Use of MESSENGER radioscience data to improve planetary ephemeris and to test general relativity
Ashok Verma, Agnes Fienga, Jacques Laskar, Herve Manche, Mickael Gastineau
AAstronomy & Astrophysics manuscript no. messenger September 26, 2018(DOI: will be inserted by hand later)
Use of MESSENGER radioscience data to improve planetaryephemeris and to test general relativity.
A. K. Verma , , A. Fienga , , J. Laskar , H. Manche , and M. Gastineau Observatoire de Besanc¸on, UTINAM-CNRS UMR6213, 41bis avenue de l’Observatoire, 25000 Besanc¸on, France Centre National d’Etudes Spatiales, 18 avenue Edouard Belin, 31400 Toulouse, France Observatoire de la Cote d'Azur, G´eoAzur-CNRS UMR7329, 250 avenue Albert Einstein, 06560 Valbonne, France Astronomie et Syst`emes Dynamiques, IMCCE-CNRS UMR8028, 77 Av. Denfert-Rochereau, 75014 Paris, FranceSeptember 26, 2018
Abstract.
The current knowledge of Mercury orbit has mainly been gained by direct radar ranging obtained from the 60s to1998 and by five Mercury flybys made by Mariner 10 in the 70s, and MESSENGER made in 2008 and 2009. On March 18, 2011,MESSENGER became the first spacecraft to orbit Mercury. The radioscience observations acquired during the orbital phaseof MESSENGER drastically improved our knowledge of the orbit of Mercury. An accurate MESSENGER orbit is obtained byfitting one-and-half years of tracking data using GINS orbit determination software. The systematic error in the Earth-Mercurygeometric positions, also called range bias, obtained from GINS are then used to fit the INPOP dynamical modeling of theplanet motions. An improved ephemeris of the planets is then obtained, INPOP13a, and used to perform general relativity testsof PPN-formalism. Our estimations of PPN parameters ( γ and β ) are more stringent than previous results. Key words. messenger - celestial mechanics - ephemerides - general relativity
1. Introduction
Mercury is the smallest and least explored terrestrial planet ofthe solar system. Mariner 10 was the first spacecraft to makethree close encounters (two in 1974 and one in 1975) to thismysterious planet, and it provided most of our current knowl-edge of the planet until early 2008 (Smith et al. 2010). In ad-dition to Mariner 10 flyby observations, ground-based radarmeasurements were the only observations to be used to studyMercury’s gravity field and its physical structure (sphericalbody with slight flattening at the poles and a mildly elongatedequator) (Anderson et al. 1987, 1996). In 2004, the NationalAeronautics and Space Administration (NASA) launched adedicated mission, MErcury Surface, Space ENvironment,GEochemistry, and Ranging (MESSENGER), to learn moreabout this planet. MESSENGER made three close encounters(two in 2008 and one in 2009) to Mercury and became the firstspacecraft to observe Mercury from its orbit.Untill now, MESSENGER has completed more than twoyears on orbit at Mercury. During the orbital period, radiotracking of MESSENGER routinely measured the Doppler andrange observables at Deep Space Network (DSN) stations.These observables are important for estimating the spacecraftstate vectors (position and velocity) and improving the knowl-edge of Mercury’s gravity field and its geophysical proper-ties (Srinivasan et al. 2007). Using the first six months of ra-
Send o ff print requests to : A. Verma, [email protected] dioscience data during the orbital period, Smith et al. (2012)computed the gravity field and gave better constraints on theinternal structure (density distribution) of Mercury. This up-dated gravity field becomes crucial for the present computationof MESSENGER orbit and for performing precise relativistictests.The primary objectives of this work are to determine theprecise orbit of the MESSENGER spacecraft around Mercuryusing radioscience data and then to improve the planetaryephemeris INPOP (Fienga et al. 2008, 2009, 2011a). The up-dated spacecraft and planetary ephemerides are then used toperform sensitive relativistic tests of the Parametrized PostNewtonian (PPN) formalism (Will 1993, 2001, 2006).Nowadays, spacecraft range measurements are the mostaccurate measurements used for constructing planetaryephemerides. These measurements cover approximately 56%of all INPOP data (Fienga et al. 2011a) and impose strongconstraints on the planet orbits and on the other solar sys-tem parameters, including asteroid masses. However, un-til now, only five flybys (two from Mariner 10 and threefrom MESSENGER) range measurements have been avail-able for imposing strong constraints to Mercury’s orbit (Fiengaet al. 2011a). Therefore, range measurements obtained byMESSENGER spacecraft during its mapping period are impor-tant for improving our knowledge of Mercury’s orbit.Moreover, high-precision radioscience observations alsoo ff ered an opportunity to perform sensitive relativistic tests by a r X i v : . [ a s t r o - ph . E P ] N ov Verma et al.: Planetary ephemeris construction and test of relativity with MESSENGER estimating possible violation of the two relativistic parame-ters ( γ and β ) of the PPN formalism of general relativity (GR)(Will 1993). The previous estimations of these parameters us-ing di ff erent techniques and a di ff erent data set, can be foundin Bertotti et al. (2003); M¨uller et al. (2008); Pitjeva (2009);Williams et al. (2009); Manche et al. (2010); Konopliv et al.(2011); Fienga et al. (2011a). However, because of Mercury’srelatively high eccentricity and its close proximity to the Sun,its orbital motion provides one of the best solar system testsof GR (Anderson et al. 1997). In addition, Fienga et al. (2010,2011a) also demonstrated, Mercury observations are far moresensitive to PPN modification of GR than other data used in theplanetary ephemerides. We, therefore, also performed the testof GR with the latest MESSENGER observations to obtain oneof the most precise value for PPN parameters.In this paper, we introduce the updated planetary ephemerisINPOP13a and summarize the technique used for estimat-ing the PPN parameters. The outline of the paper is asfollows Section 2 discusses the radioscience data analysisof the MESSENGER spacecraft. The dynamic modeling ofMESSENGER and the results obtained during orbit computa-tion are also discussed in the same section. In section 3, we dis-cuss the construction of INPOP13a using the results obtainedin section 2. In section 4, we discuss the gravitational tests us-ing updated MESSENGER and Mercury ephemerides. Section5 follows with conclusions and perspectives.
2. MESSENGER data analysis
Under NASA’s Discovery program, the MESSENGER space-craft is the first probe to orbit the planet Mercury. It waslaunched in August 3, 2004, from Pad B of Space LaunchComplex 17 at Cape Canaveral Air Force Station, Florida,aboard a three-stage Boeing Delta II rocket. On March 18,2011, MESSENGER successfully entered Mercury’s orbit af-ter completing three flybys of Mercury following two flybys ofVenus and one of Earth (Solomon et al. 2007).The MESSENGER spacecraft was initially inserted into a ∼ ∼ ∼
12 to ∼ ± / s over 10s to several minutes of integrationtime (Srinivasan et al. 2007). We have analyzed one-and-half years of tracking data collectedby the DSN during the MESSENGER orbital period. Thesedata belong to one year of the prime mission and six monthsof the first extended mission (see Table 1). The complete dataset that was used for the analysis is available on the Geosciencenode of the NASA’s Planetary Data System (PDS). For preciseorbit determination, all available observations were analyzedwith the help of the G´eod´esie par Int´egrations Num´eriquesSimultan´ees (GINS) software, which was developed by theCentre National d’Etudes Spatiales (CNES) in collaborationwith Royal Observatory of Belgium (ROB). GINS numericallyintegrates the equations of motion and the associated varia-tional equations. It simultaneously retrieves the physical pa-rameters of the force model using an iterative least-squarestechnique. The precise orbit determination is based on a full dynamicalapproach. The dynamic modeling includes gravitational (grav-itational attraction of Mercury, third-body gravity perturbationsfrom the Sun and other planets, and relativistic corrections)and nongravitational (solar radiation pressure; Mercury radia-tion pressure) forces that are acting on the spacecraft. Theseforces have been taken into account in the force budget ofMESSENGER. The latest spherical harmonic model (Smithet al. 2012) of Mercury’s gravity field, HgM002 developedup to degree and order 20, and the associated Mercury’s orien-tation model (Margot 2009) have been considered for precisecomputation.The measurement (Doppler and range) models and thelight time corrections that are modeled in GINS correspondto the formulation given by Moyer (2003). During computa-tions, DSN station coordinates were corrected from the Earth’spolar motion, from solid-Earth tides, and from the ocean load-ing. In addition to these corrections, radiometric data have alsobeen corrected from tropospheric propagation through the me-teorological data (pressure, temperature, and humidity) of thestations.The complex geometry of the MESSENGER spacecraftwas treated as a combination of flat plates arranged in the shapeof a box, with attached solar arrays, the so-called Box-Wing macro-model. The approximated characteristics of this macro-model, which includes cross-sectional area and specular anddi ff use reflectivity coe ffi cients of the components, were taken http://pds-geosciences.wustl.edu/messenger/ http://pds-geosciences.wustl.edu/missions/messenger/rs.htm http://pds-geosciences.wustl.edu/messenger/mess-v_h-rss-1-edr-rawdata-v1/messrs_0xxx/ancillary/wea/ erma et al.: Planetary ephemeris construction and test of relativity with MESSENGER 3 from (Vaughan et al. 2002). In addition to the macro-modelcharacteristics, orientations of the spacecraft were also takeninto account. The attitude of the spacecraft and of its articulatedpanels in inertial space were defined in terms of quaternions.The approximate value of these quaternions was extracted fromthe C-kernel system of the SPICE Navigation and AncillaryInformation Facility (NAIF) software. The macro-model andits orientation allowed calculation of the nongravitational ac-celerations that are acting on the MESSENGER spacecraft dueto the radiation pressure from Sun and Mercury (albedo andthermal infrared emission).For orbit computation and parameters estimation, a multi-arc approach was used to get independent estimates of theMESSENGER accelerations. In this method, we integrated theequations of motion using the time-step of 50s then, and orbitalfits were obtained from short data arcs fitted over the observa-tions span of one day using an iterative process. The short dataarcs of one day were chosen to account for the model imperfec-tions. To initialize the iteration, the initial position and velocityvectors of MESSENGER were taken from the SPICE NAIFspk-kernels . An iterative least-squares fit was performed on the complete setof Doppler- and range-tracking data arcs that correspond to theorbital phase of the mission using an INPOP10e (Fienga et al.2013) planetary ephemeris . We have processed data from May17 2011 to September 18 2012 excluding the periods of themaneuvers. A summary of these tracking data is given in Table1. MESSENGER fires small thrusters to perform momentumdump maneuver (MDM) for reducing the spacecraft angularmomentum to a safe level. Normal operations (during orbitalperiods) includes only one commanded momentum dump ev-ery two weeks. In addition, orbit correction maneuvers (OCM)were also performed (typically once every Mercury year, 88Earth days) to maintain the minimum altitude below 500 kilo-meters. Such large intervals between the MESSENGER ma-neuvers facilitate the orbit determination. The data arcs thatcorrespond to the maneuver epochs are thus not included inthe analysis. The total 440 one-day data arcs were then usedfor the analysis.Several parameters were estimated during orbit computa-tion: spacecraft state vectors at the start epoch of each data arc,for a total of 440 × = (cid:80) × nparameters, where n is the number of stations participating inthe data arc); one station bias per arc for each DSN station toaccount for the uncertainties on the DSN antenna center posi- ftp://naif.jpl.nasa.gov/pub/naif/pds/data/mess-e_v_h-spice-6-v1.0/messsp_1000/data/ck/ ftp://naif.jpl.nasa.gov/pub/naif/pds/data/mess-e_v_h-spice-6-v1.0/messsp_1000/data/spk/ tion or the instrumental delays (total of (cid:80) × n parameters);and one range bias per arc for ranging measurements to accountfor the systematic geometric positions error (ephemerides bias)between the Earth and the Mercury (total of 440 parameters). The root mean square (rms) values of the post-fitted Dopplerand range residuals give some indication about the quality ofthe orbit fit and the estimated parameters. Moreover, the qualityof the used parameters associated to the physical model canalso be judged from these residuals. Figure 1 illustrates the timehistory of the residuals estimated for each measurement type.In this figure, panel a represents the rms values of the two-and three-way Doppler residuals that were obtained for eachdata arc and are expressed in millihertz (mHz). As Mercuryhas shorter orbit than other planets, it experiences five superiorconjunctions (when the Earth, the Sun and the spacecraft lieon the same line, with the spacecraft located on the oppositeside of the Sun with respect to Earth) during the time intervalcovered by the analysis. Because of a lack of modelisation ofthe solar corona perturbations within the GINS software, nomodel of solar plasma was applied during the computations ofthe MESSENGER orbit. The peaks shown in Fig. 1, therefore,demonstrate the clear e ff ect of the solar conjunctions on thetypical fit to the Doppler and range data residuals.Excluding the solar conjunction periods (about 100 dataarcs), when Sun-Earth-Probe angle remained below 10 ◦ , an av-erage value of Doppler residuals has been found to be approx-imately 4.8 ± ∼ ± / s), which is compara-ble with values given by (Smith et al. 2012; Stanbridge et al.2011; Srinivasan et al. 2007). The mean value of the estimatedDoppler bias for each DSN station tracking pass was found tobe very small (a few tenths of mHz), which is lower than theDoppler post-fit residuals for each data arc. It demonstrated thatwe have no large bias in the modeling of the Doppler shift mea-surements at each tracking station.The range measurements were also used to assist in fittingthe Doppler data for a precise orbit determination. Panel b ofFig. 1 represents the rms values of two-way range residualsthat were obtained for each data arc. An average value of theserange residuals is 1.9 ± We fitted one scale factor per data arc for the solar radiationforce to account the inaccuracy in the force model. Panel a of Fig. 2 represents the time history of these scale factors.These scale factors are overplotted with the beta angle , whichis the angle between MESSENGER orbital plane and the vec-tor from the Sun direction. The variation in the MESSENGERorbital plane (beta angle) relative to the Sun occurs as Mercurymoves around the Sun. For example, at 10 ◦ , 100 ◦ , and 180 ◦ of Mercury’s true anomaly, the corresponding beta angles are83 ◦ , 0 ◦ , and -78 ◦ , respectively (Ercol et al. 2012). At 0 ◦ beta Verma et al.: Planetary ephemeris construction and test of relativity with MESSENGER
Table 1: Summary of the Doppler and range tracking data used for orbit determination.
Mission Begin date End date Number of Number of Number ofphase dd-mm-yyyy dd-mm-yyyy 2-way Doppler 3-way Doppler rangePrime 17-05-2011 18-03-2012 2108980 184138 11540Extended 26-03-2012 18-09-2012 1142974 23211 5709
Fig. 1: Quality of the MESSENGER orbit in terms of rms values of the post-fit residuals for each one-day data arc: (a) two- andthree-way Doppler given in millihertz (multiply by 0.0178 to obtain residuals in mm / s), and (b) two-way range given in meters.angle, the spacecraft travels directly between the Sun and theplanet, while at 90 ◦ , the spacecraft is in sunlight 100% of thetime. As one can see from panel a of Fig. 2, the solar pressurecoe ffi cients have variations that approximately follow those ofthe beta angle. This implies that, whenever MESSENGER or-bital plane approaches the maximum beta angle, it is fully il-luminated by direct sunlight (no shadow a ff ect). To protect thespacecraft from the direct sunlight, the automatic orientationof the solar panels therefore balances the need for power andthe temperature of the surface of the panel. Thus, imperfectionin the modeling of these orientations is then taken care of bythe scale factor to reduce the error in the computation of solarradiation pressure (see Fig. 2). The fitted scale factor for solarradiation pressure is, therefore, typically in the range of about2.1 ± a priori value and itreflects the imperfection in the force model due to the approxi-mate representation of the macro-model.Panel b of Fig. 2 illustrates the one-way range bias esti-mated for the ranging measurements for each data arc. Thesebiases represent the systematic uncertainties in the Earth-Mercury geometric positions. The black ( • ), brown ( (cid:78) ) and blue ( (cid:72) ) bullets in this figure correspond to INPOP10e (Fiengaet al. 2013), DE423 (Folkner 2010), and DE430 (Williams et al.2013), respectively. An average value of these range bias forINPOP10e, DE423 and DE430 is 21 ±
187 m, 15 ±
105 m, and -0.5 ±
42 m, respectively. This range bias is then used in the plan-etary ephemerides to fit the dynamical modeling of the planetmotions (see Sec. 3). Thus, MESSENGER ranging measure-ments were used to reconstruct the orbit of Mercury around theSun. The improved planetary ephemeris, INPOP13a (see Sec.3.1) was then used to re-analyze the MESSENGER radiometricdata to study the impact of planetary ephemeris over the com-putation of MESSENGER orbit and associated parameters (seeSec. 3.2).
Planetary ephemerides are a good tool for testing the grav-ity model and GR (Fienga et al. 2011a) and performing solarcorona studies (Verma et al. 2013). Moreover, it is also pos-sible to calibrate the transponder group delay from the plane-tary ephemeris. The spacecraft receives and transmits the signal erma et al.: Planetary ephemeris construction and test of relativity with MESSENGER 5
Fig. 2: History of the fitted scale factor and estimated range bias: (a) scale factor (for solar radiation acceleration) fitted overeach one-day data arc to account inaccuracy in the force model, and (b) one-way range bias, represent the systematic error in theEarth-Mercury positions, estimated for each one-day arc using INPOP10e ( • ), DE423 ( (cid:78) ), and DE430 ( (cid:72) ) planetary ephemerides.to the Earth station through the on-board transponder, whichcauses the time delay in the range measurements. This delayvaries from one spacecraft to another depending on the radiofrequency configuration. Usually an average value for this de-lay is measured at di ff erent occasions on the ground beforelaunch. However, the group delay is not perfectly stable andcan fluctuate by a few ns, depending upon variations in a num-ber of parameters such as temperature and signal strength.For MESSENGER, we estimated this group delay withthe planetary ephemeris. This procedure becomes an alter-nate method of testing the procedure and quality of the orbitfit by comparing estimated group delay with the delay testedon the ground. Since the transponder delay does not a ff ectthe Doppler measurements we were therefore, able to com-pute the precise orbit of the spacecraft without considering thetransponder delay in the range measurements. With this config-uration, we then reanalyzed the entire radio tracking data (seeTable 1). To check the precision on the knowledge of the space-craft orbit, we compared the radial, along-track, and cross-trackcomponents of the orbit for each data arc with the solution ob-tained in Section 2.2.1. An average rms value of radial, along-track, and cross-track di ff erence is 0.015m, 0.16m, and 0.19m,respectively. Less than a meter level of di ff erences in the or-bit implies that the transponder delay has negligible impact onthe orbit, since the spacecraft orbit is mostly constrained bythe Doppler tracking data. However, there is a dramatic changein the estimation of range bias, which now includes ephemerisbias plus the bias due to the transponder delay. Using theserange biases to fit the planetary ephemeris, we found a clear o ff -set in the Earth-Mercury geocentric distances of about 410 ±
20 m (two-way) during the orbital period of the MESSENGER.This estimation of transponder delay is compatible with the onefound during ground testing, which ranged from 1,356.89 ns ( ∼
407 m) to 1,383.74 ns ( ∼
415 m) (Srinivasan et al. 2007). Thusthese results also suggested that there is not a large error in thespacecraft and the planetary orbit fit procedure.
3. Improvement of planetary ephemeris, INPOP
Since 2003, INPOP planetary ephemerides have been built on aregular basis and provided to users thought the IMCCE website . The INPOP10e ephemeris was the lat-est release (Fienga et al. 2013) that was delivered as the o ffi cialGaia mission planetary ephemerides used for the navigation ofthe satellite as well as for the analysis of the data. Specific de-velopments and analysis were done for the Gaia release suchas the TCB time-scale version or an accurate estimation of theINPOP link to ICRF. With the delivery of the MESSENGERradio science data, a new opportunity was o ff ered to improvedrastically our knowledge of the orbit of Mercury and to per-form tests of gravity at a close distance from the Sun.The use of the 1.5 year range measurements deduced fromthe previous analysis ( see section 2) is then a crucial chanceto obtain better knowledge over the ∼ ff ects. Verma et al.: Planetary ephemeris construction and test of relativity with MESSENGER
The constants and dynamical modeling used for constructingthe new ephemerides, INPOP13a, are similar to INPOP10e. Acomplete adjustment of the planet initial conditions (includingPluto and the Moon), the mass of the Sun, the oblateness ofthe Sun, the ratio between the mass of the Earth and the Moon,and 62 asteroid masses is operated. Values of the obtained pa-rameters are given in Tables 2 and 3. Even if Mercury is notdirectly a ff ected by the main belt asteroids, the use of the rangemeasurements between MESSENGER and the Earth does havean impact on the Earth’s orbit and then could bring some infor-mation on asteroid masses perturbing the Earth orbit. On Table3, we only gave the masses that are significantly di ff erent fromthose obtained with INPOP10e and inducing detectable signa-tures below five meters. These masses are also compared withthe Konopliv et al. (2011) on the same table. The masses ofthe biggest objects di ff er from the two ephemerides inside theirtwo-sigma error bars, and one can notice the new determinationof the mass of (51) Nemausa inducing slightly bigger perturba-tions on Mercury (7 meters) and Venus (8 meters) geocentricdistances than on Mars (5 meters).Table 4 gives the postfit residuals obtained with INPOP13aand compared with those obtained with INPOP10e. One cansee a noticeable improvement in Mercury’s orbit over all theperiods of the fit including direct radar observations. The resultis of course more striking for MESSENGER range measure-ments that were deduced from section 2, and not used for the fitof INPOP10e. In this particular case, the improvement reachesa factor of almost 16 on the estimation of the distance betweenMercury and the Earth (see Figure 3). The extrapolated residu-als given in Table 4 are not really significant since INPOP10ewas fitted over a very similar interval of time ending at about2010.4 when INPOP13a was fitted up to 2011.4.Figure 4 plots the di ff erences between INPOP13a,INPOP10e and DE423 for Mercury geocentric right ascension,declination, and distance, and the Earth-Moon barycenter lon-gitudes, latitudes, and distances in the BCRS. These di ff er-ences give estimations of the internal accuracy of INPOP13a.By comparison, the same di ff erences between INPOP10aand DE421 are also plotted. They present the improvementsreached since INPOP10a, clearly noticeable for the Mercurygeocentric distances (a factor two between INPOP13a-DE423and INPOP10a-DE421). They are less impressive for the EMB;however, one can notice that the clear systematic trend in theINPOP10a-DE423 barycentric distances of the EMB is re-moved in INPOP13a-DE423. The fact that the di ff erences be-tween INPOP13a and INPOP10e are smaller than the di ff er-ences to DE ephemerides is mainly discussed in Fienga et al.(2011a) and Fienga et al. (2013) by a di ff erent method of com-puting the orbit of the Sun relative to the solar system barycen-ter, as well as a di ff erent distribution of planetary and asteroidmasses.In conclusion, INPOP13a shows an important improvementin the Mercury orbit especially during the MESSENGER or-bital and flyby phases of the mission. The improvement overthe EMB orbit in the BCRS is less important but still a sys-tematic trend noticeable in the EMB barycentric distance dif- ferences between INPOP10a and DE421 seems to be removedin the new comparisons. As given in Table 4, geometric distances between Earthand Mercury are ∼
16 times better in INPOP13a than theINPOP10e. To analyze the impact of the improvement of theplanetary ephemeris on the spacecraft orbit, we reanalyzed theentire one and half years of radioscience data (see Table 1) us-ing INPOP13a ephemeris. The dynamical modeling and orbitdetermination process for this analysis are the same as dis-cussed in section 2.1. To compare the results of this analysiswith the one obtained from INPOP10e (see Section 2.2.1), thedi ff erences in the Doppler and range postfit residuals alongwith the changes that occurred in the periapsis and apoapsisaltitudes of MESSENGER are plotted in Fig. 5.An average value of these di ff erences and its 1 σ meandispersion for Doppler, and range postfit residuals was esti-mated as 0.008 ± ± ± ± ff erences in theperiapsis δ p and apoapsis δ a altitudes of MESSENGER due tothe change in planetary ephemeris are plotted in panels c and d of Fig. 5. An average and 1 σ dispersion of δ p and δ a was foundas 0.05 ± ±
4. Test of general relativity
INPOP13a was built in the framework of GR in using theParametrized Post-Newtonian formalism (PPN). A detailed de-scription of the modeling used for the INPOP ephemerides isgiven in (Fienga et al. (2011a)). Specific relativistic timescaleshave been set up and integrated in INPOP (TCB, TDB) andmainly two parameters characterized the relativistic formalismin modern planetary ephemerides: the parameter β that mea-sures the nonlinearity of gravity and γ , measuring the deflex-ion of light. In GR, both are supposed to be equal to 1 and werefixed to 1 for the INPOP13a construction. The GINS softwareused for the analysis of the radio science data and the recon-struction of the MESSENGER orbit is also coded in the PPNframework, including both β and γ PPN parameters.Up to now, general relativity theory (GRT) has success-fully described all available observations, and no clear obser-vational evidence against GR has been identified. However, thediscovery of Dark Energy which challenges GRT as a com-plete model for the macroscopic universe, and the continuing erma et al.: Planetary ephemeris construction and test of relativity with MESSENGER 7
Fig. 3: MESSENGER one-way range residuals obtained with INPOP13a, INPOP10a, DE423, and DE430.Table 2: Values of parameters obtained in the fit of INPOP13a and INPOP10e to observations including comparisons to DE423and DE430.
INPOP13a INPOP10e DE423 DE430 ± σ ± σ ± σ ± σ (EMRAT-81.3000) × − (5.770 ± ± ± ± (cid:12) × − (2.40 ± ± ± (cid:12) - 132712440000 [km . s − ] (48.063 ± ± × [m] 9.0 9.0 (-0.3738 ± Table 3: Asteroid masses obtained with INPOP13a, significantly di ff erent from values found in INPOP10e, and inducing a changein the Earth-planets distances smaller than 5 meters over the fitting interval. The uncertainties are given at 1 published sigma andcompared with Konopliv et al. (2011). IAU designation INPOP13a INPOP10e Konopliv et al. (2011)number 10 x M (cid:12) x M (cid:12) x M (cid:12) ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± failure to merge GRT and quantum physics indicate that newphysical ideas should be sought. To streamline this search it isindispensable to test GRT in all accessible regimes and to thehighest possible accuracy.Among all possibilities for testing GRT, the tests of themotion and light propagation in the solar system were histori-cally the first ones, and they are still very important since theygive highest accuracy since the dynamics of the solar system is well understood and supported by a long history of observa-tional data. Concerning the Einstein field equations, the mostimportant framework used for the tests in the solar system isthe PPN formalism (such as Will (1993)). The PPN formal-ism is a phenomenological scheme with ten dimensionless pa-rameters covering certain class of metric theories of gravity,among them the β and γ parameters parts of the INPOP andGINS modelings. The tracking data of space missions give a Verma et al.: Planetary ephemeris construction and test of relativity with MESSENGER
Table 4: Statistics of the residuals obtained after the INPOP13a fit. For comparison, means and standard deviations of residualsobtained with INPOP10e are given. The label
GINS range indicates that the corresponding data set was obtained after orbitreconstruction of the spacecraft in using the GINS software. For MGS, see Verma et al. (2013).
Type of data Nbr Time Interval INPOP13a INPOP10emean 1- σ mean 1- σ Mercury range [m] 462 1971.29 - 1997.60 -108 866 -45 872Mercury Messenger GINS range [m] 314 2011.39 - 2012.69 2.8 12.0 15.4 191.8Out from SC * GINS range [m] 267 2011.39 - 2012.66 -0.4 8.4 6.2 205.2Mercury Mariner range [m] 2 1974.24 - 1976.21 -124 56 -52.5 113Mercury flybys Mess ra [mas] 3 2008.03 - 2009.74 0.85 1.35 0.73 1.48Mercury flybys Mess de [mas] 3 2008.03 - 2009.74 2.4 2.4 2.4 2.5Mercury flybys Mess range [m] 3 2008.03 - 2009.74 -1.9 7.7 -5.05 5.8Venus VLBI [mas] 46 1990.70 - 2010.86 1.6 2.6 1.6 2.6Venus range [m] 489 1965.96 - 1990.07 502 2236 500 2235Venus Vex range [m] 24970 2006.32 - 2011.45 1.3 11.9 1.1 11.9Mars VLBI [mas] 96 1989.13 - 2007.97 -0.02 0.41 -0.00 0.41Mars Mex range [m] 21482 2005.17 - 2011.45 -2.1 20.6 -1.3 21.5Mars MGS GINS range [m] 13091 1999.31 - 2006.83 -0.6 3.3 -0.3 3.9Mars Ody range [m] 5664 2006.95 - 2010.00 1.6 2.3 0.3 4.1Mars Path range [m] 90 1997.51 - 1997.73 6.1 14.1 -6.3 13.7Mars Vkg range [m] 1257 1976.55 - 1982.87 -0.4 36.1 -1.4 39.7Jupiter VLBI [mas] 24 1996.54 - 1997.94 -0.5 11.0 -0.3 11.0Jupiter Optical ra [mas] 6532 1914.54 - 2008.49 -40 297 -39 297Jupiter Optical de [mas] 6394 1914.54 - 2008.49 -48 301 -48 301Jupiter flybys ra [mas] 5 1974.92 - 2001.00 2.6 2.9 2.4 3.2Jupiter flybys de [mas] 5 1974.92 - 2001.00 -11.0 11.5 -10.8 11.5Jupiter flybys range [m] 5 1974.92 - 2001.00 -1065 1862 -907 1646Saturne Optical ra [mas] 7971 1913.87 - 2008.34 -6 293 -6 293Saturne Optical de [mas] 7945 1913.87 - 2008.34 -12 266 -2 266Saturne VLBI Cass ra [mas] 10 2004.69 - 2009.31 0.19 0.63 0.21 0.64Saturne VLBI Cass de [mas] 10 2004.69 - 2009.31 0.27 0.34 0.28 0.33Saturne Cassini ra [mas] 31 2004.50 - 2007.00 0.8 3.4 0.8 3.9Saturne Cassini de [mas] 31 2004.50 - 2007.00 6.5 7.2 6.5 7.2Saturne Cassini range [m] 31 2004.50 - 2007.00 -0.010 18.44 -0.013 18.84Uranus Optical ra [mas] 13016 1914.52 - 2011.74 7 205 7 205Uranus Optical de [mas] 13008 1914.52 - 2011.74 -6 234 -6 234Uranus flybys ra [mas] 1 1986.07 - 1986.07 -21 -21Uranus flybys de [mas] 1 1986.07 - 1986.07 -28 -28Uranus flybys range [m] 1 1986.07 - 1986.07 20.7 19.7Neptune Optical ra [mas] 5395 1913.99 - 2007.88 2 258 0.0 258Neptune Optical de [mas] 5375 1913.99 - 2007.88 -1 299 -0.0 299Neptune flybys ra [mas] 1 1989.65 - 1989.65 -12 -12Neptune flybys de [mas] 1 1989.65 - 1989.65 -5 -5Neptune flybys range [m] 1 1989.65 - 1989.65 66.8 69.6Pluto Optical ra [mas] 2438 1914.06 - 2008.49 -186 664 34 654Pluto Optical de [mas] 2461 1914.06 - 2008.49 11 536 7 539Pluto Occ ra [mas] 13 2005.44 - 2009.64 6 49 3 47Pluto Occ de [mas] 13 2005.44 - 2009.64 -7 18 -6 18Pluto HST ra [mas] 5 1998.19 - 1998.20 -42 43 33 43Pluto HST de [mas] 5 1998.19 - 1998.20 31 48 28 48Venus Vex ** range [m] 2827 2011.45 - 2013.00 51 124 52 125Mars Mex ** range [m] 4628 2011.45 - 2013.00 -3.0 11.5 4.2 27.5 * Solar corona period ** Extrapolation period good possibility to test GRT since the data is very sensitive tothe GRT-e ff ects in both dynamics of the spacecraft and sig-nal propagation. However, some factors, such as navigation unknowns (AMDs, solar panel calibrations), planet unknowns(potential, rotation, etc.), e ff ect of the solar plasma, or the cor-relation with planetary ephemerides limit this sort of gravity erma et al.: Planetary ephemeris construction and test of relativity with MESSENGER 9 Fig. 4: In Mercury panel, di ff erences in geocentric Mercury right ascension (RA), declination (DE) and distances betweenINPOP13a, INPOP10e and DE423. In EMB panel, di ff erences in BCRF longitudes, latitudes and distances of the EMB betweenINPOP13a, INPOP10e and DE423. Di ff erences between INPOP10a and DE421 are also given.test. Dynamics of the solar system are, however, less a ff ectedby poorly modeled accelerations and technical unknowns. Upto now, the best constraints for β come from the planetary datain INPOP (Fienga et al. (2011a)). Constraints on other PPNparameters can be found in Will (2006). A number of theo-retical models predict deviations of PPN parameters that aresmaller than current constraints. Typical examples here are cer-tain types of tensor-scalar theories where cosmological evolu-tion exhibits an attractor mechanism towards GRT (Damour& Nordtvedt (1993)) or string-inspired scalar-tensor theorieswhere the scalar field can decouple from matter (Damour &Polyakov (1994)).Another phenomenological test concerns the constancy ofthe Newtonian gravitational constant G in time. A variable Gis produced say by alternative theories of gravity such tensor-scalar theory (see e.g. Damour et al. (1990) and Uzan (2003))or some models of dark energy (Steinhardt & Wesley (2009);Alimi & F¨uzfa (2010)). The ratio is now constrained at the levelof 10 − with LLR analysis (Williams et al. (2004)). γ and β In this section, we propose to use the improvement ofMercury’s orbit as an e ffi cient tool for testing the consistencybetween planetary ephemerides built with MESSENGER radioscience data and non-unity PPN parameters. A first estimation of PPN β and γ is possible by leastsquare methods during the adjustment of the INPOP planetaryephemerides, and the results are given in Table 5. Figure 6 givesthe correlations between the first 71 over the 343 parametersestimated in the adjustments. As one can see in Fig. 6 no cor-relation greater than 0.3 a ff ects the determination of the PPNparameters β and γ , as well as the fit of the Sun oblateness,when the gravitational mass of the Sun is highly related to theMercury and to the Earth orbits.However to go further in the analysis of the uncertaintiesand the construction of acceptable intervals of violation of GRthrought the PPN β and γ , we also considered the same methodas the one that was used and described in Fienga et al. (2011a)for determining acceptable intervals of violation of GR whenthe PPN formalism. Small variations in these two parametersnear unity are imposed when constructing alternative plane-tary ephemerides that are fit over the whole data sets presentedin Table 4 and with the same parameters and hypothesis asINPOP13a. A minimum of three iterations in the adjustmentprocess is required for building new ephemerides, and compar-isons between these ephemerides and INPOP13a are done toscale up what variations to GR are acceptable at the level ofuncertainty of the present planetary ephemerides.The improvement of Mercury’s orbit in INPOP13a justifiesthese new estimations. Indeed, Mercury played a historical rolein testing gravity and GR in 1912 (Einstein 1912) and it is stillthe planet the most influenced by the gravitational potential of Fig. 5: Comparison between INPOP13a and INPOP10e estimations of MESSENGER orbit: (a) di ff erences in the postfit Dopplerresiduals; (b) di ff erences in the postfit range residuals; (c) di ff erences in the periapsis altitude δ p; (d) di ff erences in the apoapsisaltitude δ a.the Sun. Its orbit can then lead to the most e ffi cient constraintson β , hence on γ in the PPN formalism. Before the recent in-put of MESSENGER flyby and orbital radio science data in theINPOP construction, Mars was the most constraining planet forthe PPN parameters (Fienga et al. (2010)). The reason was thelong range of high accurate observations on Mars. The imple-mentation of the first MESSENGER flyby data reduces the in-terval of violation of β to 50 %. The first estimation of the γ interval of violation was made possible thanks to the gain inuncertainty on the Mercury orbit. With INPOP13a, even betterimprovement is achieved. The results obtained by direct least squares fit are presentedin Table 5. As expected, the estimated uncertainties are veryoptimistic and a more detailed analysis is done based on themethod proposed by Fienga et al. (2011a).Results obtained in terms of percentages of the variations inpostfit residuals between a planetary ephemeris fitted and builtup with PPN parameters di ff erent from one and INPOP13a aregiven in Figure 7. Panel (a) in Figure 7 gives the map of thevariations in percent of the full dataset postfit residuals. Panel(b) in Figure 7 gives the same map but without taking the vari-ations of the Mercury flyby data into account. For Panel (b),the Mercury flyby data are indeed used in the ephemerides fitbut not in the analysis of the postfit residuals for testing GR.The map of Panels (a) and (b) is then dramatically di ff erent:where the limits for β and γ are stringent for the map includ-ing the Mariner data, the constraints are greatly enlarged for Fig. 8: Di ff erences in geocentric Mercury distances betweenINPOP13a and a planetary ephemerides built with PPN β and γ di ff erent from 1. The indicated area shows intervals of timecorresponding to Mariner observations, MESSENGER and thefuture Bepi-Colombo.these two parameters. These phenomena were expected sincethe variations in PPN parameters induce long-term perturba- erma et al.: Planetary ephemeris construction and test of relativity with MESSENGER 11 Fig. 6: Correlation between the first 71 (over 343) parameters estimated during the fit of the planetary ephemerides. The red frameframes the correlations related to the initial conditions of planet orbits and the blue rectangle frames the correlations related tothe Sun J ( JS ), the PPN parameters β ( BE ) and γ ( GA ) and the gravitational mass of the Sun ( GS ). The magenta rectangleframes the correlations related to the gravitational masses of the first most perturbing asteroids including the gravitational massof the asteroid ring ( GR ). m , ... m expresses the initial conditions of the Mercury orbit in equinoctial coordinates: semi-majoraxis, mean motion, k = , h = , q = and p = respectively. The other planet initial conditions are indicated by the first letter of theplanet ( V for Venus, M for Mars etc...) and by the figures of the corresponding initial conditions as given for Mercury.tions in the geocentric distances of Mercury as one can seein Figure 8. Panels (c) and (d) are similar to (a) and (b), butthey are obtained with ephemeris INPOP10a. In this ephemeris,MESSENGER flyby data were included in its fits but not in theorbital data. By comparing panels (a) and (c), one can see thatthe use of the MESSENGER orbital data significantly reducethe intervals of violation for both PPN parameters by a factor10. The same manner, in the most pessimistic case and withoutconsidering the Mercury flybys in analysing of the variationsin the postfit residual, one can see in Panels (b) and (d) that theimprovement of Mercury’s orbit is again crucial for reductingthe violation intervals of PPN parameters.Table 5 collects the acceptable violation intervals ob-tained from INPOP10a and INPOP13a. Values extracted fromINPOP10a were obtained at 5% of postfit residual variations(Fienga et al. 2011a). With INPOP13a, we extracted values from i) Panel (a) of Figure 7 obtained at 5%, but also at 10%and 25% and ii) from Panel (b) of Figure 7 obtained at 5%,which is consistent with the 25% of intervals extracted fromPanel (a).All given intervals are compatible with GR with an uncer-tainty at least ten times smaller than our previous results withINPOP10a. In Table 5, comparisons to least squares estima-tions of other planetary ephemerides or Moon ephemerides likePitjeva & Pitjev (2013), Konopliv et al. (2011), M¨uller et al.(2008), and Williams et al. (2009), as well as estimations de-duced from VLBI observations Lambert & Le Poncin-Lafitte(2011), are also given. The most stringent published constraintfor the PPN parameter γ has been obtained so far during a ded-icated phase of the Cassini mission by Bertotti et al. (2003).This value is compatible with our 25% estimation when our Fig. 7: Variations in postfit residuals obtained for di ff erent values of PPN β (x-axis) and γ (y-axis). Panels (a) and (c) are obtainedby considering the variations in the whole data sets when for Panels (b) and (d), variations in the Mercury flyby data (fromMariner and MESSENGER missions) are excluded from the analysis. The dashed line indicates the limit in γ given by Bertottiet al. (2003).5% and 10% estimations give more restrictive intervals of GRviolations.Confirmations of the results presented in Table 5 will beobtained by the use of the radioscience data obtained duringthe future Bepi-Colombo mission. In addition, the recovery ofthe Mariner flyby data would also be a great help for such con-firmations. Unfortunately, the Mariner data seem to have beenlost, and access to these data seems to be unrealistic. Indeedas one can see in Figure 8, perturbations induced by a slightchange in the PPN parameters (( β -1) = × − and ( γ -1) = -2.2 × − ) inducing an e ff ect of about six meters on the Marinerrange data (12%) will induce a signature of about the same level at the Bepi-Colombo epoch. With the improved Bepi-Colombo radio science tracking, the expected accuracy in therange measurement is planned to be about 50 centimeters. Withsuch accuracy, detecting the perturbations induced by the samemodification of the PPN parameters should be done at 1200%!Two orders of magnitude are expected as a gain in the uncer-tainty for the β and γ estimations. As stated previously, the GINS software was modeled in theframework of the PPN formalism which includes β and γ pa- erma et al.: Planetary ephemeris construction and test of relativity with MESSENGER 13 Table 5: Intervals of violation for PPN parameters β and γ deduced from Figure 7 panel (a) labelled INPOP13a and Figure 7panel (b) labelled INPOP13aWF. Values from INPOP10a are extracted from Fienga et al. (2010) and Fienga et al. (2011a) witha threshold for the variations of the postfit residuals of 5% given in Column 4. K11 stands for Konopliv et al. (2011), M08 forM¨uller et al. (2008), W09 for Williams et al. (2009), B03 for Bertotti et al. (2003), P13 for Pitjeva & Pitjev (2013) and L11 forLambert & Le Poncin-Lafitte (2011). The Least squares section gives the fitted values of β and γ at 1 σ as obtained by a globalfit of INPOP presented in section 4.2.1. Ref. ( β − × ( γ −
1) INPOP13a Limit [%] ( β − × ( γ − × × All data 25 ( β -1) = (0.2 ± β -1) = (-6.2 ± γ -1) = (-0.3 ± γ -1) = (4.5 ± β -1) = (-0.15 ± γ -1) = (0.0 ± β -1) = (4 ±
24) 5 ( β -1) = (0.02 ± γ -1) = (18 ±
26) ( γ -1) = (0.0 ± β -1) = (15 ±
18) Least squares ( β − ∗∗ = (1.34 ± β -1) = (12 ±
11) ( γ − ∗∗ = (4.53 ± γ -1) = (2.1 ± β -1) = (-0.5 ± γ -1) = (-8 ±
12 ) ( γ -1) = (12.5 ± β -1) = (0.0 ± β -1) = (-2 ±
3) ( γ -1) = (0.5 ± γ -1) = (4 ±
6) 5 ( β -1) = (-0.25 ± γ -1) = (-0.1 ± γ − ∗∗ least square results given at 1 σ rameters. To analyze the combined impact of PPN parametersover the MESSENGER orbit and the planetary ephemeridesconstruction, we analyzed the entire one and half years of ra-dioscience data again using the PPN parameters that are di ff er-ent from unity. The same procedure as described in section 2.1has been used for reconstructing the MESSENGER orbit. Twosets of MESSENGER orbits were then built, one with β and γ equal to unity and the other with β and γ di ff erent from unity(in this case, β -1 = γ -1 = × − ). Hereafter, the solution ob-tained from β and γ equal to unity is referred to as SOL Re f (forthe reference solution), and the solution corresponds to β and γ di ff erent from unity referred to as SOL βγ (cid:44) .To maintain consistency in constructing the MESSENGERorbit, we used corresponding planetary ephemerides that werebuilt with the same configurations of PPN parameters as usedfor SOL Re f and SOL βγ (cid:44) . In Fig. 9, we plotted the di ff er-ences between solve-for parameters obtained for SOL Re f andSOL βγ (cid:44) . The error bars shown in the same figure representthe 1 σ uncertainties in the estimation of solve-for parameterscorresponding to SOL Re f . From this figure, one can notice thedi ff erences in the parameters are always below the 1 σ uncer-tainties. The estimated solve-for parameters for SOL βγ (cid:44) areanalogous to SOL Re f , and there is no significant change in theMESSENGER orbit due to the change in PPN parameters. Incontrast, as shown in Fig. 7, this configuration of PPN param-eters ( β -1 = γ -1 = × − ) in the construction of planetaryephemerides led to ∼
65% of change in the postfit residuals,which shows that, the planetary ephemerides are more sensitiveto GR e ff ects. This can be explained from the fitting intervals of the data set. Usually planetary ephemerides are fitted overlong intervals of times (see Table 4) to exhibit long-term ef-fects, while a spacecraft orbit is usually constructed over muchshorter intervals (usually one day to a few days) of data arcs toaccount for the model’s imperfections. The short fitting intervalof the spacecraft orbit would absorb such e ff ects.Moreover, it is worth noticing that, unlike state vectors andscale factor F S (see Panels (a)-(g) of Fig. 9), the range biasdi ff erences between SOL Re f and SOL βγ (cid:44) solutions (see Panel(h)) shows systematic behavior. This trend in the range bias canbe explained from the contribution of the relativistic deflectionof light by the Sun (a function of PPN parameter γ , Shapiro(1964)) in the light time computations. Explicitly this e ff ectwas not absorbed during the computation of range bias and itbecomes important to examine this e ff ect when constructingplanetary ephemerides.We, therefore, reconstruct the planetary ephemerides usingthe range bias obtained from SOL βγ (cid:44) and the PPN parameters β -1 = γ -1 = × − . The newly estimated postfit range biasis then compared with the range bias (prefit) corresponding toSOL βγ (cid:44) . This investigation shows that, the postfit residuals aremodified by ∼
6% for MESSENGER and ∼
1% for Mariner 10with respect to prefit residuals. This modification in the residu-als is negligible compared to a ∼
65% of change with respect toreference residuals obtained from INPOP13a. As a result, thesupplementary contributions in the range bias due to the rela-tivistic deflection between DSN station and MESSENGER didnot bring any significant change in the planetary ephemeridesconstruction.
Fig. 9: Di ff erences between solve-for parameters obtained from the solutions SOL Re f and SOL βγ (cid:44) (see text): Panels (a)-(f)represent the changes in the initial state vectors, Panel (g) represents the changes in the scale factors estimated for solar radiationpressure, Panel (h) represents the changes in the estimated range bias. The error bars represent the 1 σ uncertainties in theestimation of solve-for parameters obtained with the reference solution SOL Re f .
5. Conclusions
We analyzed one and half years of radioscience data ofthe MESSENGER spacecraft using orbit determination soft-ware GINS. An accurate orbit of MESSENGER was thenconstructed with the typical range of Doppler, and two-wayrange residuals of about 4.8 ± ∼ ± / s),and 1.9 ± ± β , γ ), small variations of these two parameters nearunity were imposed in the construction of alternative planetaryephemerides fitted over the whole data sets. The percentage dif-ference between these ephemerides to INPOP13a are then usedto defined the interval of PPN parameters β and γ . As expected, our estimations of PPN parameters are morestringent than previous results. We considered the 5%, 10 %and 25% of changes in the postfit residuals. That the PPN in-tervals correspond to these changes is compatible with GR withan uncertainty at least ten times smaller than our previous re-sults with INPOP10a. Moreover, one of the best estimation ofparameter γ has so far been estimated from the Cassini obser-vations by Bertotti et al. (2003), which is compatible with our25% estimation.To further the accuracy of the PPN parameters improve, andto confirm the results given in Table 5, one needs to analysis theradioscience data of the future Bepi-Colombo mission.
6. Acknowledgments
We are very thankful to the CNES and Region Franche-Comte,who gave us financial support. Part of this work useds GINSsoftware, so we would like to acknowledge the CNES for pro-viding access to this software. A. K. Verma is thankful to P.Rosenbatt and S. Le Maistre for fruitful discussions. We arealso thankful to G.Esposito-Farese for his constructive remarksand comments. erma et al.: Planetary ephemeris construction and test of relativity with MESSENGER 15
References
Alimi, J.-M. & F¨uzfa, A. 2010, in American Institute of PhysicsConference Series, Vol. 1241, American Institute of PhysicsConference Series, ed. J.-M. Alimi & A. Fu¨ozfa, 690–699Anderson, J. D., Colombo, G., Espsitio, P. B., Lau, E. L., &Trager, G. B. 1987, Icarus, 71, 337Anderson, J. D., Jurgens, R. F., Lau, E. L., Slade, III, M. A., &Schubert, G. 1996, Icarus, 124, 690Anderson, J. D., Turyshev, S. G., Asmar, S. W., et al. 1997,Planet. Space Sci., 45, 21Bertotti, B., Iess, L., & Tortora, P. 2003, Nature, 425, 374Damour, T., Gibbons, G. W., & Gundlach, C. 1990, PhysicalReview Letters, 64, 123Damour, T. & Nordtvedt, K. 1993, Phys. Rev. D, 48, 3436Damour, T. & Polyakov, A. M. 1994, General Relativity andGravitation, 26, 1171Ercol, C. J., M., B. S., & Holtzman, G. A. 2012, in Thermaland Fluids Analysis Workshop, TFAWS2012-PT-03Fienga, A., Laskar, J., Kuchynka, P., et al. 2010, in IAUSymposium, Vol. 261, IAU Symposium, ed. S. A. Klioner,P. K. Seidelmann, & M. H. So ff el, 159–169Fienga, A., Laskar, J., Manche, H., et al. 2011a, CelestialMechanics and Dynamical AstronomyFienga, A., Laskar, J., Morley, T., et al. 2009, Astronomy &Astrophysics, 507, 1675Fienga, A., Manche, H., a. L. J., & Gastineau, M. 2008,Astronomy & Astrophysics, 477, 315Fienga, A., Manche, H., Laskar, J., Gastineau, M., & Verma,A. 2013, ArXiv e-printsFlanigan, S. H., OShaughnessy, D. J., Wilson, M. N., &Hill, T. A. 2013, in 23rd Space Flight Mechanics Meeting,American Astronomical Society, 13–382Folkner, W. M. 2010, JPL, intero ffi ce memorandum, IOM343R-10-001Konopliv, A. S., Asmar, S. W., Folkner, W. M., et al. 2011,icarus, 211, 401Lambert, S. B. & Le Poncin-Lafitte, C. 2011, A&A, 529, A70Manche, H., Fienga, A., Laskar, J., et al. 2010, in Journ´eesSyst`emes de R´ef´erence Spatio-temporels 2010, JourneesSystemes de referencesMargot, J.-L. 2009, Celestial Mechanics and DynamicalAstronomy, 105, 329Moyer, T. D. 2003, Formulation for Observed and ComputedValues of Deep Space Network Data Types for Navigation,Vol. 2 (John Wiley & Sons)M¨uller, J., So ff el, M., & Klioner, S. A. 2008, Journal ofGeodesy, 82, 133Pitjeva, E. V. 2009, in IAU Symposium, AmericanAstronomical Society, Vol. 261, 603Pitjeva, E. V. & Pitjev, N. P. 2013, MNRAS, 432, 3431Shapiro, I. I. 1964, Physical Review Letters, 13, 789Smith, D. E., Zuber, M. T., Phillips, R. J., et al. 2012, Science,336, 214Smith, D. E., Zuber, M. T., Phillips, R. J., et al. 2010, icarus,209, 88Solomon, S. C., McNutt, R. L., Gold, R. E., & Domingue, D. L.2007, Space Sci Rev,, 131, 3 Srinivasan, D. K., Perry, M. E., Fielhauer, K. B., Smith, D. E.,& Zuber, M. T. 2007, Space Science Reviews, 131, 557Stanbridge, D. R., Williams, K. E., Taylor, A. H., et al. 2011,in AAS //