc et al. (2002) slightly improves the comparisons between observation and simulation. There is no significant contribution to the observed Na emission intensity due to Io on this date. Indeed Io is just behind Europa (see Fig. 10) and only the fastest Na atoms ejected from Io contribute. These have velocities that are too fast to be included in the spectral range described in this work. Table 4 shows that neither the case with an atmosphere (columns 3 and 4) nor the case without an atmosphere (columns 5 and 6) agree with the Potter and co-workers observations (7th column) on 11/29/2000. There is globally a factor of 2 to 2.5 lower emission intensity obtained with the simulation than observed at all positions. Such discrepancy suggests a much larger rate of ejection at this moment that at 11/28/2000. Multiplied by a factor between 2 and 2.5, the simulated signal is in good agreement with the observations. This suggests that the morphology of the cloud produced by the simulation is consistent with the observations. When the simulation intensities have been increased by a factor between 2 and 2.5, the best agreement is obtained in the case with no atmospheric collision and an energy distribution with energetic tail Observation on 11/30/2000 (Fig. 1d) In Fig. 11, the North/South asymmetry of the Na cloud has the same origin as the one for 11/29/2000, an ejection Fig. 10. Na density in Europa s orbital plane (log 10 of Na/cm 3 ) produced by the ejection of Na atoms at Io (only sputtering induced by Io s torus particle bombardment of Io s atmosphere and only the Na atoms whose velocity corresponds to the observer range of detection centered on Europa +/ 5 km/s). The positions of Europa and Io are indicated by the dark and white crosses respectively and correspond to their position at the beginning of Potter s 11/29/2000 measurements (6:54 UT). The North pole direction (the rotation vector direction of Io and Europa) is toward the reader. Jupiter is in the center of the figure. rate dependent on the distance from the jovian centrifugal equator. It is more pronounced here than on 11/29/2000 because Europa is further from the jovian centrifugal equator (Table 1). The East/West asymmetry also has the same origin, except that in this case, the leading side is on the West (positive y 0 ) and the trailing side on the East (negative y 0, see Fig. 1). The agreement of the simulated velocity distributions with the observed for the pixels, whose size and position are represented in Fig. 11 by the dark squares, is not as good as for the two other dates of observation (Fig. 12). The simulated distribution exhibits a larger peak centered at Europa s velocity and is narrower than that observed. Moreover the simulation underestimates the velocity distribution in the positive velocity range (the particles which are moving away from the observer). These differences may be in part due to Table 4 Total D2 emissions in kr in each pixel extracted from the image slicer on 11/29/2000 (last column) and results of the simulation (3rd to 6th columns) Date Position of the pixel Simulation with atmosphere (kr) Simulation without atmosphere (kr) Observations (kr) With energetic tail Without energetic tail With energetic tail Without energetic tail 11/29/00 2 East ± West ± South ± North ± East ± West ± North ± East ± West ± North ± 0.4 Position of the pixels refers to Fig. 8. The same legend as in Table 3 applies. A small North/South asymmetry of the ejecta flux has been considered.

13 Europa s Na cloud and torus 379 Fig. 11. Same legend as in Fig. 5 but for 11/30/2000. Dark squares are for pixels from top to bottom along y 0 = 0: 4 North, 3 North, 2 North, 2 South, 3 South and from left to right along z 0 = 0, 4 East, 3 East, 2 East, 2 West, 3 West, and 4 West. Europa center is 44.7 R E South of the jovian centrifugal equator. the fact that on this date the observed signal is weaker than at the two other dates as indicated by negative values of the intensity (note the level of noise on the observed signal shown by negative values of the intensity. Such values are due to the uncertainty in the background subtraction). However, the observed velocities indicate that a significantly larger velocities in the positive velocity range at all the pixel positions. Such a trend indicates that Europa Na cloud is produced mainly from the trailing hemisphere. There is no significant difference between the simulations with and without atmosphere. The role of the energetic tail is not obvious, in particular close to Europa, even if it slightly improves the agreement with the measurement far from Europa (at 4 ) at large velocity (Fig. 12, panels h, i, and j). A significant population of sodium atoms from Io is not present in the Potter observations. On 11/30/2000, Io was just ahead of Europa as shown in Fig. 13. Moreover Io s relative velocity with respect to the observer was close to that of Europa. As a consequence, the density of simulated Na atoms originating from Io that reach Europa with velocities within the range of detection of the observation is up to two orders higher than on the two previous dates. The values of the contribution in kr due to this population of Na atoms for each pixel are indicated in Table 5 (7th column). It can be seen that this contribution can dominate the emission at Europa (8th column). Since the velocity distribution of Iogenic sodium is not centered on Europa velocity a larger population in the negative velocity range is seen. This corresponds to atoms moving faster than Europa (probably launched when Io passed Europa). This characteristic was described in the context of Fig. 3. The fact that such a population is not present in the Potter and co-workers observations means that either the population is overestimated or the energy of this population is underestimated in our calculation. Therefore, depending on the relative position of Io and Europa within the jovian magnetosphere the Iogenic contribution could be significant and should be addressed using a complete model of Io Na cloud (Burger, 2003). In contrast with the previous dates, the emission predicted by the simulation is, in general, larger than that observed, as shown in Table 5 (by a factor 1.5 to 2). Once again, such a discrepancy suggests a different ejection rate at this particular position of Europa than the one used in the simulation. When the simulated emission intensity has been corrected by a factor 1.5 to 2 it is in good agreement with the observed intensities, particularly for the case with an ejecta flux having an energy distribution with an energetic tail. Table 5 also shows that the presence or absence of an atmosphere does not change the agreement between simulation and observations. 5. Discussion One of the main conclusions from Section 5 is that the discrepancies in the total emission intensity from one measurement to another in Potter s data cannot be explained solely by the differences in the geometry of obser- Table 5 Total D2 emissions in kr in each pixel extracted from the image slicer on 11/30/2000 (last column) and results of the simulation (3rd to 6th columns) Date Position of the pixel Simulation with atmosphere (kr) Simulation without atmosphere (kr) Simulation of With an Without an With an Without an Io contribution energetic tail energetic tail energetic tail energetic tail Observation (kr) 11/30/00 2 East ± West ± South ± North ± East ± West ± North ± East ± West ± North ± 0.1 Position of the pixels refers to Fig. 11. The same legend as in Table 3 applies. A small North/South asymmetry of the ejecta flux has been considered.

14 380 F. Leblanc et al. / Icarus 178 (2005) Fig. 12. Velocity distribution of the Na atoms for 11/30/2000 Potter and co-workers observation. The simulated distribution has been normalized to the total emission of the observation. (a) 2 East, (b) 2 West, (c) 2 South, (d) 2 North, (e) 3 East, (f) 3 West, (g) 3 North, (h) 4 East, (i) 4 West, (j) 4 North (Fig. 11). Same legend as in Fig. 6. vations with respect to the energetic particle flux. It is more likely that variation in the source rate from 11/28/2000 to 11/30/2000 was significant. In order to reproduce the measurements, it is necessary to introduce two to three-fold

15 Europa s Na cloud and torus 381 Fig. 13. Na density in Europa s orbital plane (log 10 of Na/cm 3 ) produced by the ejection of Na atoms at Io (only sputtering induced by Io s torus particle bombardment of Io s atmosphere and only the Na atoms whose velocity corresponds to the observer range of detection centered on Europa +/ 5 km/s). The positions of Europa and Io are indicated by the dark and gray crosses respectively and correspond to their position at the beginning of Potter s 11/30/2000 measurements (2:44 UT). The North pole direction (the rotation vector direction of Io and Europa) is towards the reader. Jupiter is at the center of the figure. larger sources of ejecta on 11/29/2000 and to decrease the rate of ejecta on 11/30/2000 by a factor between 1.5 and 2. One possible explanation for such a variation is the redistribution of the Na atoms on Europa s surface during its orbit around Jupiter. When the trailing hemisphere is also the dayside of Europa, the Na atoms are mainly ejected from this hemisphere and accumulate preferentially on the nightside of Europa, where ion sputtering, is less efficient and thermal and photon-stimulated desorptions are not active. Such geometry corresponds to Europa being West to Jupiter (see Fig. 1). Therefore, according to this scenario, the Na atoms ejected from Europa s surface are reabsorbed on the leading side of Europa without being re-ejected. This lasts for almost half an orbit. After Europa passes midnight local time in jovian frame, the Sun rises on the leading side and should contribute to the release of atoms accumulated during the previous half orbit. Therefore, the ejection of sodium from Europa s surface should be maximum between midnight and 6 a.m. local jovian time (that is at the time of 11/29/2000 observation). Brown s observations on 12/28/1999 viewed Europa at a geometry between those of 11/29/2000 and 11/30/2000 locations viewed by Potter and co-workers (see Fig. 1). The closest point to Europa in Brown s observation is at roughly 6 R E. At this position, along the East/West direction, the D1 + D2 emission intensities (Fig. 3 upper left panel) reported by Brown (2004) can be compared with the D2 emission intensity reported by Potter and co-workers for the pixels at 3 (between 4.7 and 6.5 R E from Europa center) and 4 (between 6.6 and 8.4 R E ) along the East and West directions (Tables 4 and 5 last column). Between 6 and 8 R E,Brown observed a total D1 + D2 emission between 1600 and 900 R (Rayleigh) averaged between West and East sides whereas Potter and co-workers observed a D2 emission brightness between 3900 and 1500 R on the 11/29/2000 and between 600 and 400 R on the 11/30/2000. The latter values can be converted into D1 + D2 intensities by multiplying them by 1.66, which implies an optically thin atmosphere (Brown and Hill, 1996). Therefore the equivalent D1 + D2 intensities for 11/29/2000 are 6500 R and 2500 and for 11/30/2000, 1000 and 700 R. From this comparison it appears that the general trend of a decrease of the total emission intensity observed at Europa between midnight and noon jovian local time, is consistent with both Brown and Potter and co-workers observations. During the orbital motion of Europa from midnight to noon local times, the observable region of Europa s surface, which is poorly sputtered by the incident jovian particles or by UV photons, increases in size. This region is the part of the nightside that is not fully bombarded by the jovian particles on 11/30/2000 (positive x and y part in Fig. 1d) and, therefore, accumulates sodium. Such a region begins to be gradually lit to the Sun as Europa rotates through a local time of 6 p.m. This is roughly the orbital position of the 11/28/2000 observations, consistent with the larger emission brightness observed on 11/28/2000 with respect to 11/30/2000. In order to verify this conclusion, we performed a simulation of Europa s Na cloud including a dependence of the Na ejecta flux from Europa with respect to its local time position in the jovian frame. This is similar to the simulation carried out earlier for the morphology of the sodium cloud at Mercury (Leblanc and Johnson, 2003). The variation of the Na cloud with respect to latitudinal position is described by varying the flux from a minimum to a maximum as discussed in the previous section. We found in the present case that a slightly smaller variation from minimum to maximum values gives a better agreement with observation. We fix the nominal variation induced by the latitudinal variation of Europa within the jovian magnetosphere to be between the following minimum Na/cm 2 /s and maximum Na/cm 2 /s. The variation induced by Europa motion around Jupiter (jovian local time dependent variation) is described in the following way: at noon local time, we supposed that the minimum and maximum associated to the latitudinal position of Europa are only 20% of the nominal values given previously that is Na/cm 2 /s and Na/cm 2 /s, such minimum and maximum then increase linearly up to 6 p.m. local time where they reach 90% of the nominal values ( Na/cm 2 /s and Na/cm 2 /s), these minimum and maximum increase linearly up to midnight local time where they reach 140% of the

16 382 F. Leblanc et al. / Icarus 178 (2005) Fig. 14. Comparison between Brown data (circles) with our simulation with the contribution due to Io and with an energetic tail (solid lines) considering a dependence of the ejected flux with respect to Europa jovian local time within Jupiter s magnetosphere and with respect to Europa s centrifugal equator. No Europa atmosphere was taken into account. nominal values (that is Na/cm2 /s and Na/cm2 /s), these minimum and maximum again increase linearly to reach maxima equal to 320% of the nominal values at 3 a.m. (that is Na/cm2 /s and Na/cm2 /s), between 3 a.m. and 6 a.m. these minimum and maximum quickly decrease linearly down to 15% of the nominal values (that is Na/cm2 /s and Na/cm2 /s), at the end, these minimum and maximum decrease linearly one more time up to noon local time where they reach 20% of the nominal ones. Fig. 14 provides the result of this later simulation using no atmosphere but an energetic tail for the energy distribution of the ejecta. The relatively good agreement shown in Fig. 14 between the observations of Brown and simulations is apparent not only below 20 RE but also at all the slit positions used. It is seen that a second variation of the ejecta flux, linked to Europa s local time, can also roughly account for the discrepancies observed in the previous sections, especially above 20 RE. It is also important to note that the latitude dependent variation suggested in the previous section is still the only explanation of the difference in intensities observed by Brown within 20 RE from the surface at two times separated by only few Earth hours (see discussion in Leblanc et al., 2002). There remains some discrepancy between simulation and observation along the East/West direction in Europa ecliptic plane (upper left figure) that suggests a different dependence of the ejecta flux with respect to Europa latitude than assumed in this work. Table 6 provides the comparison between this simulation (using no atmosphere but an energetic tail for the energy distribution of the ejecta) and the observations of Potter and coworkers described in Section 2. The generally good agreement between the intensities of the Na emission reported by Potter and co-workers and the simulations for the three nights of observation at all positions around Europa confirms the assumption on the variation of the ejecta flux with respect to Europa jovian local time. It is however difficult to explain that the ejected flux from Europa could vary from