Swarm Langmuir Probes’ data quality and future improvements

Swarm is ESA’s first Earth observation constellation mission, which was launched in 2013 to study the geomagnetic field and its temporal evolution. Two Langmuir Probes on board of each of the three Swarm satellites provide very accurate measurements of plasma parameters, which contribute to the the study of the ionospheric plasma dynamics. To maintain a high data quality for scientific and operational applications, the Swarm products are continuously monitored and validated via science-oriented diagnostics. This paper presents an overview of the data quality of the Swarm Langmuir Probes’ measure5 ments. The data quality is assessed by analysing short and long data segments, where the latter are selected sufficiently long to consider the impact of the solar activity. Langmuir Probes data have been validated through comparison with numerical models, other satellite missions, and ground observations. Based on the outcomes from quality control and validation activities conduced by ESA, as well as scientific analysis and feedback provided by the user community, the Swarm products are regularly upgraded. In this paper we discuss the data quality improvements introduced with the latest baseline, and how the data 10 quality is influenced by the solar cycle. The main anomaly affecting the LP measurements is described, as well as possible improvements to be implemented in future baselines.


Probe Gain
Surface Position 1 high up to Dec 2019-low onward TiN -y 2 low up to Dec 2019-high onward Au + y Table 1. List of differences between the two LPs on each Swarm satellite. The probe position is defined with respect to the spacecraft coordinate system where x is along the fly direction, y horizontally crosses the satellite toward local dusk, and z points toward the Earth.
Au probes even after baking with temperatures of up to 300 • C and after exposure to ultra-sound. Presently, it is unknown how much Au is left on the Ti surface after more than 7 years in space.
3 LP data processing 90 The L1B PLASMA processor, which is used to generate the LP data products, is organised according to simple flow-chart reported in Figure 2. It uses as inputs the L1B products containing position and velocity of the satellite, auxiliary data, and EFI-LP Level 0 (L0) data, to obtain three L1B and one Level 1A (L1A) data products. The auxiliary data contains information that support the Swarm data processing, such as geomagnetic indices, or instrumental calibration parameters obtained during ground tests. The L0 data contains raw measurements from each Swarm instrument and are essentials to generate the L1 95 products. The EFI-LP L1A product (EFIX_LP_1A) contains information about the LP configuration, ion and electron currents in different regimes, and bias voltages. The L1B product LP_X_CA_1B delivers the LP calibration parameters derived for each probe by the L1B PLASMA processor. Finally the EFIX_LP_1B and EFIXLPI_1B products, provide the plasma parameters as density, electron temperature, plasma potential, together with the spacecraft position and the flags indicating a possible source of error for each data point. The EFIX_LP_1B are available at 2 Hz sample rate. By interpolating these products at exact UTC, 100 the EFIXLPI_1B products are obtained at 1Hz sample rate. Also, the LPs operate in different modes. The "harmonic mode" (HM) consists of sinusoidal varying biases applied to the LPs. Each HM cycle lasts for 0.5 s, and during the HM currents and admittances are measured. To our knowledge this method to obtain the current-voltage (I-V) characteristic of the space plasma is being used in orbit for the first time. The HM operates most of the time, while the classical "sweep mode" occurs each 128 s, and lasts for 1 s. In sweep mode the I-V curve is measured traditionally by scanning the probe bias over a range that 105 expectantly stretches from a dominant ion current (at negative bias) to a saturated electron current (positive bias). Sweep mode data are not used in the L1B PLASMA processor, but are separately analysed and provided as an additional "advanced" product.
Furthermore, each six hours a calibration mode is activated and a calibration data packet is generated on-board. This data are used for calibration purposes and, during the calibration mode, short data gaps are registered in L1B PLASMA products. The EFI-LP data products, and the other Swarm L1B products, are provided in daily files with a latency of four days. Detailed 110 information on Swarm L1B processors and data products are described by DTU (2019a, b). The Swarm products are freely accessible through the ESA dissemination server (ESA, a). In the next two sessions are described the recent data products evolution and data quality characterization.

Evolution from product baseline 04 to 05
The product baseline is a number identifying the data that were generated in a consistent way, i.e. using the same algorithms and input parameters, and, thus, constitute a data set. The first two of the last four digits of the product filename indicate the baseline. The product baseline is incremented when algorithm or input parameter upgrades lead to significant improvements in the data quality of the related products. The first PLASMA baseline went into operation in 2015 with the number 04. Before baseline 04, LP data were processed with a provisional processor by the Swedish Institute of Space Physics (IRF) (ESA, 2015).
When the final version of the PLASMA processor was ready to be transferred into operation, it was deployed directly with 120 baseline number 04 to be aligned with the other Swarm processor baselines. Thus, baseline lower than 04 are not available for EFI data products. Since September 2018, the PLASMA baseline has the number 05. An updated version of this processor has been deployed in operation in February 2020 containing only minor evolution, thus the baseline number remained unchanged.
A complete description of all the evolution introduced with these processors is reported in the related technical notes ESA (2018, 2020b). In the following we will discuss the major differences in PLASMA products between the baselines 04 and 05, 125 consisting on the electron temperature (T e ) computation from high-gain probe, and the decoupling of PLASMA processor from MAGNET processor.

Electron temperature computation from the high-gain probe
Each of the two LPs on board the Swarm satellites can be commanded to high or low gain. By electronically coupling a second shunt resistor in parallel the mode is low gain which allows higher probe currents to be measured without ADC (Analog Digital 130 Converter) overflows. Typically, one probe is set to the low gain and the other one to the high gain. The LP product parameters can be estimated from each probe. In practice the values often differ which we suspect is because of the different probe gains.
The first analysis, preceding baseline 04, estimated the electron density N e and electron temperature T e from the high gain probe for low densities/probe currents, from the low gain probes for high densities/probe currents, and by blending the results from both probes for an intermediate range of density/probe current. This avoided sudden jumps which would be caused by 135 switching the probes at threshold values. Typically the low gain probe needs to be used at the dayside magnetic equator because of very high density in the ionization anomaly, and the high gain probe is more appropriated for other regions. In the commissioning phase it became clear that the regularly occurring transition between probes produced unphysical variations of the estimated parameters even when smoothed by the intermediate blending. Therefore the algorithm to estimate the density was changed to use the weaker ion current instead of the retarded and saturated electron currents. The ion current and 140 admittance is always and very reliably measured by the high gain probe. The density product is therefore rather an ion density product, though often designated still N e . At Swarm altitudes, in the thermosphere and F region, the ion and electron densities are expected to be equal (only in the mesosphere and D region negatively charged ions and dust particles could cause N e to be less than the positive ion density). Also for T e the blending of high and low gain estimates was eventually abandoned in order to avoid producing unphysical variations at transitions. This, however, has the drawback, that especially in the ionization 145 anomaly ADC overflow occurs in the high gain saturated electron current. The T e from the low gain probe is dropped in, with a flag value as warning. This modification has been introduced with baseline 05. The region characterized by larger plasma density are generally observed at equatorial and low latitudes. In particular, in correspondence to day side equatorial crossings, it is possible to observe the typical double peak of the plasma density. This feature is related to the equatorial fountain effect characterizing the equatorial ionisation anomaly (Kelley, 2009). Also, the ADC overflows are frequently observed at equato-150 rial latitudes. Thus, to compare the measurements from baseline 04 (where high and low gain T e measurements were blended together) and baseline 05 (where only high gain measurements are used) it is worth to consider the latitudinal variation. Figure   3 shows the differences between T e obtained from baseline 04 (Te 04 ) and baseline 05 (Te 05 ) as a function of Quasi-Dipole ysis is limited to the latitudinal range to +/-50 • because at higher latitudes the electron temperature has a level of fluctuations too strong to obtain a meaningful comparison between Te 04 and Te 05 . Figure 3 demonstrates that Te 04 is on average larger that Te 05 at higher latitudes. On the day side, the two baselines are comparable at equatorial latitudes (panel (a)), while the differences in this region are larger on the night side (panel (b)). In particular, the night side presents a negative peak between −10 and 10 degrees of QD latitude. Also, in Figure 3 (c),we note a negative peak in correspondence to equatorial latitudes, 160 and a decrease for higher latitudes. Table 2 reports the average relative differences < ∆Te/Te 05 > for each MLT range, where ∆Te = Te 04 − Te 05 . The results demonstrate that, on average, the baseline 05 measures T e , which is 5-10% larger for the lower pair (Swarm A and C). This is a very good improvement, because it has been shown that the LP measurements of baseline 04, on average, underestimate the electron temperature with respect to ground measurements (Lomidze et al., 2018). Thus, the larger T e measurements obtained with baseline 05 represent a better agreement with ground observations. 165

Decoupling between PLASMA and MAGNET processors
In the previous configuration related to baseline 04, the PLASMA processor had a dependence on the MAGNET and OR-BATT processors. The ORBATT processor is fundamental for the L1B processing chain because it generates the L1B satellite ephemeris and attitude products, which are inputs for all the other processors. The MAGNET processor generates L1B magnetic measurements data products, which also contain the satellite position and attitude for convenience. The PLASMA processor 170 needs as inputs the spacecraft position and velocity expressed in the Earth-fixed reference frame. In baseline 04, the spacecraft ESA: Swarm L1B and L2 operational processors, https://earth.esa.int/documents/10174/1514862/ Swarm-L1B-and-L2-operational-processors.pdf, 2020b.