Mass-loading, pile-up, and mirror-mode waves at comet 67P/Churyumov-Gerasimenko

. The data from all Rosetta Plasma Consortium instruments and from the ROSINA COPS instrument are used to study the interaction of the solar wind with the outgassing cometary nucleus of 67P/Churyumov-Gerasimenko. During and June 2015, the interaction was ﬁrst dominated by an increase in the solar wind dynamic pressure, caused by a higher solar wind ion density. This pressure compressed the draped magnetic ﬁeld around the comet, and the increase in solar wind electrons enhanced the ionization of the outﬂow gas through collisional ionization. The new ions are picked up by the solar wind magnetic ﬁeld, and create a ring/ring-beam distribution, which, in a high- (cid:12) plasma, is unstable for mirror mode wave generation. Two different kinds of mirror modes are observed: one of small size generated by locally ionized water and one of large size generated by ionization and pick-up farther away from the comet.


Introduction
The theory of the interaction of an outgassing comet with the solar wind magnetoplasma started with the explanation of the formation and physics of the cometary ion tails by Biermann (1953) and Alfvén (1957). With the beginning of the space age and spacecraft-flybys of comets in the last century, e.g. VEGA 1, 2, Giotto, ICE, Sakigake and Suisei by comet 1P/Halley, Giotto at 26P/  Skjellerup and ICE at 21P/Giacobini-Zinner, much has been learned about the various physical processes taking place in the plasma around the outgassing cometary nucleus.
In the current century, on 20 January 2014 the Rosetta spacecraft  was woken up after 18 months of hibernation, and the spacecraft cruised towards its rendezvous with comet 67P/Churyumov-Gerasimenko (67P/CG). On 6 August 2014 Rosetta arrived at its target, and 20 started its escort phase, following the comet along its orbit from pre-to past-perihelion. 67P/CG's perihelion was on 13 August 2015.
In this paper the data from the Rosetta Plasma Consortium instruments (RPC, Carr et al., 2007) are used to study the interaction of the outgassing nucleus of comet 67P/CG and the solar wind magnetoplasma at a time when the comet is closing in on its perihelion. Unlike the previous missions 25 mentioned above, Rosetta does not perform a quick flyby of the comet, but remains at the comet, moving at a very slow pace of ∼ 1 m/s. This means that Rosetta RPC can follow the development of the interaction of the solar wind with the increasingly more actively outgassing nucleus as comet 67P/CG heads towards perihelion, and the decreasing activity after perihelion.
After initial arrival a new phenomenon was found, now called the "singing comet" (Richter et al., 30 2015); ∼ 40 mHz waves generated by a cross-field current instability created by freshly ionized, not yet magnetized water ions within the Larmor sphere (sphere with radius of 1 Larmor radius, Sauer et al., 1998) of the comet. At that time, these newly created ions also indicated the "birth of a magnetosphere" (Nilsson et al., 2015a) for which the spatial distribution of the low-energy plasma was discussed by Edberg et al. (2015b). However, "conventional signatures" such as Alfvén waves 35 or cyclotron waves were not observed.
Later in the mission, with comet 67P/CG approaching its perihelion, the activity of the nucleus increased significantly. Various strong outbursts were observed by the Rosetta NAVCAM, see in Fig. 1, which mainly shows reflected sunlight on dust grains, and these might significantly influence the plasma interactions. Rotundi et al. (2015) discussed the link between gas and dust emissions. Indeed, 40 in the second half of July 2015, the outgassing of the nucleus was so strong that a diamagnetic cavity was created which extended well past the ∼ 180 km distance of Rosetta from comet 67P/CG (Glassmeier et al., 2015;Götz et al., 2015, see also http://blogs.esa.int/rosetta/2015/08/11/cometsfirework-display-ahead-of-perihelion/). Koenders et al. (2013Koenders et al. ( , 2014 have predicted distances of ∼ 25 km for the diamagnetic cavity distance under quiet conditions. Such strong outburst conditions 45 have not been modeled yet. In a diamagnetic cavity the outflowing neutral gas and plasma is strong enough to keep the solar wind and its embedded magnetic field at bay, pushing it away from the nucleus (see e.g. Cravens and Gombosi, 2004). This creates a magnetic field free region around the comet. However, the Rosetta RPC magnetometer did still measure a very small magnetic field, which is an indication for the not-fully corrected offsets of the magnetometer, which can be either 50 inherent or arise from stray fields from the spacecraft. In this paper the measured fields have been used to correct the offset.
In this paper a first overview and discussion is given of the events taking place on 6 and 7 June 2015. There is a ∼ 6 hours quasi-periodic variation in the neutral and plasma density (Hässig et al., 2015;Edberg et al., 2015b). First the effect of the mass loading on the induced magnetosphere is 55 discussed, including magnetic field pile-up and draping, relating it to variations in the solar wind.
Second, the behaviour of the freshly created ions and the resulting mirror-mode wave activity is investigated.
2 Mass loading of the induced magnetosphere On 6 June 2015 there was a higher than usual gas-outflow from the comet, which loaded the induced 60 magnetosphere with neutral gas and plasma. The combined data of the six instruments discussed below, for the two-day interval of 6-7 June 2015 are shown in Fig. 2. From top to bottom are shown: the Ion and Electron Spectrometer (IES, Burch et al., 2006) time-energy spectrogram, the Ion Composition Analyser (ICA, Nilsson et al., 2006) time-energy spectrogram, the low-pass filtered magnetic field components in Cometocentric Solar EQuatorial (CSEQ 1 ) coordinates from the MAGnetome-65 ter (MAG, ; the magnetic field strength, the Mutual Impedance Probe (MIP, Trotignon et al., 2006) deduced electron densities; the LAngmuir Probe (LAP, Eriksson et al., 2006) P1 current, the IES ion and electron density; the Rosetta Orbiter Spectrometer for Ion and Neutral Analysis (ROSINA, Balsiger et al., 2007) COmetary Pressure Sensor (COPS) neutral density; the location of the spacecraft with respect to the comet; the IES ion velocity in CSEQ and the angles of 70 the ion velocity with the radial direction to the comet and with the magnetic field direction.
In both the IES and the ICA, an increase in ion counts and energies in the ion channels starting at approximately 1800 UT is seen. There is an increase in energy from ∼ 10 eV to up to ∼ 500 eV for both instruments, where IES seems to show a sawtooth-like behaviour with a quasi-period of around 4 to 6 hours as shown in Fig. 2.

75
The neutral gas density measured by COPS of ROSINA are shown in Fig. 2 panel H. A semiperiodic density fluctuation at a quasi-period of ∼ 6 hours, and a few maxima at ∼ 0830 and ∼ 1545 UT and a very strong peak at ∼ 2100 UT are seen. The second and third bursts (vertical dashed lines) coincide well with the start of energy increases in the IES and ICA data in Fig. 2.
It is clear from comparing panels A, B, D and G in Fig. 2 that a severe change occurs in the 80 environment around comet 67P/CG; the magnetic field strength starts to increase around 1100 UT, when at the same time IES and ICA data show an increase in counts and energies of the ions.
As the total magnetic field strength increases, the fluctuations in the magnetic field are also enhanced: the field increases from averageB ≈ 27 nT with a standard deviation σ ≈ 11 nT during 0000 to 1200 UT toB ≈ 41 nT with σ ≈ 16 nT during 1200 to 2400 UT. In the early hours of 7 85 June the magnetic field strength has returned to a lower valueB ≈ 30 nT with σ ≈ 12 nT, and the 1 CSEQ: A cometocentric coordinate system with the x-axis pointing towards the Sun, the z-axis is aligned with the rotational axis of the Sun, and the y-axis completes the triad.
3 IES densities in Panel E return to the values as at the beginning of 6 June and the ion densities and the LAP P1 current follow the COPS neutral densities in Panel F. It should be noted that near 2400 UT on 6 June the magnetic field strength decreases to very a low value of B m ≈ 4 nT.
There is an interesting correlation between the data from ROSINA COPS neutral density and the 90 densities measured by the RPC instruments. In Fig. 2 the vertical dashed lines are coincident with the maxima in the COPS data, with the black dashed lines marking the "regular" 6-hour maxima.
The sharp density peaks at the maroon coloured dashed lines are artifacts created by reaction wheel offloading on the spacecraft. There appears to be a delay in the response in the IES time-energy spectrogram to the increased neutral density. After a neutral density maximum, the count rate and 95 the ion energy increase and drop just before a new neutral density maximum is reached again. This may be due to the ionization time, and will have consequences for when RPC-measurable ions can be observed after neutral injection. However, this is beyond the scope of this paper.
The solar wind transports magnetic fields from the Sun towards the comet. In the surroundings of the comet a conducting layer exists, created by ionization of the outflowing gas from the nucleus. As 100 discussed by Alfvén (1957) the magnetic field cannot pass unimpeded through this region near the nucleus and gets hung-up, whereas the part of the field lines further away are still moving with solar wind velocity. This leads to two phenomena: near the nucleus the magnetic field will pile-up, i.e. increase in strength, as the field is delivered faster than it can be transported away. This creates the so-called induced magnetosphere of the comet. Furthermore, the field lines wrap around the nucleus, 105 get draped, because of the difference in velocity along the field line. These phenomena have been well studied during the flybys of other comets in the last century (see e.g. Smith et al., 1986;Riedler et al., 1986;McComas et al., 1987;Raeder et al., 1987;Delva et al., 2014).

110
At ∼ 1545 UT on 6 June, COPS shows a maximum in the neutral gas density in the quasi-periodic ∼ 6-hour changes. IES shows an increase in energy and counts of the ions over the following four hours, however, note this signature looks different from what is happening after midnight on 7 June.
The IES ion (electron) density, Fig. 2 Panel E, is rather peaked and strongly variable and reaches a maximum density at ∼ 1436 (∼ 1336) UT, which is most likely the result of the increased neutral 115 density at ∼ 0830 UT. After the ∼ 1545 UT neutral density maximum the ion (electron) density starts to increase, with a slight maximum at ∼ 1740 UT.
With the increased plasma density a simultaneous increase in magnetic field strength B m is observed, see Fig. 2 Panel B. This could be a result of more magnetic pile-up because of the increased mass loading generating a layer with higher conductivity and thus a longer diffusion time. It is, 120 however, unclear if an increase in ion density can actually lead to such a strong increase in magnetic field strength through increased hang-up. Volwerk et al. (2014) posited that a decrease in ion density 4 at comet 1P/Halley could be the reason for the disappearance of the nested draped magnetic field between the flybys of Vega 1 and Vega 2. However, it is also quite possible that the increase in magnetic field strength and the increase in ion density are generated by an external source in the 125 solar wind. This will be discussed in the next section.
Interestingly though, the situation is different from what was observed at comet 1P/Halley (see e.g. Gringauz et al., 1986;Neubauer et al., 1986), where the magnetic fluctuations disappeared in the pile-up region. At comet 67P/CG the magnetic fluctuations increase in the pile-up region.
WithB ≈ 50 nT the gyro frequency of water ions is f c,H2O ≈ 40 mHz. Spectral analysis of 130 the interval 1700 -1900 UT on 6 June is performed and displayed in Fig. 3 top-left panel. The three components of the magnetic field are spectrally analysed (cf. McPherron et al., 1972), and displayed. In order to find the confidence level of the peaks, the spectra are fitted by a fourth-order polynomial, which is subtracted from the spectrum and from the residual (bottom-left panel) the ±95% confidence level is determined (see e.g. Bendat and Piersol, 1966), shown as a red solid and 135 dash-dotted line. The spectrum shows that the strongest (highest PSD) component is B y , there is a strong peak at ∼ 4.7 mHz in B x and B y and a peak at ∼ 5.5 mHz in B z , and mutual second and third peaks at ∼ 7.7 and ∼ 13 mHz. No significant signal is found at the water ion gyro frequency.

Massloading at ∼ 2100 UT
At ∼ 2100 UT COPS showed another maximum in neutral gas density. The IES ion density in-140 screases with a maximum N i ≈ 430 cm −3 at ∼ 2240 UT, after which it quickly returns to pre-event values around N i ≈ 50 cm −3 . Spectral analysis of the interval 2100 to 2300 UT of 6 June shows (see Fig. 3 right panels) that the strongest component is B x ; there is a first mutual peak at ∼ 2.8 mHz, a second, stronger, peak in B x is found at ∼ 4.7 mHz, whereas for B y a second peak is found at ∼ 6.0 mHz and for B z a second peak is found at ∼ 5.4 mHz. There seems to be little common 145 behaviour of the three magnetic field components.

Ion Motion
The deduced ion velocities from the IES instrument are shown in Fig. 2 Panel H. On 6 June the ion (H 2 O + ) velocity is aroundv ≈ (−12, −1, 2) km/s, with the magnitude of the components increasing when the mass loading starts around 1600 UT (but the increase in magnetic field strength already 150 starts about two hours earlier). Mainly v x and v z (in CSEQ coordinates) increase in magnitude with strongest change in v x . After the increase in density and the increase in magnetic field strength disappear, just before midnight, v z returns to pre-mass-loading values, but v x and v y strongly increase in magnitude withv ≈ (−23, 10, 1) km/s, lasting for many hours. This means that the ions are mainly moving anti-sunward as discussed by Nilsson et al. (2015b).

155
In order to determine the propagation direction of the ions the angle η with the radial direction to the centre of the comet (red line) is calculated, as well as the angle ζ of the velocity with the local magnetic field (blue line) in Fig. 2 Panel I. Basically, over the whole of 6 June the ions are moving perpendicular to the radial direction to the comet and nearly perpendicular the magnetic field, apart from 0900 -1500 UT, which is related to the rotation of the magnetic field discussed further below. 160 Near midnight, after the enhanced mass-loading, the situation changes: the ions are accelerated in the XY -plane and move still mainly perpendicular to the radial vector with η ≈ 110 • . However, the angle with respect to the magnetic field increases to ζ ≈ 140 • . The latter is what one would expect for newly formed ions being accelerated by the motional electric field (see also Broiles et al., 2015) whilst having an initial velocity at ionization, starting their gyration around the magnetic field, 165 creating a ring-beam distribution, which can be unstable for mirror-mode waves (Hasegawa, 1969;Tsurutani et al., 1982;Gary, 1991;Gary et al., 1993). These are the same kind of ions that, at arrival at comet 67P/CG, caused the so-called singing (Richter et al., 2015), but in a low-density and low-magnetic field environment. The Tao-model shows that the tangential component of the magnetic field B t,t slowly increases in 185 strength and after midnight from 6 to 7 June quickly reverses in sign. With the increase in B t,t the density N SW and dynamic pressure P dyn also increase. The Opitz-Dósa-model shows that the radial magnetic field, B r,o , slowly changes from negative to positive, indicating a heliospheric plasma sheet crossing, which would explain the increase in solar wind density. However, this could also be a signature of a corotating interaction region impinging on the comet's plasma surrounding (Edberg 190 et al., 2015a) As the solar wind velocity does not change during this interval, the increase in dynamic pressure is only created by an increase in ion density, which is clear through the same profiles in panels F and G. The solar wind density in the Tao-model increases by a factor of 4 from N sw ≈ 2 to N sw ≈ 8 cm −3 over ∼ 18 hours. The Opitz-Dósa-model shows a lesser increase of a factor ∼ 2.

195
As the modeling of the solar wind propagation cannot be perfect, in Fig (Glassmeier et al., 2015;Götz et al., 2015) shows that at the location of Rosetta collisions are indeed important. The increased counts/energy/density in the IES and ICA data occur during the shifted increase in solar wind density.

Pile-up and Draping
With the increase in plasma density and magnetic field strength, generated by the increased solar 210 wind dynamic pressure and density, the magnetic field is expected to get more piled-up, as observed, and possibly more draped. For the whole interval the clock (ξ) and cone (ψ) angle of the magnetic field is calculated: When the field strength starts to decreases at ∼ 2100 UT, and reaches a very low value, B m ≈ 4 nT around midnight, the cone angle ψ slowly increases to ∼ 85 • , i.e. far away from the x-direction, whereas the clock angle ξ varies stongly because of the large oscillations in the magnetic field com-225 ponents, the largest of which are also visible in the cone angle.
As there is neither undisturbed solar wind data, nor real undisturbed field around the comet, the draping analysis as proposed by  and applied to comet 1P/Halley (see also Delva et al., 2014;Volwerk et al., 2014) cannot be applied. However, the magnetic field direction and behaviour can be looked at in hedgehog-plots, as in Fig. 5 (Raeder et al., 1987). These oppositely directed magnetic fields have to be separated by current sheets and bring the possibility of magnetic reconnection in the cometary coma (see e.g. Verigin et al., 1987;Kirsch et al., 1989Kirsch et al., , 1990.
The rotation of the magnetic field, as shown in Fig i.e. moved to the other side of a current sheet. Although in principle this could be a signature of 8 component reconnection, the plasma data are too sparse to draw such a conclusion.
Using the low-pass filtered data (periods longer than 10 min), the field changes by ∆B max ≈ 21 nT over a time-span of 11 min. With a spacecraft velocity of v sc ∼ 1 m/s, assuming the rotations convect over Rosetta with this velocity, making ∆L ≈ 660 m, and using Ampère's law: to calculate the current density (neglecting the displacement current): For the second rotation the field change is ∆B max ≈ 35 over a time span of 7 min, which leads to a current density of J ≈ 66µA/m 2 . Because of the assumed slow convection velocity ∆L remains 270 small, an upper limit for ∆L can be found under the assumption of frozen in fields and a convection velocity of ∼ 10 km/s, which would significantly decrease the current density by a factor ∼ 10 4 .
5 Crossing from 6 to 7 June: mirror-mode waves Pick-up of freshly ionized ions into a streaming magnetoplasma leads to the creation of a ring/ringbeam distribution in velocity space, which is unstable (see e.g. Hasegawa, 1969;Tsurutani et al., 275 1982; Gary, 1991;Gary et al., 1993). Depending on the plasma-β this can lead to either ion cyclotron waves (low-β) or mirror-mode (MM) waves (high-β). In the case of comet 67P/CG, the plasma-β is high and thus MM waves are expected. They were also observed e.g. at comet 1P/Halley (see e.g. Schmid et al., 2014;Volwerk et al., 2014). The instability criterion for MM waves is given by: where T ⊥ and T ∥ are the ion-temperatures perpendicular and parallel to the background magnetic field and β ⊥ is the perpendicular plasma-β determined only using T ⊥ . The MM wave behaves in such a way that the perpendicular pressure p ⊥ of the plasma is in anti-phase with the magnetic pressure p B , while the total pressure remains constant.

285
On 7 June the ion density returned to pre-event values, the magnetic activity, however, remains.
To study the difference in the four hours before and after midnight, the magnetic field and plasma data are plotted in Fig. 7.
It is clear from the panels in Fig. 7 that during the last 4 hours of 6 June (left panels) the MIP electron density variations (red dots) seem to be in phase with the low frequency variations of the 290 total magnetic field. After 6 June ∼ 2300 UT there is no MIP density available anymore and after 7 June ∼ 0010 UT LAP P1 currents are available as a proxy for the plasma density. Over the first 4 hours of 7 June, Fig. 7 right panels, there often seems to be an anti-correlation between the the total magnetic field Bm and the LAP P1 current.

9
Starting at 6 June around ∼ 2300 UT quasi-periodic dips occur in the magnetic field strength, 295 some of which seem to be anti-correlated with the LAP P1 current. This could imply that the freshly mass-loaded magnetospheric magnetic field is mirror-mode unstable (see e.g. Hasegawa, 1969;Tsurutani et al., 1999;Schmid et al., 2014;Volwerk et al., 2014).
As the resolution of the plasma data is too low to check the pressure balance over the MM structures, the magnetic-field-only method by Lucek et al. (1999) is used to investigate the data for MM 300 waves. These waves are expected to have strong magnetic field variations, ∆B/B, and they are non-propagating structures, only convected by the streaming magnetoplasma in which they are embedded. This means that in an MVA the minimum variance direction should be perpendicular to the background magnetic field and the maximum variance direction along the background magnetic field. A study by Price et al. (1986) showed that the angle between maximum variance direction and A zoom-in on two ten-minute intervals of Fig. 8, and adding the density data of either MIP or LAP is shown in Fig. 9. In the first interval 2230 -2240 UT there are short periods where the criteria are almost fulfilled, the maximum variance angle ϕ is rather large. Unfortunately, the electron density estimated by MIP is unavailable when the plasma frequency is out of the frequency range of the 320 instrument, or when the electron density is small enough and the electron temperature large enough for the Debye length to be much larger than the instrument emitter-receiver length scale. This makes it difficult to find a correlation between B m and N e for the whole time series. Before 2235 UT, when θ > 80 • it is difficult to interpret the electron density and thus the inset panel zooms in once more on the interval 2231 -2232:30 UT. There it is clear that the MIP electron density is in anti-phase 325 with the non-filtered magnetic field strength (cyan).
During the second interval of 0110 -0120 UT, the LAP P1 current acts as a proxy for the plasma density. In this case it is clear in Fig. 9 right panels that θ and ϕ are close to the MM criteria. The two strong dips in B m in the first 5 minutes show that as the field strength decreases the current increases.

330
This means that the mass-loading of the induced magnetosphere of comet 67P/CG created an unstable ion population through pick-up (a ring/ring-beam distribution), which relaxes through the generation of mirror-mode waves. Indeed, such a distribution was posited above when looking at the ion velocity direction with respect to the background magnetic field. The question whether such a distribution is able to develop in the cometosheath under the above conditions is addressed in the 335 discussion section below.
On 7 June, the MM structures have, on average, a time scale 100 ≤ T mm ≤ 150 s, which will be compared to a characteristic length scale of pick-up ions, being the Larmor radius. Assuming that the newly formed ions are picked up with the local (decelerated) solar wind velocity v SW , the gyro frequency ω c,i and radius ρ c,i are given by: also assuming that v ⊥ = v SW and that the structures are transported with v SW over the spacecraft and have a size of αρ c,i the time scale is given by:

345
This means that for these assumptions the solar wind velocity drops out of the equations and the crossing time is given by known and measured quantities. For water ions at a magnetic field strength of B m ≈ 20 nT this leads to T mm ≈ 9α s. With the measured T mm mentioned above this leads to 11 ≤ α ≤ 16, which is similar to what was found by Tsurutani et al. (1999) at comet 21P/Giacobini-Zinner, α GZ ≈ 12, but much larger than was found by Schmid et al. (2014)  For the interval 2230 -2240 UT it is clear that the size of the alleged MM structure is much smaller than in the later interval discussed above. An estimate from the inset panel in Fig. 9 shows that the MM structures have a timespan of ∼ 10 s. The field strength is slightly higher at B m ≈ 25 6 Change of MM shape 365 A closer look at Fig. 7 right panel shows that the structures, identified as mirror-mode waves, are changing in shape. Indeed, in the top panel the structures seem to be mainly dips in the magnetic field strength, B m , but at later times the structures seem to become asymmetric. A zoom-in on three intervals of 20 minutes is shown in Fig. 11; the data are shifted along the y-axis in order to make the difference between them more visible. The LAP P1 current is shown as grey asterisks overplotted 370 on each interval. The three intervals are different in behaviour: the first interval 0120 -0140 UT (blue) shows mainly strong dips in B m ; the second interval 0220 -0240 UT (green) shows strong asymmetric dips in B m and a large variety in structure sizes; the third interval 0320 -0340 UT (red) shows in the beginning deformation of the waves, strong periodic peaks with moving peaks super-imposed.

375
Spectral analysis is performed on these three intervals. It is clear from Fig. 11 right panel, that the three intervals have different spectral content: the first interval (blue) has a peak at f ≈ 6 mHz and a minor peak at f ≈ 13 mHz, the second interval (green) shows a plateau-like structure around f ≈ 10 mHz; the third interval (red) shows a clear double peaked structure at f ≈ 9 and f ≈ 19 mHz with a minor peak at f ≈ 51 mHz, which explains the beat-mode that can be seen in the red 380 trace in Fig. 11 left panel. It is not very clear from the LAP P1 current to deduce that these structures are mirror-modes, although the Lucek method indicates that they are.

Discussion and conclusions
For the first time in space research history a spacecraft is following a comet along its orbit from pre-to post-perihelion, entering regions around the comet that up to now had not been accessed.

385
Also the outgassing of comet 67P/CG at arrival in August 2014 was at a much lower level than for any other mission. During the period discussed in this paper the outgassing rate is around 10 27 molecules/s, which is orders of magnitude smaller than at comets 27P/Grigg-Skjellerup  or 1P/Halley (Reinhard, 1986). This means that the interaction of the solar wind with the outgassing comet is different, which was clearly illustrated through the discovery of the "singing 390 comet" by Richter et al. (2015), an unexpected plasma instability created by the not-yet-magnetized freshly produced ions near the comet. This is the context in which the results of this paper should be interpreted: measurements much closer to a cometary nucleus than ever before, with low outgassing rate and a very slowly moving spacecraft relative to the nucleus.
The data from RPC MAG have been calibrated, however Richter et al. (2015) state that: "The 395 short boom length implies that the spacecraft is heavily contaminating the magnetic field measurements. At this stage of the investigation it was not possible to completely remove these quasi-static spacecraft bias fields from the measured magnetic field values.". In this current paper, the observations of the diamagnetic cavity (Glassmeier et al., 2015;Götz et al., 2015) have been used to obtain values for non-corrected bias fields originating from the spacecraft. Assuming the diamagnetic cav-400 ity should be field-free (see e.g. Ip and Axford, 1987), the measured fields in the cavity have been subtracted from the data. This leads to a greatly improved determination of the mirror mode waves using the magnetic-field-only technique (Lucek et al., 1999), as the examples shown in Fig. 9 would not have been selected without bias-field offset correction.
The mass loading of the induced magnetosphere of comet 67P/CG, as indicated by the Rosetta -Before the increased density and the pile-up region there was a rotation of the magnetic field. This is probably related to changes in the field direction of the solar wind magnetic field, generating nested draped fields around the comet.
-Depending on the assumption how fast Rosetta crosses this structure the current densities in 420 the current sheet are tens or µA/m 2 or several nA/m 2 .
-There is increased ionization and energization of gas from the cometary nucleus in both IES and ICA.
-The magnetic field strength increased by a factor > 3 up to ∼ 60 nT, increasing the magnetic pressure. With the ion density on the order of 100 cm −3 and the ion temperature a few 10 5 K, 425 this means that the plasma beta β ∼ 10.
-The newly created ions are accelerated by the motional electric field, however, the effect only becomes apparent after the pile-up region is exited by the spacecraft.
-In the pile-up region there is evidence for mirror-mode structures, generated by the newly created ions, with a size between one and three water-ion gyro radii.

430
-Outside the pile-up region there are clear signatures of mirror-mode waves, with a much larger size of ten to sixteen water-ion gyro radii.
times there are three dominant frequencies present, which leads to strong deformation of the mirror-mode waves signature in the MAG data.

435
The above results leave a few points to discuss which will be addressed below.
-Nested draping: The change in direction of the magnetic field as observed in the Rosetta data does not show up clearly in the propagated solar wind magnetic field. The Tao-tangential field seems to go negative for a short period in the non-shifted data in Fig. 4 at the beginning of 6 June.

440
The Opitz-Dósa-radial magnetic field basically shows a heliospheric current sheet crossing.
Because of the draping and hanging-up of the magnetic field around the comet, it is difficult to find a one-to-one correlation between the solar wind field signatures and the draped field signatures. The layer of differently directed field at Rosetta may be the result of an older interval outside that presented in the figure. The difference in field strength can be explained 445 through the compression by the solar wind pressure.
-Changes in the magnetic pile-up region: Rosetta is located well inside the MPR of comet 67P/CG, which is clear from the high mag- where the 4 fold increase in dynamic pressure leads to a magnetic field strength increase by a factor ∼ 2.5 from ∼ 20 nT to ∼ 55 nT. This agrees well with the expected increase, which 455 would be √ P dyn,max /P dyn,min .
-Ionization increase: Looking at a longer data set of the IES ion energy spectra, it is clear that this increase in counts and energy of the ions is limited to a period of ≤ 18 hours, which corresponds to the increased solar wind dynamic pressure, which is caused by an increase of the solar wind 460 density. This means that an enhanced number of solar wind electrons is also entering the pileup region, which increases collisions and ionization as observed by RPC. After this period the IES densities follow the periodicity in the COPS neutral density, indicating that the increased ionization was indeed generated by the higher solar wind density.

465
A ring/ring-beam distribution is assumed necessary for the generation of the mirror mode waves. However, do the pick-up ions have enough time to develop such a distribution? The IES ion velocity in the increased pile-up region shows that the ions are basically moving perpendicular to the magnetic field. With a magnetic field strength between 20 and 55 nT and a velocity of ∼ 12 km/s the gyro frequency is 0.1 ≤ ω ci ≤ 0.25 s −1 and the gyro radius is 470 50 ≤ ρ ci ≤ 120 km. In order for a ring distribution to occur, the collision frequency must be much smaller than the gyro frequency. The collisional time is given by where σ i ∝ 10 −16 m 2 is a typical ion-neutral collisional cross-section (A' Hearn and Festou, 1990). Using typical values n ≥ 10 6 cm −3 and v = 10 km/s this leads to τ coll ≈ 10 3 − 10 4 s.
With a gyro period of 25 ≤ τ ci ≤ 60 s this means there is ample time for the ions to create a 475 ring-beam distribution and the location of Rosetta with respect to the comet, ∼ 225 km shows that the coma is large enough for full gyrations of the ions with the gyro radii mentioned above.
-Different sizes of MMs: Within the pile-up region, in the second half of 6 June, at high density, the mirror mode waves 480 are between one and three water gyro radii in size. This is "as expected" from newly created H 2 O + , as measured e.g. at comet 1P/Halley . Many hours later, on 7 June, there are much larger MM structures in the MAG data, with a size between 10 and 16 gyro radii. The larger structures could possibly be generated by diffusion of smaller size MMs as described by Hasegawa and Tsurutani (2011): where the source size has been changed to αρ c,i . Putting in the measured values (λ(L) = 14, α = 2, u = 10 km/s) and solving for the diffusion distance L ∼ 10 5 km shows that the large structures cannot have evolved from diffusion of the small structures in the pile-up region. Thus these large structures find their origin in MMs created further upstream in the 490 comet's coma. Where exactly cannot be determined as the source size α of the MMs further upstream is unknown.
-Structure deformation: The main ion species discussed in this paper is H 2 O + , however, CO + and CO + 2 were almost equal to that of water. Hässig et al. (2015) showed that the detector signal of the ROSINA 495 instrument for all three species was on average ∼ 2 × 10 5 particles/20 s, with only variations depending on which side of the comet is facing Rosetta. Assuming that the two main frequencies in the spectrum of the third interval in Fig. 11, with deformed (beating?) MM waves, are related to gyro frequencies of pick-up ions, then the ratio of the frequencies should possibly be related to the ratio of the masses of the ions. The low frequency waves are at ∼ 9 and ∼ 19 500 mHz, which have a frequency-ratio of ∼ 0.47, the mass-ratio of water with carbon(di)oxide is 0.6/0.41. The ratios are close, which might suggest that there are indeed different kinds of MMs at the same time. This would ask for an interaction of multiple kinds of MMs in one multi-component plasma, which has not been discussed in the literature.
The Rosetta mission around comet 67P/Churyumov-Gerasimenko offers excellent opportunities 505 to investigate processes that have been observed during flybys of other comets. Due to the slow motion of the spacecraft with respect to the comet an in-depth view is obtained of the interaction of the solar wind with the outgassing comet. This paper gives a "short" first discussion of a twoday interval of the data. With the spacecraft in basically the same location near comet 67P/CG this gave the possibility to study the reaction of the induced magnetosphere with respect to the increased 510 solar wind dynamic pressure. Furthermore, in this way temporal variations in the cometosheath, e.g.
the changes in the characteristics of the mirror mode waves were studied. Numerical modeling of the events showed in this paper is underway, as well as theoretical investigations into the various mirror-mode waves in a multi-ion pick-up plasma.
Acknowledgements. Rosetta is an ESA mission with contributions from its Member States and NASA. We Lajos Földy for his computational support. The authors acknoledge the ACE and OMNI databases for solar wind data.            Right panel: The Fourier power spectra for the three intervals. The coloured arrows at the top mark the peaks discussed in the text.