Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Mass sensitivity of gravimetric satellites

Mass sensitivity of gravimetric satellites Frequency-domain expressions are found for gradiometer and satellite-to-satellite tracking measurements of a point source on the surface of the Earth. The maximum signal-to-noise ratio as a function of noise in the measurement apparatus is computed, and from that the minimum detectable point mass is inferred. A point mass of magnitude M = 100 Gt gives a signal-to-noise ratio of 3 when a GOCE-like gradiometer passes directly over the mass. On the satellite-to-satellite tracking mission GRACE-FO M = 1:3 Gt for the microwave instrument and M = 0:5 Gt for 3 3 the laser ranging interferometer. The sensitivity of future GRACE-like missions with di erent orbital parameters and improved accelerometer sensitivity is explored, and the optimum spacecraft separation for detecting point-like sources is found. The future-mission benefit of improving the accelerometer sensitivity for measurement of non-gravitational disturbances is shown by the resulting reduction of M ; to as small as 7 Mt for 500 km orbital altitude and optimized satellite separation of 900 km. © 2020 COSPAR. Published by Elsevier Ltd All rights reserved. 1. Introduction to GRAIL data to solve for regional fields of the Moon. A global gravity map is the principal data product of satel- Though they di er slightly in assumptions and results, the lite gravity missions. Previously CHAMP (Reigber et al. spherical harmonic and mascon methods are constructed to an- (2003)), GRACE (Tapley et al. (2004)), and GOCE (Drinkwa- swer the same question: what is the gravity field that is most ter et al. (2006)) collected data to map the Earth’s gravity, and consistent with measurements? Here we address a di erent GRAIL (Konopliv et al. (2013)) measured the Moon’s gravity. question: what is the limit to measurement precision of a point- Currently GRACE Follow-On (GRACE-FO, Landerer et al. like mass on the surface? This is an artificial model, a single (2020)) is extending the GRACE data record, with increased mascon, that is not directly applicable to the geodetic agenda of ranging precision a orded by its laser ranging interferometer measuring the Earth’s gravity. We do not attempt to replicate (LRI, Abich et al. (2019)). the global gravity-field inversion achieved by the usual many- As pointed out by Watkins et al. (2015), the most com- mascon analysis. Rather, the motivation for this analysis is monly used method of analyzing satellite gravity data is based twofold: to provide a single-number figure of merit, namely on global gravity fields expressed in terms of spherical har- the minimum detectable isolated point mass perturbation, and monic basis functions. An alternative to spherical harmonics to find the optimal filter for such a detection. The minimum is the mass concentration, or mascon, model. Starting with detectable mass M is defined as the point mass that a gives a Wong et al. (1971), the mascon approach was applied to single- signal-to-noise ratio  = 3 in a single orbital pass directly over satellite lunar orbital measurements to infer the surface gravity the point mass. It is calculated by applying the Wiener optimal of the moon. Mascons can be modeled as many discrete sources filter to the problem of detecting a signal of known waveform, (Pollack (1973), Watkins et al. (2005)) that cover the globe, or against a background specified by instrument noise power spec- used to solve for regional fields. Han (2013) applied mascons tral density (Wainstein and Zubakov (1970)). A comparison of https://dx.doi.org/10.1016/j.jasr.xxxx.xx.xxx 0273-1177/© 2020 COSPAR. Published by Elsevier Ltd All rights reserved. arXiv:2006.08179v9 [physics.geo-ph] 12 Dec 2020 2 Robert Spero / Advances in Space Research xx (2020) xxx-xxx M for di erent orbital configurations and instrument sensitivi- ties guides the design of future missions. Additionally, we find the optimum satellite separation in GRACE-like missions for a specified instrument noise power spectral density. 2. Gradiometer Mass Sensitivity Consider a gradiometer flying directly over a point mass M at altitude h; Figure 1 left. At orbital altitude h = 330 km; the Fig. 2. Orbiting gradiometer vertical gradient response to a point mass. Upper- Fig. 1. Measurement of gravitational field from a point mass M at along-track left: g (t); z-direction gradient in time domain, Equation 1; Upper-right: distance x, altitude h; and velocity v: Left: Vertical gradient g : Right: Di er- g ( f ) = Fourier transform, Equation 2; Lower: M = Detectable mass with z 3 ential acceleration a between spacecraft 1 and 2 with average separation L. signal-to-noise = 3, Equation 5. along-track velocity v is the orbital velocity v = 7:7 km=s and In general,  depends on what filtering is applied to the in- the along-track distance x changes at approximately constant strument output. After Flanagan and Hughes (1998), with rate, x = v t: The acceleration at the spacecraft in the vertical, 2 2 3=2 optimum filtering the maximum signal-to-noise ratio per unit z direction is a = G Mz=(x + z ) ; and the gradient in the z 2 2 2 5=2 2 2 3=2 source mass  is direction is g = da =dz = G M[3z (x + z ) (x + z ) ]: z z G = Newton’s constant of gravitation. Substituting z ! h and Z x ! v t, = 4 W ( f )d f: (4) h i 2 5=2 2 3=2 g (t) =  M 3(1 + [ f t] ) (1 + [ f t] ) (1) z g h h Prime superscripts indicate quantities normalized by the source 3 3 0 0 2 where  M = G M=h = 1:85 10 mE with M = 1 Gt and g mass:  = =M and W ( f ) = W ( f )=M : It follows that the g z 12 2 f = v =h = 23 mHz; 1 mE = 1 10 s : At t = 0, when h o minimum detectable point mass with  = 3 is the gradiometer is directly above the source mass, the gradient reaches its maximum value g = 2 M: At x =  2h; g = 0: z g z M = q : (5) The Fourier transform of p(t) is defined by F [ p(t)] 4 W ( f )d f 1 z p( f ) = dt p(t) exp(2i f t): Applying the Fourier transform to g (t): The lower panel of Figure 2 shows M as a function of or- z 3 " ! !# bital altitude h for a gradiometer limited by white spectral 4 f 2 f 2 f 2 f noise S ( f ) = 1 mE= Hz (approximated value for GOCE g ( f ) =  M K + K (2) z g 1 0 f f f f h h h h from Touboul et al. (1999), Touboul et al. (1999a)). At h = 330 km; M = 100 Gt and the minimum observable gradient at where K is the modified Bessel function of the second kind, the peak time t = 0 is 2 M = 0:37 mE: order n. From Abramowitz and Stegun (1964), Section 9.7.2, K (z)  =(2z)e ; where  indicates approximately equal for large z: At frequency f  f ; the measurement response 3. Sensitivity of GRACE-like measurements is attenuated approximately exponentially with e-folding fre- quency f =(2) = 3:7 mHz: This corresponds to harmonic or- The measurement configuration and signal parameters for der N = f =(2 f ) = 20; where f = 0:183 mHz is the orbital the low-low satellite-to-satellite tracking (SST) of GRACE and h 1 1 frequency. GRACE-FO are shown in Figure 1, right. The primary signal is The signal-to-noise ratio  depends on the signal and the the along-track di erential position of the spacecraft, measured power spectral density of the gradiometer noise, S ( f ): Define by microwave ranging or laser interferometry. the frequency-dependent signal-to-noise-ratio density SNRD as 3.1. Single Spacecraft Acceleration For simplicity, the Flat-Earth approximation (Tapley (1997)) jg ( f )j W ( f ) = : (3) is used. This approximation neglects centrifugal acceleration, S ( f ) g Robert Spero / Advances in Space Research xx (2020) xxx-xxx 3 which introduces an error of less than 20% at Fourier frequen- cies greater than 2 mHz (Ghobadi-Far et al. (2018), Figure 1; Muller ¨ (2017), Figure 1.7). In this approximation, the acceler- ation on spacecraft 1 flying over point mass M at along-track 2 2 3=2 distance x is a = G M x=(h + x ) : Define the acceleration per unit source mass, a = a =M: Then Gtv a = (6) 2 2 2 3=2 (h + v t ) Converting to frequency space, 2 f Fig. 3. Range acceleration resulting from a square mass centered under the flight 4i f GK 0 0 along-track path. Orbital altitude h, average spacecraft separation L and source a ( f ) = F [a (t)] = : (7) 1 1 mass M are indicated in the title. The separate traces are for squares of side S indicated in the legend. Left: time domain, a (t); Right: Fourier Transform, a ( f ): 3.2. Range Acceleration Signal where The acceleration experienced by spacecraft 2 is the same as v = orbital velocity =7:6 km=s spacecraft 1 at distance L, but delayed by  = L=v : The re- h = orbital altitude =500 km sulting (along-track) range acceleration between the spacecraft L = spacecraft separation =200 km (Figure 3, left) is similar to what Han (2013) computed for the response of the GRAIL spacecraft to regional lunar gravity. The f = =15:2 mHz p h peak range acceleration a is f = =76 mHz L=2 GL a = M   M; (8) R  = =26 s 3=2 2 2 h + (L=2) o GL = =0:101 nm=s =Gt: 3=2 2 2 3=2 2 2 where  = GL(h + (L=2) ) : h + (L=2) Using the identity F (delay ) = exp(2i f), the range ac- These numerical values apply to GRACE-FO. The approxi- celeration in the frequency domain, a ( f ); is given by mately exponential attenuation with frequency of ja ( f )j has e-folding frequency 2:5 mHz, corresponding to harmonic order 0 0 2i f a ( f ) = a ( f )(1 e ) (9) N = 14: R 1 0 0 To explore the valid realm of the point-mass approximation, ja ( f )j = 2ja ( f ) sin( f)j (10) R 1 Figure 3 shows the range acceleration signal from a square- shaped planar mass of side length S, computed by numerical That is, in the frequency domain the range acceleration is integration. The S=1 km result is in agreement with the point- the single-satellite acceleration multiplied by 2j sin( f)j: For mass analytical calculation, which is valid at the 20% level for f  1=; ja ( f )j / L, which is the response for the spacecraft sources as large as S=300 km. Henceforth, we restrict our anal- pair acting as a gradiometer. The response departs from that ysis to the signal from a point source. of a gradiometer at large L; most conspicuously in the form of Equation 11 gives the measurement impulse response; that high-frequency nulls where the signal vanishes. The first null is, the range acceleration frequency response to a point mass is at f = 1= = 38 mHz for low-Earth orbit and L = 200 km, null input. This facilitates the direct comparison of signal and noise as recognized by Wol (1969). In degree-variance evaluations amplitudes as computed in the following section, and yields an of measurement sensitivity, the first null is expressed as a max- expression for the minimum detectable mass for GRACE-like imum in geoid height error at degree N = f = f = 216 for null 1 measurements of point source perturbations to surface gravity. L = 200 km; and N = 86 for L = 500 km; where f = orbital frequency = 0:176 mHz. 3.3. Noise and Mass Sensitivity From Equations 7 and 10, Consider the range measurement made by the laser ranging interferometer (LRI) on GRACE-FO. Assuming the measure- ! ! 8 f G 2 f 2 f ment resolution is limited by the thermal noise of the laser ref- ja ( f )j = K sin ; (11) erence cavity (Numata et al. (2004)), the displacement noise v f f h L o 4 Robert Spero / Advances in Space Research xx (2020) xxx-xxx root power spectral density (rpsd) x ˜ and strain rpsd x ˜ =L LRI LRI are given by x ˜ ( f )=L = x = f; (12) LRI c where x is a constant. For the LRI (Abich et al. (2019)), x = c c 1 10 : The rpsd of the LRI range acceleration noise is S ( f ) = (2 f )  x ˜ ( f ): (13) LRI LRI Take for the accelerometer measurement noise rpsd on a sin- gle satellite of GRACE and GRACE-FO (Touboul et al. (1999)) S ( f ) = a ˜ 1 + : (14) ACC 0 Fig. 4. Total range acceleration noise rpsd S for various assumptions of in- strument noise. The ranging noise for the “GRACE-FO MWI+ACC” is the white displacement noise of the microwave measurement on GRACE-FO, equal 11 2 Estaimates of a ˜ and f range from 3 10 m=s = Hz 7 0 k to 6 10 m= Hz. The other two noise curves assume the ranging noise of and 10 mHz, respectively (Hauk and Wiese (2020)) to the LRI, Equation 12, shown as a dotted line. Two levels of a ˜ =[m=s = Hz] 11 13 10 2 are assumed: 7 10 for the GRACE-FO curves, and 7 10 for a future 1 10 m=s = Hz and 5 mHz, respectively (Christophe et al. mission such as Mass Change Mission (MCM). The solid black line is the signal (2010), Conklin and Nguyen (2017)) . We take as a compromise spectrum a ( f ) from a 1 Gt point mass, for h = 500 km and L = 200 km; units 11 2 a ˜ = 7 10 m=s = Hz and f = 5 mHz: Assuming that the 0 k m/s. The values of M for the three respective configurations are (Section 3.3) acceleration measurements on the two satellites are uncorre- 1:3 Gt and 0:5 Gt and 7 Mt. lated, the total accelerometer noise is double: S = 2S : ACC ACC Improved accelerometers in future missions (Christophe et al. (2010), Conklin and Nguyen (2017)) may have a ˜ = 13 2 7 10 m=s = Hz: The total instrument noise power spectral density is S = S + S : (15) a ACC LRI Figure 4 shows S for di erent ranging instrument and ac- celerometer noise spectra. The MWI ranging noise is approxi- mated by white displacement noise, x ˜ = 6 10 m= Hz: MWI This estimate is based on comparing MWI range measurements to simultaneous LRI range measurements. To guide the eye to the frequencies that have the largest signal, ja ( f )j for a 1 Gt Fig. 5. Signal-to-noise ratio integrand of Equation 16, W ( f ) for the “GRACE- point source from Figure 3 is overlaid as the solid black line. FO LRI + ACC” noise of Figure 4, signal from orbital h = 500 km and source mass M = 1 Gt and three di erent values of spacecraft separation L: The in- The units of ja j are m/s. tegrated signal-to-noise ratios  from Equation 17 are indicated in the legend. As in Section 2, define the SNRD for range acceleration 2 2 ja ( f )j =M W ( f ) = : (16) S ( f ) The corresponding detectable peak accelerations,  M ; are R 3 2 2 0:13 nm=s , 0:047 nm=s . 0 2 W ( f ) = W M is shown in Figure 5 for several values of L: Another assessment of mass sensitivity for SST laser rang- The oscillations with nulls at multiples of 1= = v =L degrade ing is inferred from Colombo and Chao (1992), who proposed for L beyond an optimum spacecraft separation. a laser ranging mission that, with (h; L) = (600; 500) km was The optimal signal-to-noise ratio per unit mass is found by simulation to have sensitivity to weekly changes of 1 1 mm water height over a square region 400 km across, or mass = 4 W ( f )d f; (17) sensitivity of 160 Mt. In comparison, we find for the LRI on GRACE-FO at the same (h; L), M = 400 Mt: The two mea- surements have di erent assumed instrument sensitivity and av- and the source mass that gives  = 3 is (cf. Equation 5) eraging times (week-to-week vs. single-pass). Figure 6 shows the mass sensitivity M as a function of M = q : (18) 0 h and L for the LRI ranging instrument with two di er- 4 W ( f )d f ent levels of accelerometer sensitivity: a ˜ = 7 10 and 13 2 From Equations 11 through 16 and Equation 18, the 7 10 m=s = Hz. The lower row of Figure 6 shows the GRACE-FO parameters with the microwave ranging instru- optimum L for a given h and the resulting M : The optimum ment (MWI) and LRI give respectively M = 1:3 Gt; 0:5 Gt: L for the LRI on GRACE-FO, operating at h = 500 km, is 3 Robert Spero / Advances in Space Research xx (2020) xxx-xxx 5 L = 900 km; which would give M = 200 Mt: That reflects a Residuals are relative to a reference field. The admittance Z( f ) potential factor of 2.5 improvement over M = 500 Mt for the is defined as the ratio of power spectra, nominal satellite separation of L = 200 km: A future mission S LOS ( f ) ; ¨ g with the improved a ˜ ; h = 500 km; and optimal satellite separa- Z( f ) = ; (22) S ( f ) tion L = 900 km has M = 7 Mt: ; ¨  ¨ where S ( f ) is the power spectrum of the MWI range accel- ; ¨  ¨ eration measurement and S LOS ( f ) is the cross-power spec- ; ¨ g trum between the range acceleration and the LOS gravity di er- ence. Keeping the shorthand notation p( f ) = F [ p(t)]; Z( f ) is a filter that transforms residual range acceleration  ¨( f ) = a ( f ) LOS LOS to g ( f ), an estimate of g ( f ) : F 12 LOS g ( f ) = Z( f ) ¨( f ): (23) Z( f ); normalized to have a maximum value of 1, is shown as the dashed trace in Figure 7. Z( f ) is the optimal filter to apply to MWI range acceleration, based on the measurement data that includes signal from the gravity field. It applies to extracting the best SNR from a residual regional or global field and does not explicitly depend on instrument noise spectra. In contrast, G( f ) is fine-tuned to the problem of detecting the specific waveform of a point mass, in the presence of known measurement noise. Since a point mass generates a field with Fig. 6. Mass sensitivity of the LRI measurement on GRACE-FO left, and of the highest possible frequency content, the G( f ) passband starts a future GRACE-like mission right. Upper row shows isomass M contours, higher in frequency than Z( f ): The LRI G( f ) passband is higher in Mt, from Equation 18. Equation 12 specifies the ranging noise, and ac- 11 13 celerometer noise is given by Equation 14 with a ˜ = (7 10 ; 7 10 ) than for the MWI because the LRI measurement has reduced m=s = Hz, with fixed f = 5 mHz: Lower row shows the optimum L as a noise at high frequency. function of h; (blue, left axis) and the resulting sensitivity M (red, right axis). A practical use for the G( f ) filter is searching for un- known point-like features, such as underground water stor- age of 100 km spatial extent. The filter would be applied to 3.4. Optimal filter range acceleration measurements after subtracting the e ect of the known field, including time-varying gravity, and non- The filter that gives maximum signal-to-noise ratio is (Wain- gravitational accelerations. stein and Zubakov (1970), Chapter 3) a ( f ) G( f ) = ; (19) S ( f ) with denoting complex conjugation. The filter’s input is the measured range acceleration. G( f ) is an example of a filter for extracting a signal of known waveform, in this case the range acceleration resulting from flying over a point mass. Dropping the multiplicative constants, the filter magnitude is 2 f 2 f f K sin f f h L jG( f )j = : (20) S ( f ) Fig. 7. Transfer functions for range acceleration data. The blue trace that peaks Normalized jG( f )j for the MWI and LRI on GRACE-FO are at 2:6 mHz is jG( f )j for the MWI on GRACE-FO, and the red trace that peaks at 5:7 mHz is for the LRI on GRACE-FO. Z( f ) is the admittance filter from shown in Figure 7. Ghobadi-Far et al. (2018) Figure 2(b), and applies to the MWI on GRACE-FO. Ghobadi-Far et al. (2018) analyzed the GRACE-FO MWI All curves are normalized to give maximum value of 1. signal in terms of the line-of-sight gravity di erence, and ap- plied the same analysis method to the GRACE-FO LRI sig- nal in Ghobadi-Far et al. (2020). In their analysis of MWI 4. Conclusion data, Ghobadi-Far et al. (2018) defined the gravimetric quan- LOS tity g ; or line-of-sight (LOS) gravity di erence, which dif- We derived the optimum sensitivity of orbiting gravimetric fers from the range acceleration residual  ¨ by  ; the residual satellites to a point source, that is a single mascon. The signal- centrifugal acceleration: to-noise ratio is found as a function of instrument noise and LOS g =  ¨ +  : (21) orbital parameters. The signal is converted to frequency space 12 6 Robert Spero / Advances in Space Research xx (2020) xxx-xxx by the Fourier transform, and the signal-to-noise ratio is derived from optimal filtering a signal of known waveform. This anal- ysis di ers from the conventional approach of spherical har- monic expansion to characterize the field from an arbitrary mass distribution. Such an expansion requires a very large harmonic order to accurately approximate the field from a point source, as shown in Appendix A. The frequency response of an orbiting gradiometer to a point mass directly under the flight track is approximated by Equation 2 that depends only on the orbital altitude and the magnitude of the point mass. Likewise, for an SST-based measurement of the gravitational field, the range acceleration is approximated by Equation 11 that includes dependency on the average satel- lite separation. Applying Wiener optimal filter theory, these Fig. A.8. Coordinate system for the spherical harmonic expansion of geopoten- responses and the noise spectra of the ranging measurement tial, Kaula (2013). The satellite constellation position, defined as the center of and of accelerometer-based measurement of non-gravitational mass for a gradiometer or (illustrated) the center of the line of sight between forces give ; the maximum achievable signal-to-noise ratio. two SST satellites relative to the center of the Earth, is ~ r(r; ; ), where r is the The resolvable mass M is defined as the magnitude of the point distance from the center of the Earth,  is the co-latitude, and  is the longitude. mass that gives  = 3: M is the ultimate mass sensitivity, and realistic non-point mass distributions that are not directly un- der the flight track will give larger M in practice. Nonetheless, Earth mass), and P is the fully-normalized associated Leg- nm M provides a figure of merit for comparing future missions endre function. The field is entirely specified by the Stokes with di erent orbits and instrument sensitivities to guide the coecients (C ; S ): nm nm design of such missions. For SST measurements M has a min- 0 0 0 0 For a known mass distribution d M = (r ;  ;  )dV with imum value at a calculable satellite separation L; giving the op- primes designating the source mass coordinates, (C ; S ) are nm nm timum separation for discovering point-like (meaning less than evaluated as the volume integral Bettadpur (2018) approximately 300 km) features such as subsurface water stor- " # Z ! " # age. Equation 20 specifies the optimal filter for such a search. 0 0 1 r C cos m nm = d M P ( ) ;(A.2) nm As a caveat, the M metric and its L optimization does not apply 0 sin m S (2n + 1)M 0 a nm e to large-scale gravimetry, such as required by oceanography. 0 0 0 where  = cos  : For a point mass at r = a " # " # Acknowledgments C cos m nm 0 = P ( ) : (A.3) nm 0 sin m S (2n + 1)M nm The author thanks Kirk McKenzie, Gabriel Ramirez, Pep e Sanjuan and David Wiese for useful discussions, and Christo- 0 0 A single point mass can be placed at the north pole, ( ;  ) = pher McCullough for key insights. The contributions of four (0; 0) without loss of generality. Then anonymous reviewers, who suggested improvements that are " # " # incorporated in this manuscript, is gratefully acknowledged. C M 1 nm = P (1) : (A.4) nm This research was carried out at the Jet Propulsion Laboratory, S (2n + 1)M nm e California Institute of Technology, under a contract with the The relationship between the fully normalized Legendre func- National Aeronautics and Space Administration. ©2020 Cal- tion P and the associated Legendre function P is ifornia Institute of Technology. Government sponsorship ac- nm nm knowledged. s (2  )(2n + 1)(n m)! m0 P = P ; (A.5) nm nm (n + m)! Appendix A. Multipole expansion and the Wahr equation for surface density where  is the Kronecker delta. Since P (1) = mn nm ; P (1) =  2n + 1: For a point mass at the pole Equa- m0 nm m0 The gravitational potential is conventionally expressed as the tion A.3 reduces to multipole expansion (Kaula (1966), Kaula (2013), Chao and " # " # Gross (1987)) C M 1 nm m0 = p : (A.6) 1 n X X n+1 S M nm e 2n + 1 G M a U (r; ; ) = P (cos ) (A.1) nm a r The potential from Equations A.1 and A.6 is independent of n=0 m=0 and is given by the multipole expansion (C cos m + S sin m): nm nm n+1 G M a As illustrated in Figure A.8 (r; ; ) = (distance from the cen- U (r; ) = P (cos ); (A.7) a r ter of the Earth, co-latitude, longitude), (a; M ) = (Earth radius, n=0 Robert Spero / Advances in Space Research xx (2020) xxx-xxx 7 the familiar expansion from electrostatics for the azimuthally The quantity T is the N = 1 limit of the truncated sum, symmetric electric field from a point charge (Jackson (2007)) defined as and from gravitational potential theory (Blakely (1996), Section 6.4.2). T () = P P N n1 n+1 The Wahr equation for surface density from (C ; S ) nm nm n=1 (Wahr et al. (1998)) is = (P + P (P + P ) 0 1 N N+1 1 n X X a 2n + 1 ave 0 0 0 1 +  (P + P ) N N+1 ( ;  ) = P (cos  ) nm = : (A.14) 3 1 + k n=0 m=0 0 0 (C cos m + S sin m ); (A.8) For a small spherical cap, << 1 (and  = cos slightly < nm nm 1), the cap area is A = (a ) : Using  = M=A and  = ave where k = Love number. n 3M =(4a ); Equation A.13 is equivalent to To study the error of a finite-degree spherical harmonic ap- T (cos ) proximation to a point mass, consider a spherical cap in the 1 = 0: (A.15) limit of small cap size. The spherical cap is centered at coor- 0 0 dinates ( ;  ) and its angular radius is and   cos : As The fractional error in  due to truncation of the summation computed by Pollack (1973), the Stokes coecients are Equation A.13 at order N is " # " # M P P C n+1 n1 cos m nm T (cos ) 0 N = P (cos  ) ;(A.9) nm  = 1  P (cos ): (A.16) N N sin m M (2n + 1) (1 ) nm e 4 See Figure A.9 for  with small spherical caps of two di erent where we use the shorthand P () = P : For a spherical cap at j j sizes. The slow reduction of j j with increasing N shows that the north pole, the unfiltered spherical harmonic expansion is ill-suited to char- " # " # M P P C  acterize the field from a point-like source. The truncation error nm n+1 n1 m0 = : (A.10) 3=2 0 is often reduced by applying a spectral localizing filter (Panet S M (2n + 1) (1 ) nm e et al. (2013), Appendix 2); see also Wahr et al. (1998), Swenson The spherical cap reduces to a point mass in the limit of = and Wahr (2002), Seo et al. (2005), and Werth et al. (2009). 0; or  = 1; substituting lim [P P ] = ( 1)(2n + 1) = into Equation A.10 !1 n+1 n1 gives Equation A.6. By comparing expressions similar to Equation A.1 and Equation A.8, Dickey et al. (1997) identifies 2n + 1 ˆ ˆ ave C + S = (C + S ) (A.11) nm nm nm nm 3 1 + k w n where  is the density of water as the transformation to con- vert geoid expansion coecients (C ; S ) to mass expansion nm nm ˆ ˆ coecients (C ; S ), p. 101 their Equation (B5). nm nm At the pole, from Equation A.8, dropping the n = 0 term that represents the total potential of the Earth, and neglecting the Earth’s elasticity by setting k = 0; n Fig. A.9. Truncation error, Equation A.16, in representing the field from a mass of small spatial extent by spherical harmonic expansion of order N: The dashed lines follow the large N asymptote envelope, P (cos )   2=(N sin ): a N ave = P (1)(2n + 1)C : (A.12) nm nm n=1 From Equation A.10, References M a ave M 1  3 Abich, K., Abramovici, A., Amparan, B., Baatzsch, A., Bachman Okihiro, X B. B., Barr, D. C., Bize, M. P., Bogan, C., Braxmaier, C., Burke, M. J., 2n + 1 P (1) (P P ) et al. (2019). In-orbit performance of the GRACE Follow-on laser ranging nm n+1 n1 3=2 (2n + 1) interferometer. Physical Review Letters, 123(3):031101. n=1 Abramowitz, M. and Stegun, I. A. (1964). Handbook of Mathematical Func- M a ave tions with Formulas, Graphs, and Mathematical Tables. Dover, New York, = P P n+1 n1 ninth Dover printing, tenth GPO printing edition. M 1  3 n=1 Bettadpur, S. (2018). GRACE L-2 Product User Manual. Center for Space M a Research, The University of Texas at Austin. GRACE 327-734, CSR-GR- ave = T : (A.13) 03-01. M 3 e 8 Robert Spero / Advances in Space Research xx (2020) xxx-xxx Blakely, Richard J. (1996). Potential Theory in Gravity and Magnetic Applica- of Geophysical Research, 78(11):1760–1768. tions. Cambridge University Press. Reigber, C., Schwintzer, P., Neumayer, K.-H., Barthelmes, F., Konig, ¨ R., Chao, B. F. and Gross, R. S. (1987). Changes in the Earth’s rotation and low- Forste, ¨ C., Balmino, G., Biancale, R., Lemoine, J.-M., Loyer, S., et al. degree gravitational field induced by earthquakes. Geophysical Journal In- (2003). The CHAMP-only Earth gravity field model EIGEN-2. Advances ternational, 91(3):569–596. in Space Research, 31(8):1883–1888. Christophe, B., Marque, J., and Foulon, B. (2010). In-orbit data verification of Seo, K.-W., Wilson, C., Chen, J., Famiglietti, J., and Rodell, M. (2005). Filters the accelerometers of the ESA GOCE mission. In SF2A-2010: Proceedings to estimate water storage variations from GRACE. In IAG Symp., 128, 607– of the Annual meeting of the French Society of Astronomy and Astrophysics, 611. Springer. volume 1, page 113. Swenson, S. and Wahr, J. (2002). Methods for inferring regional surface-mass Colombo, O. and Chao, B. (1992). Global gravitational change from space in anomalies from gravity recovery and climate experiment (GRACE) mea- 2001. In IAG Symp., 112, 71–74. surements of time-variable gravity. Journal of Geophysical Research: Solid Conklin, J. and Nguyen, A. N. (2017). Drag-free control and drag force re- Earth, 107(B9):ETG–3. covery of small satellites. In 31st Annual AIAA/USU Conference on Small Tapley, B. D (1997). Evaluation of Flat-Earth Approximation Results for Satellites. Geopotential Missions. Journal of Guidance, Control, and Dynamics, Dickey, J., Bentley, C. R., Bilham, R., Carton, J., Eanes, R., Herring, T. A., 20(2):246–252. Kaula, W., Lagerleof, G., Rojstaczer, S., Smith, W., et al. (1997). Satellite Tapley, B. D., Bettadpur, S., Ries, J. C., Thompson, P. F., and Watkins, M. M. gravity and the geosphere. National Research Council Report, 112. (2004). GRACE measurements of mass variability in the Earth system. Sci- Drinkwater, M. R., Haagmans, R., Muzi, D., Popescu, A., Floberghagen, R., ence, 305(5683):503–505. Kern, M., and Fehringer, M. (2006). The GOCE gravity mission: ESA’s Touboul, P., Willemenot, E., Foulon, B., and Josselin, V. (1999). Accelerom- first core Earth explorer. In Proceedings of the 3rd international GOCE user eters for CHAMP, GRACE and GOCE space missions: synergy and evolu- workshop, pages 6–8. tion. Boll. Geof. Teor. Appl, 40(3-4):321–327. Flanagan, E. E. and Hughes, S. A. (1998). Measuring gravitational waves from Touboul, P., Foulon, B., and Willemenot (1999a). Electrostatic space binary black hole coalescences. i. signal to noise for inspiral, merger, and accelerometers for present and future missions. Acta Astronautica, ringdown. Phys. Rev. D, 57:4535–4565. 45(10):605–617. Ghobadi-Far, K., Han, S.-C., Weller, S., Loomis, B. D., Luthcke, S. B., Mayer- Wahr, J., Molenaar, M., and Bryan, F. (1998). Time variability of the Gurr ¨ , T., and Behzadpour, S. (2018). A transfer function between line-of- Earth’s gravity field: Hydrological and oceanic e ects and their possible sight gravity di erence and GRACE intersatellite ranging data and an ap- detection using GRACE. Journal of Geophysical Research: Solid Earth, plication to hydrological surface mass variation. Journal of Geophysical 103(B12):30205–30229. Research: Solid Earth, 123(10):9186–9201. Wainstein, L. A. and Zubakov, V. (1970). Extraction of signals from noise. Ghobadi-Far,K. Han, S.-C., McCullough, C. M. Wiese, D. N., Yuan, D. PrenticeHall, Englewood Cli s, NJ. Landerer, F. W. Sauber, J, Watkins, M. M., (2020) GRACE Follow-On Watkins, M., Wiese, D. N., Yuan, D.-N., Boening, C., and Landerer, F. W. Laser Ranging Interferometer Measurements Uniquely Distinguish Short- (2015). Improved methods for observing Earth’s time variable mass distri- Wavelength Gravitational Perturbations. Geophysical Research Letters, bution with GRACE using spherical cap mascons. Journal of Geophysical 47(16):e2020GL089445. Research: Solid Earth, 120(4):2648–2671. Han, S.-C. (2013). Determination and localized analysis of intersatellite line of Watkins, M., Yuan, D., Kuang, D., Bertiger, W., Kim, M., and Kruizinga, G. sight gravity di erence: Results from the GRAIL primary mission. Journal (2005). GRACE harmonic and mascon solutions at JPL. AGU Fall Meeting, of Geophysical Research: Planets, 118(11):2323–2337. 2005:G22A–04. Hauk, M. and Wiese, D. N., (2020). New Methods for Linking Science Ob- Werth, S., Guntner ¨ , A., Schmidt, R., and Kusche, J. (2009). Evaluation of jectives to Remote Sensing Observations: A Concept Study Using Single- GRACE filter tools from a hydrological perspective. Geophysical Journal and Dual-Pair Satellite Gravimetry Architectures. Earth and Space Science, International, 179(3):1499–1515. 7(3):e2019EA000922. Wol , M. (1969). Direct measurements of the Earth’s gravitational potential Jackson, J. D. (2007). Classical electrodynamics. John Wiley & Sons. using a satellite pair. Journal of Geophysical Research, 74(22):5295–5300. Kaula, W. M. (1966). Tests and combination of satellite determinations Wong, L., Buechler, G., Downs, W., Sjogren, W., Muller, P., and Gottlieb, of the gravity field with gravimetry. Journal of Geophysical Research, P. (1971). A surface-layer representation of the lunar gravitational field. 71(22):5303–5314. Journal of Geophysical Research, 76(26):6220–6236. Kaula, W. M. (2013). Theory of satellite geodesy: applications of satellites to geodesy. Courier Corporation. Konopliv, A. S., Park, R. S., Yuan, D.-N., Asmar, S. W., Watkins, M. M., Williams, J. G., Fahnestock, E., Kruizinga, G., Paik, M., Strekalov, D., et al. (2013). The JPL lunar gravity field to spherical harmonic degree 660 from the GRAIL primary mission. Journal of Geophysical Research: Planets, 118(7):1415–1434. Landerer, F., Flechtner, F., Save, H., Webb, F., Bandikova, T., and Bertiger, WI, et al. (2020). Extending the global mass change data record: GRACE follow-on instrument and science data performance. Geophysical Research Letters, 47(12). McCullough, C., Harvey, N., Save, H., Bandikova, T (2019). Description of Calibrated GRACE-FO Accelerometer Data Products (ACT) GRACE-FO Level-1 Product Version 04 Manual JPL D-103863 Muller ¨ , V. (2017). Design considerations for future geodesy missions and for space laser interferometry Ph.D. Thesis, Gottfried Wilhelm Leibniz Univer- sitat ¨ Hannover, 2017. Numata, K., Kemery, A., and Camp, J. (2004). Thermal-noise limit in the frequency stabilization of lasers with rigid cavities. Physical Review Letters, 93(25):250602. Panet, I., Flury, J., Biancale, R., Gruber, T., Johannessen, J., van den Broeke, M., van Dam, T., Gegout, P., Hughes, C., Ramillien, G., et al. (2013). Earth system mass transport mission (e. motion): a concept for future Earth gravity field measurements from space. Surveys in Geophysics, 34(2):141–163. Pollack, H. N. (1973). Spherical harmonic representation of the gravitational potential of a point mass, a spherical cap, and a spherical rectangle. Journal http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Physics arXiv (Cornell University)

Mass sensitivity of gravimetric satellites

Physics , Volume 2021 (2006) – Jun 15, 2020

Loading next page...
 
/lp/arxiv-cornell-university/mass-sensitivity-of-gravimetric-satellites-UuZHcxjX20
ISSN
0273-1177
eISSN
ARCH-3341
DOI
10.1016/j.asr.2020.12.019
Publisher site
See Article on Publisher Site

Abstract

Frequency-domain expressions are found for gradiometer and satellite-to-satellite tracking measurements of a point source on the surface of the Earth. The maximum signal-to-noise ratio as a function of noise in the measurement apparatus is computed, and from that the minimum detectable point mass is inferred. A point mass of magnitude M = 100 Gt gives a signal-to-noise ratio of 3 when a GOCE-like gradiometer passes directly over the mass. On the satellite-to-satellite tracking mission GRACE-FO M = 1:3 Gt for the microwave instrument and M = 0:5 Gt for 3 3 the laser ranging interferometer. The sensitivity of future GRACE-like missions with di erent orbital parameters and improved accelerometer sensitivity is explored, and the optimum spacecraft separation for detecting point-like sources is found. The future-mission benefit of improving the accelerometer sensitivity for measurement of non-gravitational disturbances is shown by the resulting reduction of M ; to as small as 7 Mt for 500 km orbital altitude and optimized satellite separation of 900 km. © 2020 COSPAR. Published by Elsevier Ltd All rights reserved. 1. Introduction to GRAIL data to solve for regional fields of the Moon. A global gravity map is the principal data product of satel- Though they di er slightly in assumptions and results, the lite gravity missions. Previously CHAMP (Reigber et al. spherical harmonic and mascon methods are constructed to an- (2003)), GRACE (Tapley et al. (2004)), and GOCE (Drinkwa- swer the same question: what is the gravity field that is most ter et al. (2006)) collected data to map the Earth’s gravity, and consistent with measurements? Here we address a di erent GRAIL (Konopliv et al. (2013)) measured the Moon’s gravity. question: what is the limit to measurement precision of a point- Currently GRACE Follow-On (GRACE-FO, Landerer et al. like mass on the surface? This is an artificial model, a single (2020)) is extending the GRACE data record, with increased mascon, that is not directly applicable to the geodetic agenda of ranging precision a orded by its laser ranging interferometer measuring the Earth’s gravity. We do not attempt to replicate (LRI, Abich et al. (2019)). the global gravity-field inversion achieved by the usual many- As pointed out by Watkins et al. (2015), the most com- mascon analysis. Rather, the motivation for this analysis is monly used method of analyzing satellite gravity data is based twofold: to provide a single-number figure of merit, namely on global gravity fields expressed in terms of spherical har- the minimum detectable isolated point mass perturbation, and monic basis functions. An alternative to spherical harmonics to find the optimal filter for such a detection. The minimum is the mass concentration, or mascon, model. Starting with detectable mass M is defined as the point mass that a gives a Wong et al. (1971), the mascon approach was applied to single- signal-to-noise ratio  = 3 in a single orbital pass directly over satellite lunar orbital measurements to infer the surface gravity the point mass. It is calculated by applying the Wiener optimal of the moon. Mascons can be modeled as many discrete sources filter to the problem of detecting a signal of known waveform, (Pollack (1973), Watkins et al. (2005)) that cover the globe, or against a background specified by instrument noise power spec- used to solve for regional fields. Han (2013) applied mascons tral density (Wainstein and Zubakov (1970)). A comparison of https://dx.doi.org/10.1016/j.jasr.xxxx.xx.xxx 0273-1177/© 2020 COSPAR. Published by Elsevier Ltd All rights reserved. arXiv:2006.08179v9 [physics.geo-ph] 12 Dec 2020 2 Robert Spero / Advances in Space Research xx (2020) xxx-xxx M for di erent orbital configurations and instrument sensitivi- ties guides the design of future missions. Additionally, we find the optimum satellite separation in GRACE-like missions for a specified instrument noise power spectral density. 2. Gradiometer Mass Sensitivity Consider a gradiometer flying directly over a point mass M at altitude h; Figure 1 left. At orbital altitude h = 330 km; the Fig. 2. Orbiting gradiometer vertical gradient response to a point mass. Upper- Fig. 1. Measurement of gravitational field from a point mass M at along-track left: g (t); z-direction gradient in time domain, Equation 1; Upper-right: distance x, altitude h; and velocity v: Left: Vertical gradient g : Right: Di er- g ( f ) = Fourier transform, Equation 2; Lower: M = Detectable mass with z 3 ential acceleration a between spacecraft 1 and 2 with average separation L. signal-to-noise = 3, Equation 5. along-track velocity v is the orbital velocity v = 7:7 km=s and In general,  depends on what filtering is applied to the in- the along-track distance x changes at approximately constant strument output. After Flanagan and Hughes (1998), with rate, x = v t: The acceleration at the spacecraft in the vertical, 2 2 3=2 optimum filtering the maximum signal-to-noise ratio per unit z direction is a = G Mz=(x + z ) ; and the gradient in the z 2 2 2 5=2 2 2 3=2 source mass  is direction is g = da =dz = G M[3z (x + z ) (x + z ) ]: z z G = Newton’s constant of gravitation. Substituting z ! h and Z x ! v t, = 4 W ( f )d f: (4) h i 2 5=2 2 3=2 g (t) =  M 3(1 + [ f t] ) (1 + [ f t] ) (1) z g h h Prime superscripts indicate quantities normalized by the source 3 3 0 0 2 where  M = G M=h = 1:85 10 mE with M = 1 Gt and g mass:  = =M and W ( f ) = W ( f )=M : It follows that the g z 12 2 f = v =h = 23 mHz; 1 mE = 1 10 s : At t = 0, when h o minimum detectable point mass with  = 3 is the gradiometer is directly above the source mass, the gradient reaches its maximum value g = 2 M: At x =  2h; g = 0: z g z M = q : (5) The Fourier transform of p(t) is defined by F [ p(t)] 4 W ( f )d f 1 z p( f ) = dt p(t) exp(2i f t): Applying the Fourier transform to g (t): The lower panel of Figure 2 shows M as a function of or- z 3 " ! !# bital altitude h for a gradiometer limited by white spectral 4 f 2 f 2 f 2 f noise S ( f ) = 1 mE= Hz (approximated value for GOCE g ( f ) =  M K + K (2) z g 1 0 f f f f h h h h from Touboul et al. (1999), Touboul et al. (1999a)). At h = 330 km; M = 100 Gt and the minimum observable gradient at where K is the modified Bessel function of the second kind, the peak time t = 0 is 2 M = 0:37 mE: order n. From Abramowitz and Stegun (1964), Section 9.7.2, K (z)  =(2z)e ; where  indicates approximately equal for large z: At frequency f  f ; the measurement response 3. Sensitivity of GRACE-like measurements is attenuated approximately exponentially with e-folding fre- quency f =(2) = 3:7 mHz: This corresponds to harmonic or- The measurement configuration and signal parameters for der N = f =(2 f ) = 20; where f = 0:183 mHz is the orbital the low-low satellite-to-satellite tracking (SST) of GRACE and h 1 1 frequency. GRACE-FO are shown in Figure 1, right. The primary signal is The signal-to-noise ratio  depends on the signal and the the along-track di erential position of the spacecraft, measured power spectral density of the gradiometer noise, S ( f ): Define by microwave ranging or laser interferometry. the frequency-dependent signal-to-noise-ratio density SNRD as 3.1. Single Spacecraft Acceleration For simplicity, the Flat-Earth approximation (Tapley (1997)) jg ( f )j W ( f ) = : (3) is used. This approximation neglects centrifugal acceleration, S ( f ) g Robert Spero / Advances in Space Research xx (2020) xxx-xxx 3 which introduces an error of less than 20% at Fourier frequen- cies greater than 2 mHz (Ghobadi-Far et al. (2018), Figure 1; Muller ¨ (2017), Figure 1.7). In this approximation, the acceler- ation on spacecraft 1 flying over point mass M at along-track 2 2 3=2 distance x is a = G M x=(h + x ) : Define the acceleration per unit source mass, a = a =M: Then Gtv a = (6) 2 2 2 3=2 (h + v t ) Converting to frequency space, 2 f Fig. 3. Range acceleration resulting from a square mass centered under the flight 4i f GK 0 0 along-track path. Orbital altitude h, average spacecraft separation L and source a ( f ) = F [a (t)] = : (7) 1 1 mass M are indicated in the title. The separate traces are for squares of side S indicated in the legend. Left: time domain, a (t); Right: Fourier Transform, a ( f ): 3.2. Range Acceleration Signal where The acceleration experienced by spacecraft 2 is the same as v = orbital velocity =7:6 km=s spacecraft 1 at distance L, but delayed by  = L=v : The re- h = orbital altitude =500 km sulting (along-track) range acceleration between the spacecraft L = spacecraft separation =200 km (Figure 3, left) is similar to what Han (2013) computed for the response of the GRAIL spacecraft to regional lunar gravity. The f = =15:2 mHz p h peak range acceleration a is f = =76 mHz L=2 GL a = M   M; (8) R  = =26 s 3=2 2 2 h + (L=2) o GL = =0:101 nm=s =Gt: 3=2 2 2 3=2 2 2 where  = GL(h + (L=2) ) : h + (L=2) Using the identity F (delay ) = exp(2i f), the range ac- These numerical values apply to GRACE-FO. The approxi- celeration in the frequency domain, a ( f ); is given by mately exponential attenuation with frequency of ja ( f )j has e-folding frequency 2:5 mHz, corresponding to harmonic order 0 0 2i f a ( f ) = a ( f )(1 e ) (9) N = 14: R 1 0 0 To explore the valid realm of the point-mass approximation, ja ( f )j = 2ja ( f ) sin( f)j (10) R 1 Figure 3 shows the range acceleration signal from a square- shaped planar mass of side length S, computed by numerical That is, in the frequency domain the range acceleration is integration. The S=1 km result is in agreement with the point- the single-satellite acceleration multiplied by 2j sin( f)j: For mass analytical calculation, which is valid at the 20% level for f  1=; ja ( f )j / L, which is the response for the spacecraft sources as large as S=300 km. Henceforth, we restrict our anal- pair acting as a gradiometer. The response departs from that ysis to the signal from a point source. of a gradiometer at large L; most conspicuously in the form of Equation 11 gives the measurement impulse response; that high-frequency nulls where the signal vanishes. The first null is, the range acceleration frequency response to a point mass is at f = 1= = 38 mHz for low-Earth orbit and L = 200 km, null input. This facilitates the direct comparison of signal and noise as recognized by Wol (1969). In degree-variance evaluations amplitudes as computed in the following section, and yields an of measurement sensitivity, the first null is expressed as a max- expression for the minimum detectable mass for GRACE-like imum in geoid height error at degree N = f = f = 216 for null 1 measurements of point source perturbations to surface gravity. L = 200 km; and N = 86 for L = 500 km; where f = orbital frequency = 0:176 mHz. 3.3. Noise and Mass Sensitivity From Equations 7 and 10, Consider the range measurement made by the laser ranging interferometer (LRI) on GRACE-FO. Assuming the measure- ! ! 8 f G 2 f 2 f ment resolution is limited by the thermal noise of the laser ref- ja ( f )j = K sin ; (11) erence cavity (Numata et al. (2004)), the displacement noise v f f h L o 4 Robert Spero / Advances in Space Research xx (2020) xxx-xxx root power spectral density (rpsd) x ˜ and strain rpsd x ˜ =L LRI LRI are given by x ˜ ( f )=L = x = f; (12) LRI c where x is a constant. For the LRI (Abich et al. (2019)), x = c c 1 10 : The rpsd of the LRI range acceleration noise is S ( f ) = (2 f )  x ˜ ( f ): (13) LRI LRI Take for the accelerometer measurement noise rpsd on a sin- gle satellite of GRACE and GRACE-FO (Touboul et al. (1999)) S ( f ) = a ˜ 1 + : (14) ACC 0 Fig. 4. Total range acceleration noise rpsd S for various assumptions of in- strument noise. The ranging noise for the “GRACE-FO MWI+ACC” is the white displacement noise of the microwave measurement on GRACE-FO, equal 11 2 Estaimates of a ˜ and f range from 3 10 m=s = Hz 7 0 k to 6 10 m= Hz. The other two noise curves assume the ranging noise of and 10 mHz, respectively (Hauk and Wiese (2020)) to the LRI, Equation 12, shown as a dotted line. Two levels of a ˜ =[m=s = Hz] 11 13 10 2 are assumed: 7 10 for the GRACE-FO curves, and 7 10 for a future 1 10 m=s = Hz and 5 mHz, respectively (Christophe et al. mission such as Mass Change Mission (MCM). The solid black line is the signal (2010), Conklin and Nguyen (2017)) . We take as a compromise spectrum a ( f ) from a 1 Gt point mass, for h = 500 km and L = 200 km; units 11 2 a ˜ = 7 10 m=s = Hz and f = 5 mHz: Assuming that the 0 k m/s. The values of M for the three respective configurations are (Section 3.3) acceleration measurements on the two satellites are uncorre- 1:3 Gt and 0:5 Gt and 7 Mt. lated, the total accelerometer noise is double: S = 2S : ACC ACC Improved accelerometers in future missions (Christophe et al. (2010), Conklin and Nguyen (2017)) may have a ˜ = 13 2 7 10 m=s = Hz: The total instrument noise power spectral density is S = S + S : (15) a ACC LRI Figure 4 shows S for di erent ranging instrument and ac- celerometer noise spectra. The MWI ranging noise is approxi- mated by white displacement noise, x ˜ = 6 10 m= Hz: MWI This estimate is based on comparing MWI range measurements to simultaneous LRI range measurements. To guide the eye to the frequencies that have the largest signal, ja ( f )j for a 1 Gt Fig. 5. Signal-to-noise ratio integrand of Equation 16, W ( f ) for the “GRACE- point source from Figure 3 is overlaid as the solid black line. FO LRI + ACC” noise of Figure 4, signal from orbital h = 500 km and source mass M = 1 Gt and three di erent values of spacecraft separation L: The in- The units of ja j are m/s. tegrated signal-to-noise ratios  from Equation 17 are indicated in the legend. As in Section 2, define the SNRD for range acceleration 2 2 ja ( f )j =M W ( f ) = : (16) S ( f ) The corresponding detectable peak accelerations,  M ; are R 3 2 2 0:13 nm=s , 0:047 nm=s . 0 2 W ( f ) = W M is shown in Figure 5 for several values of L: Another assessment of mass sensitivity for SST laser rang- The oscillations with nulls at multiples of 1= = v =L degrade ing is inferred from Colombo and Chao (1992), who proposed for L beyond an optimum spacecraft separation. a laser ranging mission that, with (h; L) = (600; 500) km was The optimal signal-to-noise ratio per unit mass is found by simulation to have sensitivity to weekly changes of 1 1 mm water height over a square region 400 km across, or mass = 4 W ( f )d f; (17) sensitivity of 160 Mt. In comparison, we find for the LRI on GRACE-FO at the same (h; L), M = 400 Mt: The two mea- surements have di erent assumed instrument sensitivity and av- and the source mass that gives  = 3 is (cf. Equation 5) eraging times (week-to-week vs. single-pass). Figure 6 shows the mass sensitivity M as a function of M = q : (18) 0 h and L for the LRI ranging instrument with two di er- 4 W ( f )d f ent levels of accelerometer sensitivity: a ˜ = 7 10 and 13 2 From Equations 11 through 16 and Equation 18, the 7 10 m=s = Hz. The lower row of Figure 6 shows the GRACE-FO parameters with the microwave ranging instru- optimum L for a given h and the resulting M : The optimum ment (MWI) and LRI give respectively M = 1:3 Gt; 0:5 Gt: L for the LRI on GRACE-FO, operating at h = 500 km, is 3 Robert Spero / Advances in Space Research xx (2020) xxx-xxx 5 L = 900 km; which would give M = 200 Mt: That reflects a Residuals are relative to a reference field. The admittance Z( f ) potential factor of 2.5 improvement over M = 500 Mt for the is defined as the ratio of power spectra, nominal satellite separation of L = 200 km: A future mission S LOS ( f ) ; ¨ g with the improved a ˜ ; h = 500 km; and optimal satellite separa- Z( f ) = ; (22) S ( f ) tion L = 900 km has M = 7 Mt: ; ¨  ¨ where S ( f ) is the power spectrum of the MWI range accel- ; ¨  ¨ eration measurement and S LOS ( f ) is the cross-power spec- ; ¨ g trum between the range acceleration and the LOS gravity di er- ence. Keeping the shorthand notation p( f ) = F [ p(t)]; Z( f ) is a filter that transforms residual range acceleration  ¨( f ) = a ( f ) LOS LOS to g ( f ), an estimate of g ( f ) : F 12 LOS g ( f ) = Z( f ) ¨( f ): (23) Z( f ); normalized to have a maximum value of 1, is shown as the dashed trace in Figure 7. Z( f ) is the optimal filter to apply to MWI range acceleration, based on the measurement data that includes signal from the gravity field. It applies to extracting the best SNR from a residual regional or global field and does not explicitly depend on instrument noise spectra. In contrast, G( f ) is fine-tuned to the problem of detecting the specific waveform of a point mass, in the presence of known measurement noise. Since a point mass generates a field with Fig. 6. Mass sensitivity of the LRI measurement on GRACE-FO left, and of the highest possible frequency content, the G( f ) passband starts a future GRACE-like mission right. Upper row shows isomass M contours, higher in frequency than Z( f ): The LRI G( f ) passband is higher in Mt, from Equation 18. Equation 12 specifies the ranging noise, and ac- 11 13 celerometer noise is given by Equation 14 with a ˜ = (7 10 ; 7 10 ) than for the MWI because the LRI measurement has reduced m=s = Hz, with fixed f = 5 mHz: Lower row shows the optimum L as a noise at high frequency. function of h; (blue, left axis) and the resulting sensitivity M (red, right axis). A practical use for the G( f ) filter is searching for un- known point-like features, such as underground water stor- age of 100 km spatial extent. The filter would be applied to 3.4. Optimal filter range acceleration measurements after subtracting the e ect of the known field, including time-varying gravity, and non- The filter that gives maximum signal-to-noise ratio is (Wain- gravitational accelerations. stein and Zubakov (1970), Chapter 3) a ( f ) G( f ) = ; (19) S ( f ) with denoting complex conjugation. The filter’s input is the measured range acceleration. G( f ) is an example of a filter for extracting a signal of known waveform, in this case the range acceleration resulting from flying over a point mass. Dropping the multiplicative constants, the filter magnitude is 2 f 2 f f K sin f f h L jG( f )j = : (20) S ( f ) Fig. 7. Transfer functions for range acceleration data. The blue trace that peaks Normalized jG( f )j for the MWI and LRI on GRACE-FO are at 2:6 mHz is jG( f )j for the MWI on GRACE-FO, and the red trace that peaks at 5:7 mHz is for the LRI on GRACE-FO. Z( f ) is the admittance filter from shown in Figure 7. Ghobadi-Far et al. (2018) Figure 2(b), and applies to the MWI on GRACE-FO. Ghobadi-Far et al. (2018) analyzed the GRACE-FO MWI All curves are normalized to give maximum value of 1. signal in terms of the line-of-sight gravity di erence, and ap- plied the same analysis method to the GRACE-FO LRI sig- nal in Ghobadi-Far et al. (2020). In their analysis of MWI 4. Conclusion data, Ghobadi-Far et al. (2018) defined the gravimetric quan- LOS tity g ; or line-of-sight (LOS) gravity di erence, which dif- We derived the optimum sensitivity of orbiting gravimetric fers from the range acceleration residual  ¨ by  ; the residual satellites to a point source, that is a single mascon. The signal- centrifugal acceleration: to-noise ratio is found as a function of instrument noise and LOS g =  ¨ +  : (21) orbital parameters. The signal is converted to frequency space 12 6 Robert Spero / Advances in Space Research xx (2020) xxx-xxx by the Fourier transform, and the signal-to-noise ratio is derived from optimal filtering a signal of known waveform. This anal- ysis di ers from the conventional approach of spherical har- monic expansion to characterize the field from an arbitrary mass distribution. Such an expansion requires a very large harmonic order to accurately approximate the field from a point source, as shown in Appendix A. The frequency response of an orbiting gradiometer to a point mass directly under the flight track is approximated by Equation 2 that depends only on the orbital altitude and the magnitude of the point mass. Likewise, for an SST-based measurement of the gravitational field, the range acceleration is approximated by Equation 11 that includes dependency on the average satel- lite separation. Applying Wiener optimal filter theory, these Fig. A.8. Coordinate system for the spherical harmonic expansion of geopoten- responses and the noise spectra of the ranging measurement tial, Kaula (2013). The satellite constellation position, defined as the center of and of accelerometer-based measurement of non-gravitational mass for a gradiometer or (illustrated) the center of the line of sight between forces give ; the maximum achievable signal-to-noise ratio. two SST satellites relative to the center of the Earth, is ~ r(r; ; ), where r is the The resolvable mass M is defined as the magnitude of the point distance from the center of the Earth,  is the co-latitude, and  is the longitude. mass that gives  = 3: M is the ultimate mass sensitivity, and realistic non-point mass distributions that are not directly un- der the flight track will give larger M in practice. Nonetheless, Earth mass), and P is the fully-normalized associated Leg- nm M provides a figure of merit for comparing future missions endre function. The field is entirely specified by the Stokes with di erent orbits and instrument sensitivities to guide the coecients (C ; S ): nm nm design of such missions. For SST measurements M has a min- 0 0 0 0 For a known mass distribution d M = (r ;  ;  )dV with imum value at a calculable satellite separation L; giving the op- primes designating the source mass coordinates, (C ; S ) are nm nm timum separation for discovering point-like (meaning less than evaluated as the volume integral Bettadpur (2018) approximately 300 km) features such as subsurface water stor- " # Z ! " # age. Equation 20 specifies the optimal filter for such a search. 0 0 1 r C cos m nm = d M P ( ) ;(A.2) nm As a caveat, the M metric and its L optimization does not apply 0 sin m S (2n + 1)M 0 a nm e to large-scale gravimetry, such as required by oceanography. 0 0 0 where  = cos  : For a point mass at r = a " # " # Acknowledgments C cos m nm 0 = P ( ) : (A.3) nm 0 sin m S (2n + 1)M nm The author thanks Kirk McKenzie, Gabriel Ramirez, Pep e Sanjuan and David Wiese for useful discussions, and Christo- 0 0 A single point mass can be placed at the north pole, ( ;  ) = pher McCullough for key insights. The contributions of four (0; 0) without loss of generality. Then anonymous reviewers, who suggested improvements that are " # " # incorporated in this manuscript, is gratefully acknowledged. C M 1 nm = P (1) : (A.4) nm This research was carried out at the Jet Propulsion Laboratory, S (2n + 1)M nm e California Institute of Technology, under a contract with the The relationship between the fully normalized Legendre func- National Aeronautics and Space Administration. ©2020 Cal- tion P and the associated Legendre function P is ifornia Institute of Technology. Government sponsorship ac- nm nm knowledged. s (2  )(2n + 1)(n m)! m0 P = P ; (A.5) nm nm (n + m)! Appendix A. Multipole expansion and the Wahr equation for surface density where  is the Kronecker delta. Since P (1) = mn nm ; P (1) =  2n + 1: For a point mass at the pole Equa- m0 nm m0 The gravitational potential is conventionally expressed as the tion A.3 reduces to multipole expansion (Kaula (1966), Kaula (2013), Chao and " # " # Gross (1987)) C M 1 nm m0 = p : (A.6) 1 n X X n+1 S M nm e 2n + 1 G M a U (r; ; ) = P (cos ) (A.1) nm a r The potential from Equations A.1 and A.6 is independent of n=0 m=0 and is given by the multipole expansion (C cos m + S sin m): nm nm n+1 G M a As illustrated in Figure A.8 (r; ; ) = (distance from the cen- U (r; ) = P (cos ); (A.7) a r ter of the Earth, co-latitude, longitude), (a; M ) = (Earth radius, n=0 Robert Spero / Advances in Space Research xx (2020) xxx-xxx 7 the familiar expansion from electrostatics for the azimuthally The quantity T is the N = 1 limit of the truncated sum, symmetric electric field from a point charge (Jackson (2007)) defined as and from gravitational potential theory (Blakely (1996), Section 6.4.2). T () = P P N n1 n+1 The Wahr equation for surface density from (C ; S ) nm nm n=1 (Wahr et al. (1998)) is = (P + P (P + P ) 0 1 N N+1 1 n X X a 2n + 1 ave 0 0 0 1 +  (P + P ) N N+1 ( ;  ) = P (cos  ) nm = : (A.14) 3 1 + k n=0 m=0 0 0 (C cos m + S sin m ); (A.8) For a small spherical cap, << 1 (and  = cos slightly < nm nm 1), the cap area is A = (a ) : Using  = M=A and  = ave where k = Love number. n 3M =(4a ); Equation A.13 is equivalent to To study the error of a finite-degree spherical harmonic ap- T (cos ) proximation to a point mass, consider a spherical cap in the 1 = 0: (A.15) limit of small cap size. The spherical cap is centered at coor- 0 0 dinates ( ;  ) and its angular radius is and   cos : As The fractional error in  due to truncation of the summation computed by Pollack (1973), the Stokes coecients are Equation A.13 at order N is " # " # M P P C n+1 n1 cos m nm T (cos ) 0 N = P (cos  ) ;(A.9) nm  = 1  P (cos ): (A.16) N N sin m M (2n + 1) (1 ) nm e 4 See Figure A.9 for  with small spherical caps of two di erent where we use the shorthand P () = P : For a spherical cap at j j sizes. The slow reduction of j j with increasing N shows that the north pole, the unfiltered spherical harmonic expansion is ill-suited to char- " # " # M P P C  acterize the field from a point-like source. The truncation error nm n+1 n1 m0 = : (A.10) 3=2 0 is often reduced by applying a spectral localizing filter (Panet S M (2n + 1) (1 ) nm e et al. (2013), Appendix 2); see also Wahr et al. (1998), Swenson The spherical cap reduces to a point mass in the limit of = and Wahr (2002), Seo et al. (2005), and Werth et al. (2009). 0; or  = 1; substituting lim [P P ] = ( 1)(2n + 1) = into Equation A.10 !1 n+1 n1 gives Equation A.6. By comparing expressions similar to Equation A.1 and Equation A.8, Dickey et al. (1997) identifies 2n + 1 ˆ ˆ ave C + S = (C + S ) (A.11) nm nm nm nm 3 1 + k w n where  is the density of water as the transformation to con- vert geoid expansion coecients (C ; S ) to mass expansion nm nm ˆ ˆ coecients (C ; S ), p. 101 their Equation (B5). nm nm At the pole, from Equation A.8, dropping the n = 0 term that represents the total potential of the Earth, and neglecting the Earth’s elasticity by setting k = 0; n Fig. A.9. Truncation error, Equation A.16, in representing the field from a mass of small spatial extent by spherical harmonic expansion of order N: The dashed lines follow the large N asymptote envelope, P (cos )   2=(N sin ): a N ave = P (1)(2n + 1)C : (A.12) nm nm n=1 From Equation A.10, References M a ave M 1  3 Abich, K., Abramovici, A., Amparan, B., Baatzsch, A., Bachman Okihiro, X B. B., Barr, D. C., Bize, M. P., Bogan, C., Braxmaier, C., Burke, M. J., 2n + 1 P (1) (P P ) et al. (2019). In-orbit performance of the GRACE Follow-on laser ranging nm n+1 n1 3=2 (2n + 1) interferometer. Physical Review Letters, 123(3):031101. n=1 Abramowitz, M. and Stegun, I. A. (1964). Handbook of Mathematical Func- M a ave tions with Formulas, Graphs, and Mathematical Tables. Dover, New York, = P P n+1 n1 ninth Dover printing, tenth GPO printing edition. M 1  3 n=1 Bettadpur, S. (2018). GRACE L-2 Product User Manual. Center for Space M a Research, The University of Texas at Austin. GRACE 327-734, CSR-GR- ave = T : (A.13) 03-01. M 3 e 8 Robert Spero / Advances in Space Research xx (2020) xxx-xxx Blakely, Richard J. (1996). Potential Theory in Gravity and Magnetic Applica- of Geophysical Research, 78(11):1760–1768. tions. Cambridge University Press. Reigber, C., Schwintzer, P., Neumayer, K.-H., Barthelmes, F., Konig, ¨ R., Chao, B. F. and Gross, R. S. (1987). Changes in the Earth’s rotation and low- Forste, ¨ C., Balmino, G., Biancale, R., Lemoine, J.-M., Loyer, S., et al. degree gravitational field induced by earthquakes. Geophysical Journal In- (2003). The CHAMP-only Earth gravity field model EIGEN-2. Advances ternational, 91(3):569–596. in Space Research, 31(8):1883–1888. Christophe, B., Marque, J., and Foulon, B. (2010). In-orbit data verification of Seo, K.-W., Wilson, C., Chen, J., Famiglietti, J., and Rodell, M. (2005). Filters the accelerometers of the ESA GOCE mission. In SF2A-2010: Proceedings to estimate water storage variations from GRACE. In IAG Symp., 128, 607– of the Annual meeting of the French Society of Astronomy and Astrophysics, 611. Springer. volume 1, page 113. Swenson, S. and Wahr, J. (2002). Methods for inferring regional surface-mass Colombo, O. and Chao, B. (1992). Global gravitational change from space in anomalies from gravity recovery and climate experiment (GRACE) mea- 2001. In IAG Symp., 112, 71–74. surements of time-variable gravity. Journal of Geophysical Research: Solid Conklin, J. and Nguyen, A. N. (2017). Drag-free control and drag force re- Earth, 107(B9):ETG–3. covery of small satellites. In 31st Annual AIAA/USU Conference on Small Tapley, B. D (1997). Evaluation of Flat-Earth Approximation Results for Satellites. Geopotential Missions. Journal of Guidance, Control, and Dynamics, Dickey, J., Bentley, C. R., Bilham, R., Carton, J., Eanes, R., Herring, T. A., 20(2):246–252. Kaula, W., Lagerleof, G., Rojstaczer, S., Smith, W., et al. (1997). Satellite Tapley, B. D., Bettadpur, S., Ries, J. C., Thompson, P. F., and Watkins, M. M. gravity and the geosphere. National Research Council Report, 112. (2004). GRACE measurements of mass variability in the Earth system. Sci- Drinkwater, M. R., Haagmans, R., Muzi, D., Popescu, A., Floberghagen, R., ence, 305(5683):503–505. Kern, M., and Fehringer, M. (2006). The GOCE gravity mission: ESA’s Touboul, P., Willemenot, E., Foulon, B., and Josselin, V. (1999). Accelerom- first core Earth explorer. In Proceedings of the 3rd international GOCE user eters for CHAMP, GRACE and GOCE space missions: synergy and evolu- workshop, pages 6–8. tion. Boll. Geof. Teor. Appl, 40(3-4):321–327. Flanagan, E. E. and Hughes, S. A. (1998). Measuring gravitational waves from Touboul, P., Foulon, B., and Willemenot (1999a). Electrostatic space binary black hole coalescences. i. signal to noise for inspiral, merger, and accelerometers for present and future missions. Acta Astronautica, ringdown. Phys. Rev. D, 57:4535–4565. 45(10):605–617. Ghobadi-Far, K., Han, S.-C., Weller, S., Loomis, B. D., Luthcke, S. B., Mayer- Wahr, J., Molenaar, M., and Bryan, F. (1998). Time variability of the Gurr ¨ , T., and Behzadpour, S. (2018). A transfer function between line-of- Earth’s gravity field: Hydrological and oceanic e ects and their possible sight gravity di erence and GRACE intersatellite ranging data and an ap- detection using GRACE. Journal of Geophysical Research: Solid Earth, plication to hydrological surface mass variation. Journal of Geophysical 103(B12):30205–30229. Research: Solid Earth, 123(10):9186–9201. Wainstein, L. A. and Zubakov, V. (1970). Extraction of signals from noise. Ghobadi-Far,K. Han, S.-C., McCullough, C. M. Wiese, D. N., Yuan, D. PrenticeHall, Englewood Cli s, NJ. Landerer, F. W. Sauber, J, Watkins, M. M., (2020) GRACE Follow-On Watkins, M., Wiese, D. N., Yuan, D.-N., Boening, C., and Landerer, F. W. Laser Ranging Interferometer Measurements Uniquely Distinguish Short- (2015). Improved methods for observing Earth’s time variable mass distri- Wavelength Gravitational Perturbations. Geophysical Research Letters, bution with GRACE using spherical cap mascons. Journal of Geophysical 47(16):e2020GL089445. Research: Solid Earth, 120(4):2648–2671. Han, S.-C. (2013). Determination and localized analysis of intersatellite line of Watkins, M., Yuan, D., Kuang, D., Bertiger, W., Kim, M., and Kruizinga, G. sight gravity di erence: Results from the GRAIL primary mission. Journal (2005). GRACE harmonic and mascon solutions at JPL. AGU Fall Meeting, of Geophysical Research: Planets, 118(11):2323–2337. 2005:G22A–04. Hauk, M. and Wiese, D. N., (2020). New Methods for Linking Science Ob- Werth, S., Guntner ¨ , A., Schmidt, R., and Kusche, J. (2009). Evaluation of jectives to Remote Sensing Observations: A Concept Study Using Single- GRACE filter tools from a hydrological perspective. Geophysical Journal and Dual-Pair Satellite Gravimetry Architectures. Earth and Space Science, International, 179(3):1499–1515. 7(3):e2019EA000922. Wol , M. (1969). Direct measurements of the Earth’s gravitational potential Jackson, J. D. (2007). Classical electrodynamics. John Wiley & Sons. using a satellite pair. Journal of Geophysical Research, 74(22):5295–5300. Kaula, W. M. (1966). Tests and combination of satellite determinations Wong, L., Buechler, G., Downs, W., Sjogren, W., Muller, P., and Gottlieb, of the gravity field with gravimetry. Journal of Geophysical Research, P. (1971). A surface-layer representation of the lunar gravitational field. 71(22):5303–5314. Journal of Geophysical Research, 76(26):6220–6236. Kaula, W. M. (2013). Theory of satellite geodesy: applications of satellites to geodesy. Courier Corporation. Konopliv, A. S., Park, R. S., Yuan, D.-N., Asmar, S. W., Watkins, M. M., Williams, J. G., Fahnestock, E., Kruizinga, G., Paik, M., Strekalov, D., et al. (2013). The JPL lunar gravity field to spherical harmonic degree 660 from the GRAIL primary mission. Journal of Geophysical Research: Planets, 118(7):1415–1434. Landerer, F., Flechtner, F., Save, H., Webb, F., Bandikova, T., and Bertiger, WI, et al. (2020). Extending the global mass change data record: GRACE follow-on instrument and science data performance. Geophysical Research Letters, 47(12). McCullough, C., Harvey, N., Save, H., Bandikova, T (2019). Description of Calibrated GRACE-FO Accelerometer Data Products (ACT) GRACE-FO Level-1 Product Version 04 Manual JPL D-103863 Muller ¨ , V. (2017). Design considerations for future geodesy missions and for space laser interferometry Ph.D. Thesis, Gottfried Wilhelm Leibniz Univer- sitat ¨ Hannover, 2017. Numata, K., Kemery, A., and Camp, J. (2004). Thermal-noise limit in the frequency stabilization of lasers with rigid cavities. Physical Review Letters, 93(25):250602. Panet, I., Flury, J., Biancale, R., Gruber, T., Johannessen, J., van den Broeke, M., van Dam, T., Gegout, P., Hughes, C., Ramillien, G., et al. (2013). Earth system mass transport mission (e. motion): a concept for future Earth gravity field measurements from space. Surveys in Geophysics, 34(2):141–163. Pollack, H. N. (1973). Spherical harmonic representation of the gravitational potential of a point mass, a spherical cap, and a spherical rectangle. Journal

Journal

PhysicsarXiv (Cornell University)

Published: Jun 15, 2020

References