Swarm Absolute Scalar Magnetometers first in-orbit results

Swarm Absolute Scalar Magnetometers first in-orbit results

Acta Astronautica 121 (2016) 76–87 Contents lists available at ScienceDirect Acta Astronautica journal homepage: www.elsevier.com/locate/actaastro ...

4MB Sizes 0 Downloads 21 Views

Acta Astronautica 121 (2016) 76–87

Contents lists available at ScienceDirect

Acta Astronautica journal homepage: www.elsevier.com/locate/actaastro

Swarm Absolute Scalar Magnetometers first in-orbit results Isabelle Fratter a,n, Jean-Michel Léger b, François Bertrand b, Thomas Jager b, Gauthier Hulot c, Laura Brocco c, Pierre Vigneron c a

Centre National d'Etudes Spatiales (CNES), 18, avenue Edouard Belin, 31401 Toulouse Cedex 9, France CEA-Leti (French Atomic Energy and Alternative Energies Commission/Electronics and Information Technology Laboratory), Grenoble, France c Institut de Physique du Globe de Paris (IPGP), Sorbonne Paris Cité, Université Paris Diderot, CNRS, Paris, France b

a r t i c l e in f o

abstract

Article history: Received 27 December 2014 Received in revised form 20 October 2015 Accepted 20 December 2015 Available online 30 December 2015

The ESA Swarm mission will provide the best ever survey of the Earth's magnetic field and its temporal evolution. This will be achieved by a constellation of three identical satellites, launched together on the 22nd of November 2013. In order to observe the magnetic field thoroughly, each satellite carries two magnetometers: a Vector Field Magnetometer (VFM) coupled with a star tracker camera, to measure the direction of the magnetic field in space, and an Absolute Scalar Magnetometer (ASM), to measure its intensity. The ASM is the French contribution to the Swarm mission. This new generation instrument was designed by CEA-Leti and developed in close partnership with CNES, with scientific support from IPGP. Its operating principle is based on the atomic spectroscopy of the helium 4 metastable state. It makes use of the Zeeman's effect to transduce the magnetic field into a frequency, the signal being amplified by optical pumping. The primary role of the ASM is to provide absolute measurements of the magnetic field's strength at 1 Hz, for the in-flight calibration of the VFM. As the Swarm magnetic reference, the ASM scalar performance is crucial for the mission's success. Thanks to its innovative design, the ASM offers the best precision, resolution and absolute accuracy ever attained in space, with similar performance all along the orbit. In addition, thanks to an original architecture, the ASM implements on an experimental basis a capacity for providing simultaneously vector measurements at 1 Hz. This new feature makes it the first instrument capable of delivering both scalar and vector measurements simultaneously at the same point. Swarm offers a unique opportunity to validate the ASM vector data in orbit by comparison with the VFM's. Furthermore, the ASM can provide scalar data at a much higher sampling rate, when run in "burst" mode at 250 Hz, with a 100 Hz measurement bandwidth. An analysis of the spectral content of the magnetic field above 1 Hz becomes thus possible. These different ASM new capabilities have been operated on the three Swarm satellites since the beginning of the mission. The calibration and validation activities have been carried out until the end of 2014. In this paper, we will present and discuss the first results from these various operating modes. From the results obtained in-orbit with the ASM vector mode, we will then assess the potential future prospects. & 2015 IAA. Published by Elsevier Ltd. All rights reserved.

Keywords: Magnetometer Geomagnetism Swarm Scalar and vector measurements

n Corresponding author. Tel.: þ 33 5 61 27 44 27; fax: þ 33 5 61 27 42 28. E-mail address: [email protected] (I. Fratter).

http://dx.doi.org/10.1016/j.actaastro.2015.12.025 0094-5765/& 2015 IAA. Published by Elsevier Ltd. All rights reserved.

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

77

Fig. 1. Earth's magnetic field, credits ESA.

1. Introduction This paper first introduces the main goals of the European Space Agency Swarm mission, launched at the end of 2013. The role of the Absolute Scalar Magnetometer, the French contribution to this mission, is next detailed with a description of the instrument’s principle and its innovative capabilities. The results already obtained over the first nine months in orbit are then reported, including the scalar performance and the demonstration that the ASM can operate simultaneously as an absolute scalar magnetometer and an autonomous vector field magnetometer, which is a world first. Work in progress and future prospects are finally presented.

2. The Swarm mission Swarm is one of the Earth Explorer Missions run by the European Space Agency (ESA); its main goal is to study the way the Earth's magnetic field varies in space and time, see Friis-Christensen et al. [1]. It has taken over from the Danish Ørsted mission, launched in 1999 and the German CHAMP mission (2000–2010). For more information about previous magnetic missions, see Acuna [2], Olsen et al. [3], and Hulot et al. [4]. The Earth has a large and complicated magnetic field (Fig. 1), the major part of which is produced by a selfsustaining dynamo operating in the liquid outer core. Magnetic measurements taken at or near the surface of the Earth reflect the superposition of this dynamo field with magnetic fields caused by magnetized rocks in the Earth's crust, electrical currents flowing in the ionosphere, magnetosphere and oceans, as well as in the solid Earth, induced by the time-varying external fields. The challenge for scientists is to separate and model these various sources of the magnetic field by taking advantage of their specific characteristics in space and time [4]. Swarm results will offer new insights into the Earth's system by improving our understanding of the composition and processes in the interior and of the Sun's influence within the Earth's system. The mission consists of a constellation of three identical satellites, launched by a single Rockot launcher from Plesetsk in Russia on November, 22, 2013, for a planned operational lifetime of 4.25 years. The three satellites were injected simultaneously, on a near-polar orbit, at an altitude of 490 km. Several orbital manoeuvres were then executed to reach the operational constellation, which was achieved on the 17th of April 2014. Two satellites (Alpha

Fig. 2. Swarm constellation, © P. CARRIL, 2013.

and Charlie) orbit side-by-side, at an initial altitude of about 462 km, expected to naturally decay to 300 km by the end of the mission. The third satellite (Bravo) maintains an altitude of about 510 km. The orbit inclination difference between the lower pair and the higher satellite gradually produces a relative drift in local time. The higher satellite should then cross the trajectory of the other two at an angle of 60° by the third year of the mission (Fig. 2). Data will then be acquired simultaneously at different altitudes and local times. This will help distinguish the contributions of the different sources of the magnetic field. In order to study the magnetic field thoroughly, each satellite carries two Magnetometers: a Vector Field Magnetometer (VFM) coupled with a star tracker camera (STR) to measure the direction of the magnetic field in space, and an Absolute Scalar Magnetometer (ASM) to measure its strength. In addition, an accelerometer and an electrical field instrument provide information on the state of the ionospheric environment surrounding the satellites. The payload is complemented by a GPS receiver and a laser retroreflector for precise orbit determination. To reduce as much as possible the magnetic disturbances generated by the equipment located within the satellite body, the Absolute Scalar Magnetometers sensors have been installed at the tip of a 4-metre boom (Fig. 3). The optical bench supporting the Vector Field Magnetometer and the three star trackers are mounted halfway along the boom. The ASM electronic boxes (Data Processing Units, DPU) have been installed in the main body of the satellite and are connected to the sensors by a bundle of optical fibres and electrical cables, which form the ASM harness. Particular care was taken all along the development phase to meet the magnetic cleanliness requirements inherent to the mission. For example, the first magnetic tests carried out on ground at satellite level revealed a magnetic perturbation of several nanoTeslas in the vicinity of the ASM. This was found to be induced by thermal gradients in metals (Titanium bracket and connectors caps) when the ASM heaters were activated. This led to

78

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

Fig. 3. Magnetometers (credits EADS).

accommodation

on

Swarm

satellites

Fig. 5. 4He energy diagram and ASM principle of operation.

Fig. 4. Absolute Scalar Magnetometer (Credits: Gérard Cottet).

several replacements of hardware to avoid this thermomagnetic effect. The titanium interface bracket was replaced by a carbon fibre–reinforced polymer bracket and the ASM electrical connectors were taken out. These modifications were successful and the remaining magnetostatic signature related to the ASM heaters activation was found to be negligible.

3. The Absolute Scalar Magnetometer 3.1. ASM objectives The ASM is the French contribution to the Swarm mission (Fig. 4). This new generation instrument, based on atomic spectroscopy of the helium 4, was designed by CEA-Leti in Grenoble, and developed in close partnership with CNES, with scientific support from IPGP. The change of technology for Swarm was decided to overcome the limitations of the Overhauser Nuclear Magnetic Resonance sensors, also developed by CEA-Leti in collaboration with CNES and flown with success on the Ørsted and CHAMP missions, see Léger et al. [5]. As the magnetic reference of the Swarm mission, the ASM provides continuous absolute measurements of the Earth's magnetic field strength, with resolution and accuracy characteristics unequalled for space instruments. In particular its performance is independent of the field modulus and orientation. It thus allows an in-flight calibration of the VFM. Given the crucial role of the ASM, each Swarm satellite carries two identical models, a nominal and a redundant, in cold redundancy (only one is active at a time). In addition to this main mission, the ASM also offers extended capabilities. It nominally provides a set of measurements every second, with anti-aliasing filters limiting

the ASM bandwidth to 0.4 Hz in this mode. However, the intrinsic bandwidth of the ASM is much higher. This allowed the implementation of a burst mode at 250 Hz, corresponding to a 100 Hz bandwidth measurement. This mode is useful to analyze the spectral content of the magnetic field at low frequencies with a very high resolution. In addition, on an experimental basis, the innovative design of the ASM also enables it to take continuous vector field measurements, taking advantage of the combination of its high resolution and extended bandwidth. As a result the ASM is the first space magnetometer capable of taking both scalar and vector measurements simultaneously at the same point. 3.2. ASM physical principle The ASM instrument is an optically pumped helium 4 magnetometer based on an electronic magnetic resonance whose effects are amplified by a laser pumping process. A fraction of the helium atoms are first excited to the 23S1 metastable state by means of a high frequency discharge. In the presence of a magnetic field B0, this level is split into three sub-levels whose energy levels are separated, via the Zeeman effect, by an energy ΔE that is directly proportional to the applied field (Fig. 5). The determination of this separation is a direct method of measurement of B0. The ASM uses conventional magnetic resonance techniques to measure this energy, the signal being amplified thanks to optical pumping. The measurement of B0 is performed by exciting and detecting the paramagnetic resonance between the Zeeman sublevels: a radiofrequency field BRF is applied on the 4He gas cell and when its frequency F matches the Larmor frequency ω0 of the Zeeman sublevels, the magnetic resonance occurs and transitions are induced between the sublevels. Once the magnetometer resonance frequency F is determined, the magnetic field modulus B0 can be directly derived using the electron gyromagnetic ratio γ in the helium 4 metastable 23S1 state: B0 ¼F/γ, with γ/2π E28 GHz/Tesla The ASM can therefore be considered as a magneticfield-to-frequency converter. However, at thermal equilibrium the sublevels are almost equally populated so that no significant change is induced at resonance. In order to detect it, a selective pumping from one of the Zeeman sublevel to the 23P0 state is performed, thanks to a

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

frequency-tuned linearly-polarized laser light. The resulting disequilibrium between the populations of the Zeeman sublevels amplifies the resonance signal amplitude by several orders of magnitude. The monitoring of the intensity of the laser light transmitted through the 4He cell allows its detection. The light source thus realizes two functions simultaneously for the magnetometer: on one hand it performs the optical pumping necessary to more easily detect the resonance, and on the other hand, it allows the detection of the magnetic resonance frequency. This is achieved by an innovative athermal fibre laser (i.e. a laser with a doped optical fibre as gain medium, see Fourcault et al. [6]), included in the ASM electronic unit (DPU). To our knowledge, this is the first time a fibre laser is operated in space. For space applications, it was essential to define an anisotropic architecture of the ASM probe fully independent of its spatial position and orientation, to avoid resonance signal extinction and dead zones. For a linearly polarized pumping beam, the amplitudes of the resonance signals are optimum when both the RF excitation field and the polarization direction are maintained perpendicular to the ambient magnetic field. This was achieved thanks to a special design of the sensor, using a piezoelectric motor to simultaneously rotate the polarizer and the RF excitation coils placed respectively upstream and around the helium 4 cell. This allowed directions of both the RF excitation field and the beam polarization to be controlled. To measure the magnetic field's direction, three orthogonal coils allow the superposition of three low frequency AC modulations on the static field B0, along three orthogonal directions (vector modulations amplitude E50 nT, vector modulations frequenciesE8 Hz, 11 Hz and 13 Hz, adjustable in orbit) (Fig. 6). A real time spectral analysis of the resulting scalar measurement with simple deconvolution operations then simultaneously provides a direct estimate of the projections of the static magnetic field along the three modulation directions, in addition to the static field determination. Vector field is then rebuilt in the ASM orthogonal reference frame thanks to a specific calibration process, see Gravrand et al. [7]. It has been extensively verified on ground, using differential measurements with an absolute reference, that the ASM scalar performance was not affected by these vector modulations. The undeniable advantage of that configuration is that the same instrument simultaneously delivers absolute scalar and vector data, with the same filtering. This means that no synchronization operation between the scalar and vector measurements is necessary and that both filtering and synchronization errors are suppressed. Furthermore, the vector measurements are derived from measurements made by the scalar sensor. As a result, both scalar and vector measurements correspond to the same ambient magnetic field. This avoids having to deal with natural magnetic gradients between the two measurements (which can be quite significant when measured by two separate instruments). Moreover, ASM vector data do not exhibit any offset, which considerably simplifies the calibration process: in principle, only six unknown parameters have then to be determined (the 3 transfer functions of the

79

vector coils and their deviations from orthogonality, defined by 3 angles). Last but not least, the ASM also offers auto-calibration capabilities. The calculation of a scalar residual, defined as the difference between the field intensity measured by the ASM and the norm of the three components of the field after calibration, provides an easy means for real time data quality assessment. Should the ASM be affected by an external perturbation, both scalar and vector measurements would see it the same way, so that no preliminary correction for spacecraft’s stray fields is necessary for the calibration process which can therefore be carried out directly on the raw data. While the main advantage of this instrument lies in its lack of intrinsic offsets, its vector precision is by design significantly lower than the scalar one and is inversely proportional to the static field amplitude B0 (Rscalar E1 pT/ √Hz independent of B0 whereas Rvector E1 nT/√Hz for B0 ¼ 25 mT).

4. First in-orbit results 4.1. ASM Instruments early health check after first switch on The ASM instruments have been switched on four days after launch and tested in their various operating modes. The good health of the instruments was first assessed, taking advantage of the different diagnosis tools available on board. Then all measurement modes were tested: the scalar mode (scalar measurements only at 1 Hz), the burst mode (scalar measurements only at 250 Hz) and finally the vector mode (simultaneous scalar and vector measurements at 1 Hz). All functional verifications were successful on the 3 satellites. A quick look at the magnetometers' intrinsic noise confirmed their excellent resolution, around 1 pT/√Hz over the entire [DC-100 Hz] operating range, in accordance with ground measurements. The same verifications were next carried out on the three redundant ASMs. Similar successful results were obtained on Swarm Alpha and Bravo. The redundant ASM on Swarm Charlie, however, displayed a major malfunction, with no resonance signal. This instrument had been tested on ground during the launch campaign at Plesetsk and proven to be fully operational. Unfortunately, it thus appeared that this ASM had not survived the launch. It is now considered definitively lost. This was fortunately detected before the beginning of the orbit change manoeuvres. Based on this early failure and on additional considerations related to other instruments, and as the mission's lifetime of the satellite to be placed on the upper orbit was then already known to likely be extended, it was then decided by ESA to assign this orbit to the Bravo satellite with both ASM instruments in good health. The two Charlie and Alpha satellites were consequently moved to the lower orbit. After these early functional verifications, the detailed assessment of the ASM performance started on each satellite, but only on the nominal ASM instruments.1

80

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

Fig. 6. Vector mode principle.

Fig.7. Scalar field noise analysis in burst mode.

4.2. Learnings from the ASM burst mode This mode was operated early on to allow a spectral analysis of the ASM signal in the bandwidth [DC-100 Hz] and provide a direct assessment of the instruments' noise floor. This noise was confirmed to be of order 1 pT/√Hz for the three satellites, confirming tests previously carried out on ground (Fig. 7). The low frequency part of this spectrum 1 After this paper (related to the first nine months in space) was first submitted, an ASM failure unfortunately occured on Swarm Charlie on the 5th of November 2014, resulting in the loss of the nominal ASM model. A commission of enquiry concluded that the most likely cause of this was a failure of a micro-circuit damaged by a heavy ion in the ASM DC/DC power converter. Since the redundant unit was inoperable since launch (for another reason), ASM data are since then no longer available on Charlie. ASM instruments on Alpha and Bravo still flawlessly provide data, identical in quality to the data presented and discussed in this paper. Both also still have their redundant units. Despite the loss of the ASM on Charlie, calibration of the VFM instrument on this satellite (to produce L1b data) can still be achieved in an indirect way, using ASM information from the neighbouring Alpha satellite, with only a modest penalty in terms of data quality.

reveals the ambient Earth magnetic field spectrum, with contributions from both the natural temporal field fluctuations and satellite motion related effects (both expected signals of geophysical origin). This mode could also be used to gain a very thorough insight into the instrument's operation: for instance, it was found the ASM sensor's piezoelectric motor activations generate perturbations of up to 4 nT, that could only be detected in flight using such burst mode data. These perturbations were not expected and were not observed on ground because it was not possible to operate the burst mode on ground in a simulated changing magnetic field, due to the ground facilities limitations (noise level of the field simulator). A post processing of the burst raw data was consequently implemented to detect and suppress these peaks. Fortunately, further analysis confirmed by additional dedicated ground tests revealed that the impact of such peaks on the ASM 1 Hz scalar measurements (low-pass filtered at 0.4 Hz) was in fact negligible (around 10 pT). However, the same did not hold for the ASM 1 Hz vector

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

81

Fig. 8. Spectrogram of the scalar measurement in burst mode.

Fig. 9. Example of crosstalk with the heater's harmonics, in burst mode.

measurements, and the vector data affected by this occasional effect had to be flagged as outliers. A noise spectrogram of the corrected burst data is given in Fig. 8. On this spectrogram, a clear 30.1 Hz frequency signature can be seen, which corresponds to an expected internal ASM RF frequency modulation aliasing. Another 3.05 Hz frequency signature can also be seen. This harmonic corresponds to expected remaining signatures of the ASM internal laser D0 wavelength control loops. As previously explained, the lower frequencies otherwise include the ambient Earth magnetic field spectrum of scientific interest. The analysis of the burst mode data also revealed unexpected signals of up to 1 nT (Fig. 9), regularly distributed along the orbits, observed at the same geographical location by the three ASM (Fig. 10). An in depth analysis led to the identification of a crosstalk with the ASM heater’s harmonics as the root cause. Indeed, heaters use a square AC signal at 58 kHz, which impacts the

measurements when its harmonics match the ASM's Larmor frequency. The impact of such perturbations on the ASM scalar measurements at 1 Hz (filtered at 0.4 Hz) has been assessed and was found to be negligible. Occasional perturbations are however visible in the ASM 1 Hz vector measurements and these had to be flagged as outliers. The burst mode thus appeared to be a powerful tool to investigate unexpected perturbations that could not be detected at nominal 1 Hz sampling frequency. The ASM's burst mode was further operated for a total of seven days in January and February 2014, in different magnetic (Kp) conditions, to check for the presence of potential ‘high frequency’ geophysical signals and assess the interest of this mode for scientific exploitation (currently under way). Note, however, that the ASM burst data is not processed to level 1b by ESA's Swarm ground segment, and therefore not yet distributed.

82

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

Fig. 10. Geographical distribution of the perturbations due to the heater's harmonics (satellite Alpha on January 7, 2014; green dots mark locations where conditions for cross-talk with the heater to possibly produce perturbations are met, red cross mark locations where perturbations have indeed been detected). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 11. Comparisons between the scalar measurements from the different satellites on 21/12/2013.

4.3. Validation of the nominal scalar data (1 Hz sampling rate) During the early commissioning phase, all ASM instruments were also run in their scalar mode over five orbits to verify that this mode, only providing scalar measurements at 1 Hz sampling rate, was operational. The instruments were next switched to vector mode on all three satellites and have since been providing simultaneous scalar and vector measurements at 1 Hz sampling rate (expect for a few days dedicated to specific operations, such as additional burst mode sessions). Given the ASM's operation principle, scalar measurements are absolute by design and do not require any calibration. The ASM raw scalar data are processed up to level 1b thanks to an algorithm defined by CEA-Leti, which includes corrections for

the residual sensor’s remanent and induced effects and systematic errors. These had been properly determined on ground, thanks to differential measurements with NMR reference sensors. Maximum error after level 1b correction is around one millionth of the magnetic field's magnitude to be measured (65 pT worst case, depending on instruments). This represents the ASM accuracy at instrument level. The in-flight ASM's performance assessment relied on the comparison of the noise floor measured in orbit with the one that had been measured on ground. As already presented in the previous section, this noise floor was found to be compliant with ground measurements. The following characteristics were verified in orbit: 1. Measurement range: 4[18.3 μT–52.6 μT], no dead zones, 2. Scalar bandwidth (BW)/sampling rate (Fs):

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

 [0–0.4 Hz]/1 Hz, in scalar or vector mode,  [0–100 Hz]/250 Hz, in burst mode, 3. Scalar resolution/precision (relies on the scalar resolution analysis performed in burst mode)

 Resolution: Ro1.4 pT/√Hz [DC-100 Hz], independent of the field modulus,

 Precision: σ o1 pT (Fs¼1 Hz, BW ¼0.4 Hz, leading to σ ¼R*√BWo1 pT). Compared to the characteristics of the Overhauser (OVH) scalar magnetometers flown on the Ørsted and CHAMP satellites, the ASMs offer an improvement in resolution by a factor of 4–50 depending on the value of the ambient field (since the resolution of the OVH instruments was strongly dependent on the magnetic field modulus). All other ASM characteristics were checked on ground before launch and were considered fulfilled in orbit, given the similar ground and in-orbit behaviour of the ASM instruments: 1. Scalar accuracy: maximum accuracy error after level 1b correction: σ o65 pT (worst case), which represents an improvement by a factor 5 when compared to the Overhauser magnetometers flown on Ørsted and CHAMP, 2. Stability: demonstrated to be better than 25 pT for 15 days at the Chambon-La-Forêt observatory (IPGP). To further assess the quality of the scalar measurements and test their accuracy, it was also decided to take advantage of the constellation aspect of the mission by carrying out comparisons between scalar measurements from the different satellites. This was done early on, on 21/12/2013 (local times of descending/ascending nodes were 0 h/12 h). This day was selected because it was magnetically quiet, with no satellite manoeuvre, and because of the availability of midnight data with minimum ionospheric signal. In addition, the three satellites were then still very close to each other, Alpha being located between Charlie and Bravo, with a maximum longitudinal distance of 20 km between the orbits of Alpha and Bravo, and of 32 km between those of Alpha and Charlie. The following method was used: 1. For each Alpha position, selection of the nearest Bravo and Charlie positions, for spatial co-localization, 2. For each selected scalar data, computation of a scalar residual, by subtraction of the intensity predicted by a main field model (accounting for secular variation but ignoring the crustal field) using the then available MSPMAG-1 model of Alken et al. [8] 3. For each pair of, e.g., Alpha and Bravo scalar data residuals, computation of the difference, next plotted as a function of their (practically) identical Quasi-Dipole (QD) latitude (a latitude defined with respect to the magnetic rather than the geographic equator, see Richmond [9]), distinguishing night side data (in blue) from day side data (in red).

83

Results are plotted in Fig. 11 (Bravo–Alpha: upper plot, Alpha–Charlie: middle plot, Bravo–Charlie: bottom plot). As can be seen, residuals between the different ASM readings were found to be very low, despite the fact that they could still include geophysical signals, especially near the poles and the magnetic equator. No clear day/night bias effect was found. Further selecting data over the Pacific ocean (where the crustal field is weak), with QD latitude less than 60° (thus excluding high latitudes, where the natural magnetic field variations are highest), and only at the midnight local time (to further minimize ionospheric contributions), led to the lowest residuals, with sigmas of less than 0.3 nT when comparing Alpha to either Bravo or Charlie. This result is to be compared with the Swarm specification for the accuracy of the magnetic field strength measurement, which is of 0.3 nT (2 σ, σ being the standard deviation). These ASM scalar measurements could also straightforwardly be used to produce maps of the magnetic field strength by stacking the scalar data acquired at the same location by the three satellites, whenever possible. Fig. 12 below presents the most recent maps one could build in this way for August 2014. These maps were derived from Alpha and Charlie ASM level 1B data acquired between 01/ 08/2014 and 29/08/2014, at an altitude between 460 and 480 km above the Earth's mean radius of 6371 km (since Bravo is now orbiting at a higher altitude, its data were not used here). These maps clearly show a particularly important manifestation of the dynamics of the Earth’s magnetic field, which is the very marked anomaly, known as the “South Atlantic Anomaly”, where the field is particularly weak. This anomaly changes in time. Previous space missions had shown that it has expanded over the past 30 years and that the minimum magnetic field strength dropped by roughly 8% of its value in that region during that period of time (see e.g., Hulot et al. [4]). This has a significant impact on satellites in Low Earth Orbits, as these suffer much more from radiation when they cross this zone. ASM scalar data, processed up to level 1b, are available since mid-May 2014 as one of the products released by the European Space Agency (see https://earth.esa.int/web/ guest/swarm/data-access). 4.4. Calibration and validation of the ASM vector data (1 Hz sampling rate) 4.4.1. Calibration of the ASM vector data Vector data from the ASM need to be calibrated, in order to determine the transfer functions of the vector coils (3 parameters) and their deviations from orthogonality (3 angles), also taking into account the operational conditions which could impact them. This was achieved using a least-squares algorithm, which minimizes the scalar residuals, defined as the difference between the absolute scalar measurement of the magnetic field and the modulus derived from its three components. To improve the vector reconstruction, the following parameters also had to be taken into account:

84

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

– the vector coils thermal expansion coefficient (one parameter, fixed), which impacts the coils transfer functions, – the sensor’s rotor angular position (9 parameters), to account for residual laser polarization and fibre coupling effects as well as minor sensor alignment imperfections. Such calibration can be performed on the ASM raw measurements (in ASM time and reference frame), and is therefore not sensitive to any potential spacecraft's stray field corrections or level 1b processing errors. It was done on a daily basis, using a sliding window including three days of 1 Hz vector data (no down-sampling). The analysis of the scalar residuals provided a real time quality assessment of the vector data, whose results are presented below.

Fig. 12. Maps of the strength of the Earth's magnetic field at Swarm Alpha and Charlie altitude, August 2014.

A short-term analysis is first shown on one day of Swarm Alpha ASM vector data (Fig. 13). The blue plot shows the raw residuals (bandwidth 0.4 Hz), while the red plot shows low-pass filtered scalar residuals (with a cut off frequency set to 5 mHz, to remove most of the ASM vector noise and focus on the low frequency orbital evolutions, which are the signals of greatest interest for main field modelling purposes). Unfortunately, the ASM vector data appear to be significantly noisier on Charlie than on Alpha and Bravo, with scalar residuals observed to be about twice larger. An investigation showed that this noise is in fact related to the vector modulations. As can otherwise be seen, the ASM scalar residuals do not show any obvious day/night effect. Nevertheless, remaining orbital low frequencies signals are still under investigation. A statistical analysis of the ASM scalar residuals showed that the expected global Gaussian behaviour of their distribution was in fact distorted by outliers. These were investigated and partly explained, thanks to the results of the burst mode analysis (see above, Section 4.2). Part of the outliers appears to be due to the sensor's piezoelectric motor activations, while additional outliers are clearly due to the already mentioned crosstalk of the ASM with the heater's operating frequency. A few outliers related to the magnetic field derivative (dB0/dt, dHx/dt, dHy/dt, dHz/dt) and attributed to inherent limitations in the ASM vector signals dynamics were also detected. All these outliers were identified and flagged to allow selection for scientific exploitation of the ASM vector data. Thanks to the daily calibration, a fairly stable performance was obtained throughout the 9-month period investigated in this study. The corresponding results are presented in Fig. 14 for Swarm Alpha.

Fig. 13. ASM scalar residuals (raw and 5 mHz low pass filtered), 01/09/2014 – Swarm Alpha.

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

85

Fig. 14. ASM scalar residuals obtained with a daily calibration over a 9-month period.

Fig. 15. Evolution of the ASM vector coils functions over a 9-month period, with a daily calibration.

As can be seen, a few outliers remained, but these could be correlated with easily identified satellite specific operations, such as manoeuvres. An additional analysis was also carried out using a static set of calibration parameters throughout the same time period, to assess the long-term stability of the instrument when relying on such a single calibration. This revealed a substantial seasonal variation. While this effect has no impact on the production of high quality vector data, since the calibration process can always be run on a

daily basis, as done so far, it provides extra very powerful insight into the instruments fine operation. The origin of this variation has been traced back to the slow evolution of the vector coils transfer functions (Fig. 15) and its root cause is still under investigation. A likely candidate is the effect of sun illumination on the sensors mounted at the tip of the boom. Ultimately, once our hypothesis confirmed, it is our intention to take this phenomenon into account and update the overall vector calibration process of the instruments.

86

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

Fig. 16. Evolution of the relative orientation of ASM and VFM reference frames (satellite Alpha).

4.4.2. Vector performance summary The ASM vector mode performance having been assessed in-flight, a summary of the results is given below. Please keep in mind that vector performance is inversely proportional to the field modulus: 1. 2. 3. 4.

Measurement range: 752.6 μT Sampling rate: 1 Hz Bandwidth: [DC - 0.4 Hz] Scalar residuals: on Swarm Alpha and Bravo (with current daily calibration):

 σ o2.7 nT (based on unfiltered raw measurements),  σ o2 nT (after outlier suppression) 5. Vector resolution E2 nT/√Hz for B0 ¼ 50 mT Recall that vector calibration optimization is still under way. 4.4.3. Validation of the ASM vector data Using the calibration parameters as defined above, the ASM vector data were processed up to level 1b, thanks to algorithms jointly defined by CEA-Leti and IPGP. This consisted in converting the raw values into physical values, taking into account the spacecraft’s stray field corrections (spacecraft induced and remanent models, stray field electrically induced by magneto-torquers, thrusters, batteries, solar panels, …). The ASM vector data, in the ASM time and reference frame, were then synchronized with UTC time and expressed in both the ASM instruments and NEC (North East Center) reference frames. It is important to realize that since the ASMs are not mounted on the optical benches of the satellites, any relative movement between the ASM and the star tracker will result in additional errors, not directly related to the instrument’s performance. It was therefore also necessary to assess the stability of the alignment between the optical bench and the ASM sensor prior to any exploitation of the ASM vector data, either for the determination of a

geomagnetic field model or for intercalibration with the VFM. This could be achieved by computing a rotation matrix between the ASM and VFM reference frames and next comparing the evolution of the vector data delivered by both instruments. Although such an estimation will obviously also reflect errors affecting the two instruments, it made it possible to evaluate a worst-case boom stiffness. Fig. 16 shows the result of this procedure applied to the Alpha satellite, and reveals slow long-term deformations with a maximum amplitude of 50 arcsec. Given these very encouraging results, it was decided that IPGP, in collaboration with DTU, would derive geomagnetic field models tentatively relying on ASM data only. These attempts were met with substantial success. In particular, a candidate model could be submitted on time for the International Geomagnetic Reference Field (IGRF 2015-2020). This model, see Vigneron et al. [10], got approved and is one of the models that went into the building of the official IGRF model recently released (see Thébault et al. [11] and http://www.ngdc.noaa.gov/IAGA/ vmod/). Most recently, an even more advanced model could also be built, using eleven months of data, demonstrating the possibility of using the vector data delivered by the ASM to also derive models of the lithospheric field, see Hulot et al. [12]. This already was a very significant achievement for the ASM vector mode. Finally, it should be recalled that the ASM's vector mode has been operated only as an experimental mode so far, the nominal vector data being provided by the VFM (after calibration, but only using the ASM scalar data). Therefore, ASM vector data are currently not yet freely distributed by ESA. Nevertheless, the ASM instruments having been continuously operated in vector mode on Swarm Alpha and Bravo since the beginning of the mission, with a vector performance fulfilling all expectations, the ASM vector data may eventually become an official product of the mission.

I. Fratter et al. / Acta Astronautica 121 (2016) 76–87

5. Summary and perspectives The ASM instruments have been operated on the three Swarm satellites since the 26th of November, 2013, a few days after launch. First results confirm that our initial goals were fully met. This was achieved thanks to a close partnership between CNES, CEA-Leti and IPGP and thanks to ESA who agreed and helped to accommodate this innovative instrument on board Swarm. During their first nine months in space, all ASMs have continuously been providing simultaneous scalar and vector data at a 1 Hz sampling rate, except during a few days when specific operations were carried out. The ASMs burst mode was shown to be a powerful tool to assess the instruments' performance and good health during the commissioning phase. It could be used to demonstrate the very low noise level of the magnetometers and check the cleanliness of the electromagnetic environment required for the ASMs to properly operate. Further analysis of the signals collected by this burst mode during the commissioning phase is currently being carried out by IPGP to look for potential geophysical signals of interest. The ASM scalar data was shown to offer the best performance ever attained in space. Consistency of the readings delivered by the different instruments has also been demonstrated. This should allow scientists to study very tiny signals, not detectable before Swarm. Moreover, and despite the challenge it represented, a direct in-orbit verification of the Swarm mission's requirement for the accuracy of the magnetic field's magnitude could be carried out, taking full advantage of the constellation configuration of the mission. The vector mode experiment has also been shown to be very successful. The ASM ability to function simultaneously as an absolute scalar magnetometer and an autonomous vector field magnetometer has now been validated in orbit. This is a world first. This capability also provided a unique opportunity to cross-calibrate two different types of vector instruments directly in orbit (the Vector Fluxgate Magnetometer and the ASM in vector mode). The vector mode data have also been demonstrated to have their own scientific merit. In particular, ASM data were successfully used to derive geomagnetic field models, without having to resort to any VFM data. These vector mode data may become an official product of the Swarm mission at some point. The new “two in one” instrumental concept demonstrated by the ASM should provide interesting possibilities for future missions based on space magnetometry. Coupled with a star tracker, it could lead to a significant simplification of the payload, thus reducing the mass of the spacecraft and the cost of the mission. Concerning the ASM vector mode itself, several improvements motivated by the first inflight data analysis reported here are already under study by CEA-Leti. New technologies to replace the piezoelectric motor and simplify the laser are being evaluated in partnership with CNES. CEA-Leti is also working on the miniaturization of the helium magnetometer and on minor evolutions of the sensor overall design to allow its exploitation at very low fields, making it suitable for future missions in planetary exploration, see Rutkowski et al. [13].

87

For more information, visit the CNES website dedicated to the Swarm mission (https://swarm.cnes.fr/). Acknowledgement The authors acknowledge the ESA Swarm team for computing and providing additional data access on Swarm satellites orbital characteristics for the long-term analysis of the ASM instrument behaviour. GH, LB and PV also gratefully acknowledge support from the Centre National d’Etudes Spatiales (CNES) within the context of the ‘Travaux préparatoires et exploitation de la mission SWARM’ project. This is IPGP contribution no. 3705.

References [1] E. Friis-Christensen, H. Lühr, G. Hulot, Swarm: a constellation to study the Earth's magnetic field, Earth Planets Space 58 (2006) 351–358. [2] M.H. Acuna, Space-based magnetometers, Rev. Sci. Instrum. 73 (2002) 3717–3736. [3] N. Olsen, G. Hulot, T.J. Sabaka, Measuring the Earth's magnetic field from space: concepts of past, present and future missions, Space Sci. Rev. 155 (2010) 65–93. [4] G. Hulot, T.J. Sabaka, N. Olsen, A. Fournier, The Present and Future Geomagnetic Field, in: Gerald Schubert (Ed.), 2nd edition, Treatise on geophysics, 5, Elsevier, Oxford, 2015, pp. 33–78. [5] J.-M. Léger, F. Bertrand, T. Jager, I. Fratter, Spaceborn scalar magnetometers for Earth's field studies, in: Proceedings of the 62nd International Astronautical Congress, Cape Town, South Africa, Oct. 3–7, 2011, paper: IAC-11-B1.3.9. [6] W. Fourcault, J.-M. Léger, V. Costes, I. Fratter and, L. Mondin, Athermal fiber laser for the Swarm Absolute Scalar Magnetometer, in: Proceedings of the 2010 International Conference on Space Optics, Rhodes, Greece, Oct. 4–8, 2010. [7] O. Gravrand, A. Khokhlov, J.-L. Le Mouël, J.-M. Léger, On the calibration of a vectorial 4He pumped magnetometer, Earth Planets Space 53 (2001) 949–958. [8] P. Alken, S. Maus, H. Lühr, R.J. Redmon, F. Rich, B. Bowman, S.M. O’Malley, Geo-magnetic main field modeling with DMSP, J. Geophys. Res. Space Phys. 119 (2014) 4010–4025, http://dx.doi.org/ 10.1002/2013JA019754. [9] A.D. Richmond, Ionospheric electrodynamics using magnetic Apex coordinates, J. Geomagn. Geoelectr. 47 (1995) 191–212. [10] P. Vigneron, G. Hulot, N. Olsen, J.-M. Léger, T. Jager, L. Brocco, O. Sirol, P. Coïsson, X. Lalanne, A. Chulliat, F. Bertrand, A. Boness, I. Fratter, A 2015 international geomagnetic reference (IGRF) candidate model based on Swarm’s experimental Absolute Scalar Magnetometer vector mode data, Earth Planets Space 67 (2015) 95, http://dx.doi. org/10.1186/s40623-015-0265-4. [11] E. Thébault, C.C. Finlay, C.D. Beggan, P. Alken, J. Aubert, O. Barrois, F. Bertrand, T. Bondar, A. Boness, L. Brocco, E. Canet, A. Chambodut, A. Chulliat, P. Coïsson, F. Civet, A. Du, A. Fournier, I. Fratter, N. Gillet, B. Hamilton, M. Hamoudi, G. Hulot, T. Jager, M. Korte, W. Kuang, X. Lalanne, B. Langlais, J.-M. Léger, V. Lesur, F.J. Lowes, S. Macmillan, M. Mandea, C. Manoj, S. Maus, N. Olsen, V. Petrov, V. Ridley, M. Rother, T.J. Sabaka, D. Saturnino, R. Schachtschneider, O. Sirol, A. Tangborn, A. Thomson, L. Tøffner-Clausen, P. Vigneron, I. Wardinski, T. Zvereva, International geomagnetic reference field: the twelfth generation, Earth Planets Space 67 (79) (2015) http://dx. doi.org/10.1186/s40623-015-0228-9. [12] G. Hulot, P. Vigneron, J.-M. Léger, I. Fratter, N. Olsen, T. Jager, F. Bertrand, L. Brocco, O. Sirol, X. Lalanne, A. Boness, V. Cattin, Swarm's absolute magnetometer experimental vector mode, an innovative capability for space magnetometry, Geophys. Res. Lett. (2015) 42, http://dx.doi.org/10.1002/2014GL062700. [13] J. Rutkowski, W. Fourcault, F. Bertrand, U. Rossini, S. Gétin, S. Le Calvez, T. Jager, E. Herth, C. Goreki, M. Le Prado, J.-M. Léger, S. Moralès, Towards a miniature atomic scalar magnetometer using a liquid crystal polarization rotator, Sens. Actuators A 216 (2014) (2014) 386–393.