The gravitational redshift monitored with RadioAstron from near Earth up to 350,000 km
The gravitational redshift monitored with RadioAstron from near Earth up to 350,000 km
Nunes, N. V.;Bartel, N.;Bietenholz, M. F.;Zakhvatkin, M. V.;Litvinov, D. A.;Rudenko, V. N.;Gurvits, L. I.;Granato, G.;Dirkx, D.
We report on our eorts to test the Einstein Equivalence Principle by mea- suring the gravitational redshift with the VLBI spacecraft RadioAstron, in an eccentric orbit around Earth with geocentric distances as small as 7,000 km and up to 350,000 km. The spacecraft and its ground stations are each equipped with stable hydrogen maser frequency standards, and measure- ments of the redshifted downlink carrier frequencies were obtained at both 8.4 and 15 GHz between 2012 and 2017. Over the course of the 9 d or- bit, the gravitational redshift between the spacecraft and the ground stations 10 10 varies between 6:8 10 and 0:6 10 . Since the clock oset between the masers is dicult to estimate independently of the gravitational redshift, only the variation of the gravitational redshift is considered for this analysis. We obtain a preliminary estimate of the fractional deviation of the gravi- Corresponding author Email address: firstname.lastname@example.org (N. V. Nunes) Preprint submitted to Advances in Space Research April 3, 2019 arXiv:1904.01060v1 [gr-qc] 1 Apr 2019 tational redshift from prediction of = 0:016 0:003 0:030 with stat syst the systematic uncertainty likely being dominated by unmodelled eects in- cluding the error in accounting for the non-relativistic Doppler shift. This result is consistent with zero within the uncertainties. For the rst time, the gravitational redshift has been probed over such large distances in the vicin- ity of Earth. About three orders of magnitude more accurate estimates may be possible with RadioAstron using existing data from dedicated interleaved observations combining uplink and downlink modes of operation. Keywords: Test of general relativity; Einstein Equivalence Principle; RadioAstron; space VLBI 1. Introduction The incompatibility of general relativity and quantum theory is a funda- mental problem in our understanding of the physical world. The Einstein Equivalence Principle (EEP) is a cornerstone of general relativity, and it leads to there being a gravitational redshift (Will, 1993). Speci cally, lo- cal position invariance requires that time
ows slower for an observer close to a massive body than for one farther away. Similarly, an electromagnetic wave is seen to be redshifted when its source is located close to a massive body in comparison to when the source is located far away from it. Accurate measurements of the gravitational redshift are thus a way of verifying EEP and, in turn, general relativity. Comparisons with predictions are therefore of prime importance. A milestone in measuring the gravitational redshift was reached with the Gravity Probe A (GP-A) mission in 1976. A Scout rocket lifted a spacecraft with a hydrogen maser frequency standard on board to an altitude of 10,000 km. The frequency from the onboard maser was compared with that from a maser on the ground. The resulting gravitational redshift measurement was consistent with prediction and had a fractional uncertainty of 1:4 10 (Vessot & Levine, 1979; Vessot, 1989). Recently, the two teams working on the GREAT project, using data from the Galileo 5 and 6 satellites erro- neously launched into slightly elliptical orbits with an eccentricity of 0.166 and semi-major axis of 27,980 km, have published results that reduce this uncertainty by more than a factor of 4 (Herrmann et al., 2018) and 5.6 (Delva et al., 2018) respectively. In the future, the ACES experiment, on board the International Space Station in a nearly circular orbit at an altitude of 400 2 6 km, is expected to reduce the uncertainty down to the 10 level (Meynadier et al., 2018). We conducted a similar experiment to GP-A in conjunction with the Russian-led international space Very Long Baseline Interferometry (space VLBI) mission RadioAstron (Kardashev et al., 2013). The RadioAstron spacecraft was launched on 2011 July 18 into a highly elliptical orbit around Earth with a perigee as small as 7,000 km and an apogee as large as 350,000 km from the geocenter. Since early on, the spacecraft has been used for astrophysical space VLBI observations of compact radio sources. The spacecraft has a hydrogen maser frequency standard (H-maser) on board that operated until mid-2017. Ground testing of this H-maser showed an Allan deviation of 2 10 for an averaging time of 1 hour. Prior to the failure of the onboard H-maser, the downlink carrier frequencies of 8.4 and 15 GHz, locked to the frequency of the maser, were recorded at ground sta- tions, also equipped with H-masers, during science observations. The ground stations for RadioAstron are the 26-m Pushchino tracking station in Russia (PU) and, from 2013 onwards, the 43-m Green Bank tracking station at the West Virginia site of the National Radio Astronomy Observatory in the USA (GB). We refer to this space-to-ground mode of operation as the 1-way mode. During the 9 d elliptical orbits, the spacecraft travels through the gravita- tional potential of Earth, which according to the EEP, should cause a varying gravitational redshift on the downlink signals with a relative frequency shift 10 10 predicted to vary between 6:8 10 and 0:6 10 . In order to measure the gravitational redshift, various eects need to be subtracted from the ob- served frequencies, the largest being the non-relativistic Doppler shift. This must be computed using the orbit of the spacecraft. In addition to the 1-way mode, the spacecraft also operates in a 2-way mode where the downlink signals are phase coherently locked to the uplink carrier frequency from the ground station. Figure 1 gives a block diagram for the two modes. In the 2-way mode, the Doppler shift is doubled while the gravitational redshift is cancelled. A combination of both the 1-way and 2-way modes that cancels the non-relativistic Doppler shift but retains the gravitational redshift allows the latter to be measured with much higher ac- curacy. This combination is similar to the GP-A mode of operation. The 1-way mode and the combination of the 1-way and 2-way modes each have advantages and disadvantages. The 1-way mode is compatible with normal science observations and thus allowed almost continuous daily monitoring of the predicted varying gravitational redshift for 5 yr, albeit with relatively 3 D.A. Litvinov etD.A al. /. Litvino PhysicsvLeetttal. ers/ A Ph382 ysics (2018) Letters2192–2198 A 382 (2018) 2192–2198 2195 2195 (a) (b) Fig. 4. Operation Fig.m4odes . Operofation the RmadioAs odes of tron theradio RadioAs linktr s:on a)r“H-Maser” adio links: (1-w a) “H-Maser” ay); b) “Coher (1-went” ay); b) (2-w “Coher ay); c) ent” “Semi-Coher (2-way); c)ent” “Semi-Coher (2-way). ent” (2-way). rely on anyre fe lya ton ur ean s of y fe the atu signal res of spectrum the signal and spectrum may be and re ama lizeyd be refo alri zae dnumb fo err of a numb ﬁne efer fect ofs ﬁne contributing effects contributing to the right-hand to the right-hand side of side of Figure 1: Block diagram of RadioAstron's modes of operation (Litvinov et al., 2018). In with telescopes equipped either with 8.4 or 15 GHz receivers. with telescopes equipped either with 8.4 or 15 GHz receivers. Eq. (5): Eq. (5): the 1-way mode (a), the downlink carrier frequencies from the spacecraft are locked to The secondThe op tion second fo r op the tion Doppler for the com Doppler pensation com in pensation volves involves the onboard H-maser. Measuring these frequencies at the station, also equipped with an recording the rec 15 ordGHz ing the data 15 doGHz wnlink data signal downlink in the signal “Semi-Coher in the “Semi-Coher ent” ent” • second- and • second- third-or and der thir reld-or ativider stic re kinematic lativistic ef kinematic fects: com- effects: com- mode of sync mode hr onization of synchr of onization the on-boar of the d scientiﬁc on-board and scientiﬁc radio and radio H-maser, allows the gravitational redshift to be monitored. Input the ed 2-w from aput ythe mo e dorbital fr de om(b), the data orbital the (velocity data de (v teelocity rmination det eaccur rmination acy accuracy equipment eq uipment , which  is ,a whic kindh ofis half-w a kinda yof b ehalf-w tweena ythe be tw 1-w een ay the 1-way of δv ∼ 2mofm δ/sv  ∼ 2m is m su/ﬃs  cient); is suﬃcient); onboard H-maser is not used. A combination of these modes is used to mostly cancel the “H-maser” and “H-maser” 2-way and “Coher 2-w ent” ay “Coher modes.ent” In this modes. mode In the this 7. mode 2 GH zthe 7.2 GHz • gravitational • gr apo vitational tential dif po fer tential ence bdif etw fereen ence the be tw spacecr een the aft and spacecr aft and non-relativistic Doppler shift and the eect of the troposphere. uplink toneuplink , the 8.4 toneG,Hz the do 8.4 wnlink GHz to do nwnlink e and the ton ecar and rier the of the car rier of the the ground the sta tgr ioound n: com stput atioend: frcom omput thee dorbital from the data orbital using the data using the 15 GHz data 15 do GHz wnlink data ardo e wnlink phase-loc arek ephase-loc d to the kgr edound to the H-maser ground H-maser Earth gravitational Earth gr po avitational tential model potential  model (the position  (the er rposition or of error of signal, while signal, the while modulation the modulation frequency (72 frequency MHz) (72 of the MHz) data of the data ∼ 200 m pr∼ovide 200dm by pr ra ovide dio dra n byg inragd  io ranisg isu ngﬃ  cientis fo su r ﬃ dis- cient for dis- downlink is phase-locked to the on-board H-maser signal (Fig. 4c). downlink is phase-locked to the on-board H-maser signal (Fig. 4c). low accuracy. Switching between 1-way and 2-way modes, however, is not tances ! 40,000 tanceskm, ! 40,000 laser rakm, ngin laser g req ra uin regdin ogth reerqw uiis ree) d ;otherwise); This approac This h re appr lies oac on hthe re lbr ieoadband s on the br (∼oadband 1GHz) natur (∼1G e H ofz )the natur 15 e of the 15 • residual ionospheric frequency shift: computed from • residual ionospheric frequency shift: computed from compatible with normal science observations and requires dedicated inter- GHz signal GHz modulat signal ed modulat using queadd rusing ature qu phase-shif adraturet phase-shif keying (QPSK) t key ing (QPSK) 2-frequency2-fr measur equency ement measur s (8.4ement and 15 s (8.4 GHz), and ionospheric 15 GHz), ionospheric total total and the possibility and the of possibility turning it ofs turning spectrum it sint spectrum o a comb-lik into ea fo comb-lik rm e form leaved observations and so the varying gravitational redshift was monitored electron cont electr ent on (TEC) cont maps ent (TEC)  and maps mapping  and functions mapping  functions , , by transmitting by tr ansmitting a predeﬁne d a pr periodic edeﬁne data d periodic sequence data ( Fseq ig. uence 5). As we (Fig. 5). As we for relatively short periods, but with expected higher onsit accuracy e GNSSonsit .recHere eei vGNSS er measur w reecere- iement ver measur s; ements; have shownha inv e sho , dif wn fer in ent  subt , difones ferent of subt the ones resu lof tin the g spectrum resultin gact spectrum act • frequency• shif freqtuency due to shif the t due tidal to gr the av itational tidal gr aﬁeld vitational of the ﬁeld of the like separatlik e link e separ s of athe te link GP-sA of sc the heme GP and -A sc can heme be and arrang can ed be in ar sof rang t- ed in soft- port on our analysis of 1-way mode measurements recorded at the ground Sun and Moon Sun  and :Moon comput  ed: fr com omput thee dplane fromtar the y and plane lunar tary and lunar ware postprwa ocessing re pos tpr intocessing o a combination into a combination similar to that similar of Eq. to that (5), of Eq. (5), stations and give preliminary results. ephemerides ephemerides (JPL DE430); (JPL DE430); which is free whic from h is the free 1s frt-or omder the Doppler 1st-order and Doppler tropospheric and tr opospheric effects effects (the ionospheric term persists). • phase cent • er phase motion cent of er the mo tion on-boar of the d and on-boar tracking d and st atr tiac onking an- station an- (the ionospheric term persists). Despite some Despit advantag e some es adv of the antag second es of the option second from op alg tion orithmic from algorithmic ten nas: com teput nnaesd: fr com omput the ed orbital from the and orbital housek and eeping housek dataeeping data 2. Modelling the 1-way data and operational and oper point ational s of vie point w, we s of gi vie vew pr , ef we er ence give pr toef the erence inter to - the inter  - ; ; leaved measur leavement ed measur s appr ement oach sas appr it pr oac ovides h as it fo rpr ao vides larger fonumb r a larg er er number • temperature dependence of the on-board H-maser: computed • temperature dependence of the on-board H-maser: computed of participating of participating GRTs due to GRT the s due larg to er the ground larg er fo ogr tpound rint of fo the otp rint of the from the H-maser from the sensiti H-maser vity de sensiti termine vityd de during termine grdound during test gr s ound tests The fractional frequency shift, f =f , measured at a station is given by obs on-board ant on-boar enna at d ant 8.4 enna GHz at and 8.4 wider GHz av and ail awider bility av of a8.4 ilab GHz ility of re -8.4 GHz re- and housekand eeping housek data;eeping data; the equation: ceivers at GRT ceivs.ers at GRTs. • magnetic • ﬁeld magne dependence tic ﬁeld dependence of the on-boar of the d H-maser on-board frH-maser e- fre- quency: com quput enceyd: fr com omput the ed H-maser from the sensiti H-maser vity de sensiti termine vity dde termined 3. Data processing 3. Data pr and ocessing ﬁne ef fand ects ﬁne effects during ground during test gr s, ound the magne tests, tic the ﬁeld magne model tic ﬁeld  model and the  and the orbital data;orbital data; The primaryThe data primar for the y data experiment for the e ar xperiment e the spacecr are the aft signals spacecr aft signals • ground station motion due to solid Earth tides: computed from • ground station motion due to solid Earth tides: computed from at 8.4 and/or at 8.4 15 GHz and/or re c15 eiv eGHz d and re cre eicvoerdd eand d at re ac ogr rdound ed at st aa tgr ioound n. station. Earth models Earth  models . . The majority of radio astronomy and geodetic radio telescopes are The majority of radio astronomy and geodetic radio telescopes are equipped with equippe H-maser d with st aH-maser ndards and sta n8.4 dard GHz s and re c8.4 eiv eGHz rs, enabling receiver s, enabling After the grAf avteitational r the gr afrveq itational uency shif freqt uency has be shif en tmeasur has beeedn ex measur - ed ex- them to take them part to in take the part ex pin er im the en tex . Re pecroim rdein ng t. of Re cthe ord inspace- g of the space- perimentallperimentall y in a series y in of aobserv seriesations of observ at vaations rious dis at tva ances riou son dis at ances on a craft signal crisaf perf t signal orme isd perf in the orme grdound in the H-maser ground timescale H-maser using timescale using single orbit, we ﬁt it against the gravitational potential difference single orbit, we ﬁt it against the gravitational potential difference standard VLBI sta nbac dard k-end VLBI ins bac trumentation. k-end instrumentation. Initial data Initial processing data pr is ocessing is according toaccor Eq. ding (2), thus to Eq. obt(2) aining , thus a obt single aining measur a single ement measur of ε.ement The of ε. The based on the algorithms developed originally for PRIDE (Planetary based on the algorithms developed originally for PRIDE (Planetary accuracy ofaccur the re acy su lof t of the 2–3 re sye ulta rof s of 2–3 planne yearsd of data planne accumulation d data accumulation Radio InterfRa erdome io Int tryerf and erome Doppler try and Experiment) Doppler Experiment)  for re c over-for recover- depends on the number of experiments performed and their pa- depends on the number of experiments performed and their pa- ing the phase ing of the the phase rece iof ve dthe signal. rece ivDe edtails signal. of the De tails algorithm of the and algorithm and rameters. Base ramde te on rs . the Base experiment d on the e er xperiment ror budg eer t r or budg ande ttaki  ng and taking software modiﬁcations software modiﬁcations required to re pr quocess ired to int er prlocess eaved int data erl ea will ved data will be given in an upcoming publication . Here we brieﬂint y ode- accountint the o account observ ations the observ performe ations d so perf faorme r andd those so far planne and those d planned be given in an upcoming publication . Here we brieﬂy de- −5 −5 scribe the appr scriboac e the hes appr for oac corh recting es for the cor recting recove re the d signal recove phases red signal phases we exp ect the we accur expecatcy the of accur the te acy st to of re the ac h te st δε to∼ re 10ach. δε ∼ 10 . 2 2 2 f D v v (v n) (v n) (v n) obs s e s e s = + 2 2 f c 2c c (1) f f f f v grav clock trop ne + + + + + O( ) f f f f c as described in Biriukov et al. (2014), see also Sazhin et al. (2010). The rst term is the non-relativistic Doppler shift, of order 10 , where D is the range rate or radial velocity between the spacecraft and ground station and c is the speed of light. The second and third terms are the relativistic contributions to the Doppler shift of order 10 , where v and v , respectively, are the e s velocity vectors of the station and spacecraft in the J2000 reference frame and n is the unit vector in the direction opposite to that of signal propagation. The fourth term is the observed gravitational redshift between the ground station and spacecraft and is of order 10 . The fth term, f =f , of clock order 10 , is due to the clock oset between the H-masers which drift relative to each other over time, and includes both a constant clock oset, y , and to rst order a linear clock drift, y , on the order of 10 /d with 0 1 f =f = y + y t. The sixth term, f =f , of order up to 10 , is clock 0 1 trop due to the troposphere. Eects of order 10 and smaller arising from the ionosphere, antenna phase center motion eects and the gravitational eects 3 3 of the Moon and Sun are grouped together in f =f . Finally, O(v =c ), ne of order 10 , consists of third order or higher terms that need not be considered given the accuracy achievable with our analysis. In Figure 2, we plot the total frequency shift and predicted gravitational redshift for the month of January in 2014 together with the geocentric dis- tance, jr j, of the spacecraft. The non-relativistic Doppler shift, which dom- inates the total frequency shift, varies due to the motion of the spacecraft and Earth's rotation. The largest fractional frequency shifts are more than 5 4 10 . The gravitational redshift is about 10 times smaller. Due to the large eccentricity of the orbit, the spacecraft spends much more time near apogee, where the gravitational redshift is near its maximum value of 6:8 10 . Near perigee, the gravitational redshift changes rapidly, reaching a minimum that depends on the orbit, which is constantly evolving, but can be as small as 0:6 10 . 5 5 -4 Total Frequency Shift Gravitational Redshift Spacecraft Distance -5 -6 -7 -8 -9 -10 -11 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 Time [d] Figure 2: A plot of the predicted total fractional frequency shift between the spacecraft and the GB ground station for a number of 9 d orbits, dominated by the non-relativistic Doppler shift, D=c, along with the much smaller gravitational redshift, f =f , with grav absolute values plotted against the logarithmic scale on the left side. In addition, the geocentric distance of the spacecraft, jr j, is plotted with the scale on the right side. Time is given in days since 2014 January 1. 3. Feasibility of determining the violation parameter A violation of the predicted gravitational redshift can be parametrized by introducing a violation parameter, , de ned as: z = (1 + )z (2) obs where z U=c with U being the dierence in Earth's gravitational potential between the ground station and the spacecraft. The parameter is what we seek to measure with = 0 corresponding to EEP being valid. To look for a possible violation using measured fractional frequency shifts, the state parameters for the spacecraft must be known with sucient ac- curacy to isolate the gravitational redshift from other eects, particularly the much larger non-relativistic Doppler eect. Orbit determination for the RadioAstron mission is done using a variety of sources of data, among them the same frequency observations made at the PU and GB ground stations Fractional Frequency Shift Geocentric Distance [km] used in this analysis (Zakhvatkin et al., 2013, 2018). The position and ve- locity of the spacecraft are determined using a least-squares minimization in which no violation is assumed. We conducted a covariance analysis, done in- dependently by two groups within our collaboration, to investigate whether the determination of would be biased by this assumption. No signi cant correlation between and the state parameters was found when they are esti- mated concurrently. However, as was expected, a signi cant correlation was found between and the constant clock oset, y . Since the constant clock oset is dicult to measure independently of the gravitational redshift, it is necessary to modify the approach to determining by instead looking at the variation of the gravitational redshift over the course of an orbit. We de ne a new quantity, the observed biased gravitational redshift as: f f f obs doppler trop z = obs f f f (3) f f grav clock = + f f obtained by subtracting the modelled Doppler eects ( rst three terms in Equation 1) and the modelled tropospheric eects from f =f , leaving obs the gravitational redshift biased by the clock oset. The tropospheric eect is computed using a tropospheric delay model based on seasonal averages (Collins, 1999). Contributions from eects on the order of 10 or smaller including the ionosphere and the tidal eects of the Moon ( 10 ) and Sun ( 10 ) are too small compared to the accuracy achievable with our present analysis and are ignored. The observed biased gravitational redshift can thus be parametrized as: z = (1 + )z + y + y t (4) 0 1 obs By taking the dierence between pairs of observations, the constant in the clock oset is removed: z = (1 + )z + y t (5) obs However, the expected uncertainty in the estimate of determined in this way increases by more than an order of magnitude as the variation of the gravi- tational redshift, z, is at least an order of magnitude smaller, on average, than the value of the gravitational redshift, z. 7 -11 8.4GHz 15GHz 4.0 2.0 0.0 -2.0 -4.0 8.4GHz 15GHz 0 2000 4000 6000 8000 500 1000 1500 2000 2500 3000 Time [s] No. Measurements Figure 3: A sample of the 100,000 residuals from tting the frequency measurements at 8.4 and 15 GHz with a polynomial for a single session of 1 h at Pushchino. The right panel shows the distributions of all the residuals at each downlink carrier frequency with almost complete overlap and with the dashed lines being Gaussian ts. 4. Observations and raw data About 3,900 sessions of downlink carrier frequency measurements were performed between 2012 and 2017, about 2,700 at PU and the remaining 1,200 at GB. Each session is about 1 h long and most have data for both the downlink carrier frequencies of 8.4 and 15 GHz. During each session, frequencies were recorded every 40 ms, resulting in over 100,000 frequency measurements at each of the two downlink carrier frequencies. A total of 850 million frequency measurements were obtained. At 40 ms intervals, frequency measurements are aected by Gaussian noise with a fractional frequency shift on the order of 10 and must be smoothed in time. The frequency mea- surements from each session are t with a polynomial that is used to arrive at a single frequency measurement at the mid-point of the session. Figure 3 displays the residuals from such a t for one session. Sessions after June 2017 are excluded as at that time the onboard H-maser began to show signs of failure prior to running out of hydrogen later in the summer. GB ses- Residual Frequency Shift sions in 2013 when the station was temporarily on a Rubidium standard are also excluded. In addition, 12% of the remaining sessions where residuals aren't suciently Gaussian distributed are also excluded. These additional excluded sessions are spread somewhat evenly over the full 5 years for both stations. 5. Data analysis Our goal is to estimate . Starting with the interpolated frequency mea- surement from each session, we obtain z by subtracting the Doppler eects obs computed using NOVAS for the position and velocity vectors of the stations and the determined orbital state of the spacecraft. The geodetic coordinates of GB are known from VLBI observations to an accuracy more than sucient for our purposes. However, the coordinates for PU are only known within a few meters which aects how accurately the non-relativistic Doppler shift can be computed and which, in turn, could have an impact on the estimate of . In Figure 4, panels (a) and (b) show observed biased gravitational red- shift measurements at PU and GB from 2012 to 2017, which are mostly due to the gravitational redshift including a possible violation due to 6= 0 and the clock oset between the onboard and station H-masers. Panels (c) and (d) zoom in on 60 d over which the variation of the gravitational redshift is clearly visible and can be compared to the predicted value for z. The eect of the clock oset drifting over time is shown in Figure 5. Starting with N sessions, we formed pairs of sessions at times t and t , i j where i < j N and t = t t , over a maximum time interval, t , j i max where 0 < t t . While a maximum of N 1 independent pairs are max possible by pairing each of the N sessions with another session, the restriction on t results in fewer than N 1 pairs. The choice of which speci c ses- max sions to pair within the time interval is made to maximize the magnitude of the gravitational redshift dierence, jzj, and thus maximize the sensitivity to a possible violation. By dierencing the observed biased gravitational red- shift to obtain z , the constant clock oset, y , is cancelled. Fitting these obs dierences to a model based on Equation (5) allows the violation parameter, , to be determined along with y . However, if the mean of the ratio of t NOVAS is a software library for astrometry-related numerical computations provided by the United States Naval Observatory. http://aa.usno.navy.mil/software/novas/ novas_info.php 9 -10 -10 10 10 7.5 7.5 8.4GHz 8.4GHz PU GB 15GHz 15GHz 7.0 7.0 6.5 6.5 6.0 6.0 5.5 5.5 0 500 1000 1500 2000 0 500 1000 1500 2000 Time [d] Time [d] (a) (b) -10 -10 10 10 7.4 7.4 8.4GHz 8.4GHz PU GB 15GHz 15GHz Predicted Predicted 7.2 7.2 7.0 7.0 6.8 6.8 6.6 6.6 6.4 6.4 6.2 6.2 610 620 630 640 650 660 670 610 620 630 640 650 660 670 Time [d] Time [d] (c) (d) Figure 4: Panels (a) and (b) show the observed biased gravitational redshift, z , from obs each session at PU and GB between 2012 and 2017. For each session, one point is plotted for each of the two downlink carrier frequencies. Panels (c) and (d) zoom in on 60 d to show the variation in the gravitational redshift in detail. The predicted gravitational redshift, z, is shown in green for comparison. The oset between the two is mostly due to the clock oset between the onboard and station H-masers. Time is given in days since 2012 January 1. The periods where there are fewer data points correspond to the summer months when the number of experiments with RadioAstron dropped o due to constraints on spacecraft attitude limiting the visibility of radio sources of interest. During these months, most sessions were closer to perigee, resulting in smaller gravitational redshifts. Fractional Frequency Shift Fractional Frequency Shift Fractional Frequency Shift Fractional Frequency Shift -11 -11 10 10 5.0 5.0 8.4GHz 8.4GHz PU GB 15GHz 15GHz 4.0 4.0 3.0 3.0 2.0 2.0 1.0 1.0 0.0 0.0 -1.0 -1.0 0 500 1000 1500 2000 0 500 1000 1500 2000 Time [d] Time [d] (a) (b) Figure 5: Panel (a) and (b) show the observed biased gravitational redshift minus the predicted gravitational redshift from each session for PU and GB between 2012 and 2017. This dierence is mostly due to the clock oset between the onboard and ground station H- masers drifting over time, apart from any contribution due to a possible non-zero violation parameter. Time is given in days since 2012 January 1. and z over time, ht=zi, is suciently small, which can be arranged by using a t that is relatively short, say < 10 d, then the eect of the clock max drift (as shown in Figure 5) is small compared to the accuracy of achievable in our analysis and can be ignored. 6. Estimate of For estimating , we maximize the signal and minimize systematics and noise by considering only those session pairs with the largest gravitational redshift dierences. We choose a minimum magnitude of the predicted grav- itational redshift dierence of jzj = 1 10 over a time interval of less min than t = 4:5 d, about half an orbit, resulting in a weighted mean of max ht=zi = 4 10 d for PU and GB. This choice allows the drift parameter, y , to be ignored but it still produces a sucient number of pairs for our statistical analysis. The parameter is computed for each pair. Outliers are eliminated using a lter based on a 3 criterion resulting in fewer than 6% of pairs being excluded. In Table 1, we list the mean values for our estimates of , together with their standard errors, separately for PU and GB and for the two downlink signals. Individual estimates of , for each session pair, Fractional Frequency Shift Fractional Frequency Shift Table 1: Mean values for with uncertainties PU GB 8.4 GHz 15 GHz 8.4 GHz 15 GHz -0.018 -0.018 -0.006 -0.002 0.002 0.002 0.005 0.005 stat 0.020 0.015 0.040 0.040 syst; t 0.025 0.020 0.055 0.060 syst No. pairs 967 965 191 182 are plotted in Figure 6. The solutions for the mean of at the two down- link carrier frequencies and for the two stations are all within a combined 0.5 . However, the solutions show a trend with time that we t by a stat parabola to estimate systematic errors. The t is very similar for the two downlink carrier frequencies at a given station and somewhat similar between the stations. This trend is expected to be due to unmodeled eects, possibly related to the error in accounting for the non-relativistic Doppler shift. We tried excluding sessions at times when the uncertainty in the orbital state parameters is highest, for example, during the summer months when there are fewer sessions as well as near perigee, but found the resulting ts to be within 1 of the combined statistical uncertainty. In fact, no subset of the data has been found that eliminates the systematic trend. We also analysed the eect of not accounting for the clock drift between the H-masers and, as expected, found that it does not explain the systematic trend. We esti- mate the magnitude of the systematic eect using the maxima and minima of the parabolas in the time range of the observations given in the table as . Our nal estimate of the systematic errors, , enlarges this range syst; t syst by taking into account the 68% con dence bands of the t. Our average for , weighted by the number of pairs, with statistical and systematic errors is: = 0:016 0:003 0:030 stat syst Both linear and parabolic ts were tried giving systematic errors with 68% con dence bands included of 0.017 to 0.028, respectively. We lean away from a cubic t or higher as these aren't suggested by the scatter of the data and, instead, use a parabolic t yielding a more conservative systematic error of 0.030 after rounding up. 12 0.4 0.4 8.4GHz 8.4GHz PU GB 15GHz 15GHz 0.3 0.3 0.2 0.2 0.1 0.1 0.0 0.0 -0.1 -0.1 -0.2 -0.2 -0.3 -0.3 -0.4 -0.4 0 500 1000 1500 2000 0 500 1000 1500 2000 Time [d] Time [d] (a) (b) Figure 6: Estimates of for each downlink carrier frequency computed using z from obs each pair of 1 h sessions where jzj = 1 10 and t = 4:5 d. Each point is min max plotted at the mid-point of the session pair. The rms scatter of 0.07 at PU and 0.08 at GB, is 3 times smaller than what would be expected from the uncertainty due to errors in the orbital velocity of the spacecraft found to be 1:36 mm/s (Zakhvatkin et al., 2018) in any direction. The parabolic ts, almost indistinguishable at the two downlink carrier frequencies, highlight a trend which we consider to be a systematic error. Time is given in days since 2012 January 1. This estimate of is consistent with zero within the combined uncertainties. 7. Discussion We have presented an estimate of on the basis of 1-way frequency mea- surements at two stations and at two downlink carrier frequencies. Since the constant clock osets between the onboard H-maser and the H-masers at the stations is dicult to measure independently of the gravitational redshift, only dierenced frequency measurements are used to estimate . Given our criteria for pairing observations, the dierence in gravitational redshift is, on average, 30 times smaller than the absolute eect, and therefore the expected statistical noise
oor is about 30 times larger. The statistical un- certainty for is approximately at the expected level and cannot be reduced much further using the 1-way data obtained at the ground stations. The sys- tematic uncertainty, however, is tenfold larger. If the systematic trend in the t of over the ve years of observations were to be accounted for, the total error would be reduced signi cantly. This systematic eect is possibly due Violation Parameter Violation Parameter _ to the error in accounting for the non-relativistic Doppler shift, D=c, nec- essary when using 1-way data alone. Our dedicated interleaved observations of 1-way and 2-way data, however, oer the possibility of mostly eliminat- ing the non-relativistic Doppler shift and potentially allow three orders of magnitude more accurate estimates (Litvinov et al., 2018). 8. Conclusions In summary: For 5 yr a varying gravitational redshift was observed with the Ra- dioAstron spacecraft, which is in an eccentric orbit from near Earth out to a distance of 350,000 km. The observations were made with the ground stations at Green Bank, USA, and Pushchino, Russia, in the 1-way operating mode at the down- link carrier frequencies of 8.4 and 15 GHz. These are the rst measurements of the varying
ow of time over such large distances in the vicinity of Earth. We estimate the violation parameter = 0:016 0:003 0:030 stat syst This estimate of is consistent with zero within the combined sta- tistical and systematic uncertainties and therefore consistent with the predictions of the Einstein Equivalence Principle which is foundational to general relativity. While the statistical uncertainty is at the expected level, the systematic uncertainty is tenfold larger, possibly due to the error in accounting for the non-relativistic Doppler shift. Dedicated interleaved observations using the 1-way and 2-way modes mostly eliminate the non-relativistic Doppler shift and promise to re- duce the total uncertainty of by possibly three orders of magnitude. 9. Acknowledgements The RadioAstron project is led by the Astro Space Center of the Lebedev Physical Institute of the Russian Academy of Sciences and the Lavochkin 14 Scienti c and Production Association under a contract with the Russian Federal Space Agency, in collaboration with partner organizations in Russia and other countries. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. The work of DAL, MVZ and VNR is supported by the RSF grant 17-12-01488 (estimation of error covariance matrix for RadioAstron gravitational redshift experiment). Research at York University was supported by a grant from the NSERC of Canada. References Biriukov, A.V., Kauts, V.L., Kulagin, V.V. et al.. 2014. Gravitational redshift test with the space radio telescope \RadioAstron". Astronomy Reports 58, 783-795. Collins, J. 1999. Assessment and Development of a Tropospheric Delay Model for Aircraft Users of the Global Positioning System. Department of Geodesy and Geomatics Engineering Technical Report No. 203, Univer- sity of New Brunswick, Fredericton, N.B., Canada. Delva, P., Puchades, N., Schnemann, E. et al. 2018. Gravitational Redshift Test Using Eccentric Galileo Satellites. Phys. Rev. Let., 121, 231101. Herrmann, S., Finke, F., Llf, M. et al. 2018. Test of the Gravitational Redshift with Galileo Satellites in an Eccentric Orbit. Phys. Rev. Let., 121, 231102. Kardashev, N.S., Khartov, V.V., Abramov, V.V. et al. 2013. \RadioAstron" - A telescope with a size of 300 000 km: Main parameters and rst obser- vational results. Astronomy Reports, 57, 153-194. Litvinov, D.A., Rudenko, V.N., Alakoz, A.V. et al. 2018. Probing the grav- itational redshift with an Earth-orbiting satellite. Physics Letters A, 382, 2192-2198. Meynadier, F., Delva, P., le Poncin-La tte, C. et al. 2018. Atomic clock ensemble in space (ACES) data analysis. Class. Quantum Grav., 35, 035018 Sazhin, M.V., Vlasov, I.Y., Sazhina, O.S. and Turyshev, V.G. 2010. Ra- dioAstron: relativistic frequency change and time-scale shift. Astronomy Reports, 54, 11, 959-973. 15 Vessot, R.F.C., Levine, M.W. 1979. A test of the equivalence principle using a space-borne clock. General Relativity and Gravitation, 10, 181-204. Vessot, R.F.C. 1989. Clocks and Spaceborne Tests of Relativistic Gravitation. Adv. Space Res., 9, 921-928. Will, C.M. 1993. Theory and experiment in gravitational physics. Cambridge University Press, Cambridge. Zakhvatkin, M.V., Ponomarev, Y.N., Stepanyants V.A. et al. 2013. Naviga- tion Support for the RadioAstron Mission. Cosmic Research, 52, 342-352. Zakhvatkin, M.V., Andrianov, A.S., Kostenko, V.I. et al. 2018. RadioAstron orbit determination and evaluation of its results using correlation of Space- VLBI observations. arXiv preprint arXiv:1812.01623.
http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.pngGeneral Relativity and Quantum CosmologyarXiv (Cornell University)http://www.deepdyve.com/lp/arxiv-cornell-university/the-gravitational-redshift-monitored-with-radioastron-from-near-earth-8uOGhCv4LC
The gravitational redshift monitored with RadioAstron from near Earth up to 350,000 km