Swarm Instruments Overview
Figure 1: Swarm side annotation
Absolute Scalar Magnetometer
Figure 2: ASM sensor
ASM Algorithm Overview
The overall processing is sketched in the Figure below. First, the raw output (Level 1a.ASM. Sci.E, rate: 1 Hz, timestamps t0,ASM) from the ASM is converted to physical units (nT), corrected for the Bloch-Siegert effect, and corrected for delays in timestamps (from t0,ASM to tASM). This constitutes the Level 1bInst.ASM data, the ASM instrument Level 1b scalar product containing the perturbated magnetic field intensity measurements at corrected time instants, tASM. Note: this includes stray fields from the ASM itself.
Then outlier detection and/or rejection is performed. Followed by corrections for magnetic disturbances (stray fields) of the ASM itself, of the VFM, and of the rest of the spacecraft. These disturbances need to be filtered according to the intrinsic filter of the ASM instrument (see [AD-7] for a detailed description of ASM processing). This constitutes the fully calibrated and corrected scalar magnetic measurements, FASM, at instrument time instants (tASM). These data are then adjusted for the group delay of the ASM intrinsic filter, i.e. the measurements are shifted in time from tASM to tout,ASM.
Finally, these data are interpolated to yield the scalar elements of the Level 1b.Mag-L product at UTC seconds. The interpolation process also bridges small gaps (a few samples) in the ASM data stream typically caused by missing telemetry (missing ISP) or outlier rejection.
Figure 3: ASM processing overview
Vector Field Magnetometer
Figure 4: Swarm Optical Bench
The Vector Field Magnetometer (VFM) measures the magnetic field vector at the tip of the optical bench on the boom. The sensor is a 3-axis Compact Spherical Coil (CSC) with a 3-axis Compact Detector Coil (CDC) inside. The instrument operates as a closed-loop system adjusting the compensating CSC currents to maintain a null field at the detector coils within the sphere. The currents in the CSC coils are measured and digitized (by an ADC) and this constitutes the raw measurements of the instrument. See [AD-8] for a detailed description of the instrument, and for the Level 1b algorithm described by the instrument supplier.
The VFM is an analogue instrument and as such subject to temporal changes due to radiation and aging effects of the electronics. The effects are only significant in the bias and linear scale parameters of the characterization; hence these parameters are estimated daily through a comparison of the VFM output with the ASM scalar magnetometer output. Additionally, the non-orthogonality angles of the VFM sensor may also be estimated in this process. The allowed change from day to day in the parameters is controlled group-wise by weight parameters specified in the CCDB. The parameters estimated daily are stored in a Temporal Calibration File (TCF) as an auxiliary data product.
VFM Algorithm Overview
The overall VFM processing is sketched in the Figure below. Only data with a common DPU_id is to be processed, minority data (DPU_id wise) is rejected. Next, the raw output (Level 1a.VFM.Sci.EU, rate: 50 Hz, timestamps t0,VFM) of the VFM is corrected for timestamp, processing, and filter delays. Then, it is corrected and converted to physical units (nT) using the CCDB.VFM.Cal parameters and the Level 1a.VFM.Sci.TX temperatures. This is the Level 1bInst VFM vector product. The rate is 50 Hz and the time-instants tVFM.
Then, the preliminary vector field measurements, B'VFM, are computed using the TCF.VFMinit parameters. TCF.VFMinit is the most recent (w.r.t. data being processed) TCF.VFM record among the TCF.VFMinput (from the previous Level 1b Product) and the CCDB.L1BP. VFM.TCF_Aux array - with DPU_id corresponding to Level 1a.VFM.Sci.DPU_id. Formally, let t0 = Level 1a.VFM.Sci.t0, then:
Next, outliers are detected and accordingly rejected or flagged as suspicious samples. Now, the magnetic stray fields from the rest of the S/C at the VFM sensor position - at the time instants of the VFM measurements, tout,VFM (= tVFM) - are computed. The phase linearity and fast response of the VFM Bessel filter makes it unnecessary to apply a filter to the stray fields as was the case for the ASM. However, the characterization of the AOCS torquer coil disturbances possibly needs to take the VFM filter into account.
Next, the internal temporal changes of the VFM electronics and possibly any change in the non-orthogonalities of the CSC are modeled by the TCF.VFM parameters which are estimated by comparison of VFM data with the fully corrected ASM scalar data (FASM). The new estimates of the TCF.VFM parameters, TCF.VFMoutput, are applied to all the VFM data of the actual day. Together with the correction for stray magnetic fields the Level 1b.Mag-H.BVFM data are obtained. I.e. fully converted and corrected magnetic vector data in the orthogonal VFM sensor frame. The rate is 50 Hz and the time instants tout,VFM.
Then Level 1b.Mag-L.BVFM, the 1 Hz magnetic vector product in the VFM frame at UTC seconds, is extracted. Finally, the magnetic field vectors BVFM at 50 Hz and 1 Hz are transformed via the Common Reference Frame (CRF) into the NEC frame. This completes the generation of the Level 1b.Mag-H.BNEC and Level 1b.Mag-L.BNEC products.
Note: the rotation from the VFM to the CRF frame is estimated pre-flight and refined inflight. Hence periodic updates of CCDB.Structure.STR_q_VFM are foreseen and consequently reprocessing of the final step above is required regularly.
Figure 5: VFM processing overview
The Star Tracker (STR) is comprised of three Camera Head Units (CHUs) mounted on the innermost end of the optical bench. Nominally, the attitudes of all three heads are provided simultaneously at 1 Hz rate, however one head is regularly blinded by the Sun leaving the attitudes of just two heads. The attitudes of the 2-3 CHU are combined into one attitude, the attitude of the STR Common Reference Frame (CRF). The combination uses the method described in [RD-2]. The attitudes of CRF are then interpolated to obtain the CRF and S/C attitudes at required time instants: 50 Hz VFM measurements, 2 Hz EFI measurements and UTC seconds.
STR Algorithm Overview
First, the time-stamps are corrected for any delays: tout,STR = t0,STR - CCDB.STR.Delay. Then, the aberrational correction is verified and if necessary applied. The valid attitudes of the STR are combined into common attitude solutions, qCRF←ICRF, providing the CRF → ICRF transformations. This is the Level 1bInst.STR product. Then these are interpolated to the time instants to be used (UTC seconds, tout,VFM, etc.) using a smoothing, cubic B-spline method. Finally, the various required output transformations are generated by combining series of transformations.
Electric Field Instruments
Figure 6: Electric Field Instruments
The Electric Field Instrument determines the ion density, the ion drift velocity, and the electric field at the S/C front panel (in-flight). The instrument consists of two components: the Langmuir Probe (LP) and the Thermal Ion Imager (TII).
EFI Algorithm Overview
The detailed descriptions of the algorithms are formed in two documents:
A) The Langmuir Probe (LP) algorithms can be found in [AD-9],
B) The Thermal Ion Imager algorithms can be found in [AD-10].
The S/C ephemeris and magnetic field vector data needed by the EFI algorithms are taken from various sources:
Through the GPS antenna, the GPS receiver (GPSR) receives the signals from all of the antenna visible GPS satellites. The L1b processing corrects for known effects related to the Swarm instruments and satellite. The external errors e.g. due to the GPS segment are corrected in the orbit determination processing.
Eleven different input packet definitions from the ISP exists, these are
MDH, Measurement Data Header format
However, for the GPS algorithms processing only seven of them are used. The collection of the data record from the CAP, CAA, COP, and GNE has to be repeated N times for the N viewed satellites with a 0.1 Hz update frequency.
GPSR Algorithm Overview
The processing flow of algorithms is separated into three major steps:
The overall data flow from L0 to L1b is shown in the figure below. An overview of the sequential walk-through of the data flow can be found in [AD-7].
Figure 7: Level 0 to Level 1b processing flow for GPSR
The Level 1b instrument processing of the Accelerometer (ACC) is described in [AD-7] and briefly repeated here.
The intrinsic ACC processing is described in [AD-11], cf. Appendix L in [AD-7]. The Level 1a data comprising the acceleration vectors (linear and rotational) in engineering units (eu) and the required house-keeping information (e.g. temperatures and polarization voltage) in physical units. The Level 1a data shall (optionally, i.e. user selectable) be stored as Level 1a Products.
The processing of Level 1a data to Level 1b ACC data is sketched in the Figure below.
Figure 8: ACC processing overview
It consists of the following tasks: