First Results From REPTile‐2 Measurements Onboard CIRBE

CIRBE (Colorado Inner Radiation Belt Experiment), a 3U CubeSat, was launched on 15 April 2023 into a sun synchronous orbit (97.4° inclination and 509 km altitude). The sole science payload onboard is REPTile‐2 (Relativistic Electron and Proton Telescope integrated little experiment—2), an advanced version of REPTile which operated in space between 2012 and 2014. REPTile‐2 has 60 channels for electrons (0.25–6 MeV) and 60 channels for protons (6.5–100 MeV). It has been working well, capturing detailed dynamics of the radiation belt electrons, including several orders of magnitude enhancements of the outer belt electrons after an intense magnetic storm, multiple “wisps”‐ an electron precipitation phenomenon associated with human‐made very low frequency (VLF) waves in the inner belt, and “drift echoes” of 0.25–1.4 MeV electrons across the entire inner belt and part of the outer belt. These new observations provide opportunities to test the understanding of the physical mechanisms responsible for these features.


Introduction
Trapped energetic electrons in the Earth's magnetosphere usually have a two-belt structure with a slot region in between.The inner belt, centered near L = 1.5 (where L is a parameter describing the magnetic shell that represents the geocentric distance in Earth radii (R E ) at the equator of the shell if the Earth's magnetic field is approximated as a dipole), consists of 10-100s of keV, occasionally to MeV, electrons, and multi-MeV-GeV protons.The outer radiation belt, centered near L = 4, consists of energetic electrons with 10s of keV to multi-MeV.The slot region, where fewer energetic electrons reside during quiet times, separates the two radiation belts.Such a description may give a static impression of the radiation belts.In fact, the electron radiation belts are constantly waning, waxing, and reforming, much more so for the outer belt (and the slot region) and during magnetic storms.Each reformed belt may have an extent, center location, and energydependent fluxes different from its predecessor (e.g., X. Li & Temerin, 2001;Ripoll et al., 2020;Schulz & Lanzerotti, 1974).
We report on the first results from CIRBE/REPTile-2, which can be considered a continuation of the kind of data previously gathered by the Van Allen probes.Our understanding of the dynamic variations of the radiation belt electrons has been enhanced since the launch of Van Allen Probes, two identical spacecraft, into a geo-transfer-like orbit, with perigees of ∼600 km altitude, apogees around 5.8 R E from the center of Earth and inclination of 10° (Kessel et al., 2012;Mauk et al., 2012), which operated between 2012 and 2019.The orbit period of the Van Allen Probes is ∼9 hr, meaning the Probes go through a certain L region twice in each orbit, spending more time near the apogees, and can measure most radiation belt electrons.A satellite in a low Earth orbit (LEO) will go through a certain L region four times in ∼1.5 hr (orbit period) but can only measure the electrons with their mirror points at or below the satellite's location, meaning a LEO satellite can only measure a fraction of the total electrons.Fortunately, there is a strong correlation between electrons with different equatorial pitch angles, or between electrons reaching LEO and electrons mirroring near the equator (e.g., Kanekal et al., 2001;X. Li, Baker, et al., 2017;X. Li et al., 2013).In other words, LEO measurements do reflect variations of the total radiation belt electrons.
In this letter, we briefly describe the CIRBE mission and its sole science payload, REPTile-2, focusing on electrons and showing the three populations of trapped, quasi-trapped, and untrapped electrons measured by REPTile-2 based on its location, and then show the measurement results and discussions, followed by a summary.Proton measurements will be discussed in a different paper.
Three days after launch, REPTile-2 was turned on and started collecting science data at a cadence of 1 s Figure 1 shows the locations of three populations of electrons: trapped, quasi-trapped, and untrapped, that are measured by REPTile-2 along its orbit for the first 40 days of its mission.The determination of which population a measured electron belongs to is based on REPTile-2's position (altitude, latitude, and longitude) and the local magnetic field calculated from the 13th generation International Geomagnetic Reference Field (IGRF-13) evaluated for the year of 2020 with the assumption that electrons are locally mirroring and that an electron will be lost if it reaches 100 km altitude.CIRBE has an active attitude control system which ensures REPTile-2's look direction is nearly perpendicular to the local magnetic field.REPTile-2 has a ∼51°field of view.Based on REPTile-2's position at a given time, measured electrons are either (a) in the bounce loss cone (BLC, red area: untrapped), that is, lost at its conjugate point; (b) in the drift loss cone (DLC, green area: quasi-trapped), that is, lost after drifting to the South Atlantic Anomaly (SAA); or (c) stably trapped (with different margins in the SAA indicated by different colorcontours), that is, able to complete a drift orbit around the Earth.

Dynamic Variations of the Electrons Associated With Magnetic Storms
Figure 2 shows five "snapshots" of radiation belt electrons before, during, and after an intense magnetic storm, the largest one in term of Dst index since 2017 on 24 April 2023 and before and after a moderate storm on 6 May 2023.Electron differential flux in Figures 2 and 3 are obtained by j(E i ) = n i /[(GΔE) i Δt] for channel i where n i (t) is the counts over Δt at channel i measured by REPTile-2, (GΔE) i is the effective geometric factor and effective energy bin width for channel i that was determined by the bowtie method (Khoo et al., 2022), a commonly used approach to convert count rates to differential flux due to its simplicity of implementation (e.g., Claudepierre et al., 2021;Selesnick & Blake, 2000).
The two-belt structure with a slot region in between is apparent in panel (b), during a quiet time in terms of the Dst and AE indices, shown in the top-left panel.Some fine-scale features in both the inner and outer belts are visible, thanks to the high energy resolution of REPTile-2.During the main phase of the storm, most of the electrons at L > 3.7 are gone as shown in panel (c); in a previous pass (not shown) at UT ∼22:03 all electrons at L > 3.5 were gone, suggesting that L = 3.5 was the last closed drift shell (LCDS) (e.g., Tu et al., 2014) during that time.After the storm, REPTile-2 measured electron flux enhancements by several orders of magnitude, especially at higher energies, seen throughout the outer belt and slot region, as shown in panel (d), in which the distinction between the inner and outer belt is diminished and the previously apparent slot region, in panels (b) and (c), has disappeared.Later, the two-belt structure with a slot region in between reappeared, as shown in panel (e), with somewhat different fine-scale structures compared to panel (b), for example, more higher energy electrons present; a dispersionless injection of electrons (an energy-independent enhancement) up to ∼2 MeV is discernible near L = 5.6, indicated by the vertical dashed line in panel (e).Then after a moderate storm on May 6, the reformed structures of the radiation belts, as shown in panel (f), include a remnant of higher energy (>∼1 MeV) electrons (e.g., Baker et al., 2013) that seemed to have remained at low L (<3.6), indicated by the "blob" located between L = 2.8-3.6 with energy >∼1 MeV, suggesting that most of the higher energy electrons initially located at L > 3.6, in panel (e), were lost while lower energy electrons were significantly enhanced throughout the outer belt and the slot region.A different version of Figure 2, where the electron fluxes are plotted versus time for the same measurements, is included in Supporting Information S1.
Figure 2 demonstrates variations of radiation belt electrons across a wide range in energy: 0.25-6 MeV and L: 1-7.Some of these features have been reported before in different formats, such as energy-dependent deep penetration of electrons in which very few multi-MeV electrons were observed below L = 2.6 (Baker et al., 2014;Hogan et al., 2021;X. Li, Baker, et al., 2017;X. Li et al., 1993X. Li et al., , 2021;;O'Brien et al., 2023;Zhao & Li, 2013a) while less energetic (∼1 MeV) electrons are often seen penetrating into the slot region and sometimes even into the inner belt (L < 2) (Baker et al., 2004;Claudepierre et al., 2019;Kim et al., 2016;Zhao & Li, 2013b;Zhao et al., 2023;Zheng et al., 2006).Dispersionless injections have mostly been studied using observations near the equatorial plane (Gabrielse et al., 2016;X. Li et al., 1998;Sarris et al., 2002).However, when a LEO measurement showing the characteristic feature of a dispersionless injection, as indicated by the dashed vertical line in panel (e), suggests that electrons of all equatorial pitch angles are injected simultaneously, something that has not been well modeled and understood (e.g., W. Li & Hudson, 2019;Ripoll et al., 2020).

Wisps
Quasi-trapped electrons are mostly produced by pitch angle scattering of trapped electrons by waves, such as plasmaspheric hiss, chorus, and EMIC waves, which are more prominent in the outer zone and slot region but much weaker or rarely measured in the inner belt (e.g., Zhao et al., 2019).Quasi-trapped electron fluxes in the lower L (<2) and higher L regions are vastly different since there are limited natural waves scattering electrons in the inner belt.So quasi-trapped electron fluxes are much lower, although lightning generated waves can leak into the magnetosphere and pitch angle scatter electrons (Abel & Thorne, 1998;Ripoll et al., 2019;Xiang et al., 2020).A distinctive phenomenon of energetic electron precipitation is associated with powerful ground-based VLF transmitter signals in the frequency range 10-25 kHz, which can also leak into magnetosphere when a transmitter is on the nightside (e.g., Liu et al., 2022;Starks et al., 2008), and scatter trapped electrons into quasi-trapped region, also known as the drift loss cone (DLC), leading to precipitation loss (Cunningham et al., 2020;Imhof et al., 1981a;Ma et al., 2017;Sauvaud et al., 2008;Selesnick et al., 2013).
Figure 3 shows two examples of measured "wisp" features by REPTile-2 inside the inner belt (L < 2), which is likely associated with the North West Cape (NWC) transmitters in Australia based on the measurement locations.f), of the first 50 channels of REPTile-2 electron measurements, which are color-coded and sorted in L. All measurements were taken when CIRBE was mostly above the SAA region (trapped areas).The time duration for each "snapshot" is listed on the top right corner of each panel.Count rates ≤1 are excluded to reduce the noise.The Dst index is provisional from WDC for Geomagnetism, Kyoto (http://wdc.kugi.kyoto-u.ac.jp/wdc/Sec3.html)and the AE index is from https://lasp.colorado.edu/space_weather/dsttemerin/dsttemerin.html(X.Li et al., 2007;Luo et al., 2013).
Panel (a) shows a measurement in the quasi-trapped region (or DLC) in the northern hemisphere, the corresponding CIRBE track is indicated in Figure 1, labeled "3a."A wisp is evident at ∼19:40 UT (L ∼ 1.7), which is likely created by first order cyclotron resonant scattering on field lines near the NWC VLF transmitter.The scattered electrons bounce between their mirror points in the north and south while drifting eastward toward the SAA where they precipitate into the atmosphere.Panel (b) shows a measurement in the southern hemisphere, starting from trapped region (larger L), with high electron fluxes, moving northward to lower L into the quasitrapped region.The CIRBE track is indicated in Figure 1, labeled "3b."Multiple "wisps" at L ∼ 1.7-1.9 between 18:27 and 18:30 UT with the strongest one near 18:30 UT are present.Most previous studies have focused on a single wisp, which are common.Here we show that CIRBE/REPTile-2 captured a case of multiple wisps.Using observations from the PROBA-V satellite, Cunningham et al. (2020) reported that enhanced quasi-trapped fluxes induced by the NWC transmitter can appear at multiple L-shells, which might be a multiple wisp event.However, due to energy resolution limitation (seven energy channels from 0.5 to 20 MeV), PROBA-V/EPT (Cyamukungu et al., 2014)  Other notable features in Figure 3 include: (a) some discernible injections of electrons occurred in the outer belt, indicated by the vertical dashed lines, and (b) multiple local peaks in panel (a) between 19:46 and 19:47 UT, which will be further discussed in Section 3.3.

Drift Echoes
For this section we use the least-squares inversion method (Khoo et al., 2022;Selesnick et al., 2018;Tarantola, 2005;Tarantola & Valette, 1982) to obtain differential fluxes of electrons.The least-squares inversion method differs from the bowtie method, which is a simpler approach and faster to calculate as it assumes a shape of the flux spectrum to obtain a single, representative channel response and energy (e.g., Khoo et al., 2022;Selesnick & Blake, 2000).The least-squares inversion method is more accurate (and more involved or timeconsuming) because it applies the detailed instrument response function and is accomplished by starting with an assumed prior flux model and then iteratively varying it to minimize the least-squares criterion between the observed and modeled count rates until the most probable flux spectrum is determined.
Figure 4a presents color-coded logarithmic electron flux as a function of time and energy, accompanied by a secondary x axis showing the corresponding L, during a pass through the center of the SAA, which is one orbit after the data in Figure 2c.CIRBE's track is labeled as "4" in Figure 1, between 0:53 and 1:18 UT on April 24, when the magnetic storm was approaching the end of its main phase (Dst panel of Figure 2).Moving from north to south, REPTile-2 measured the trapped electrons in the inner belt and then the quasi-trapped electrons in the outer belt, crossing L-shells from 1.1 to 6.5.The drift-echoes ("zebra stripes") can be seen in the entire inner belt (down to L = 1.17) through the measured energy range: 0.25-1.4MeV.In addition, similar stripes are also visible near the inner edge of the outer belt.Black curves on top of the color-coded electron fluxes are contours of electron drift period in hr, which are calculated through particle tracing, based on the IGRF-13 model (Alken et al., 2021).Detailed description of the drift period calculation is given in Supporting Information S1.
To extract the peaks and trough locations of the drift echo features shown in Figure 4a for further analysis, we compute moving averages of the logarithm of the fluxes over ±19% in energy and subtract the resulting smoothed fluxes from the measurements (Lejosne & Mozer, 2020;Liu et al., 2016).The smoothed fluxes are displayed in Figure 4b, in the same format of Figure 4a. Figure 4c shows the detrended electron fluxes after subtraction.Red indicates peaks of the drift echoes, while blue indicates troughs.Drift echoes of energetic electrons have previously been reported only either in the inner or outer belt (Imhof et al., 1981b;Lejosne & Mozer, 2020;Osmane et al., 2023;Sauvaud et al., 2013;Ukhorskiy et al., 2014).Here they are seen through the entire inner belt and part of the outer belt simultaneously, from L = 1.17 to L = 3.6.We are not aware that such features have been reported previously.Furthermore, stably trapped electrons are seen up to 1.4 MeV in the inner belt, in contrast to previous reports that no MeV electrons (Fennell et al., 2015) or no >1.1 MeV electrons (e.g., Claudepierre et al., 2019) were measured in the inner belt during the Van Allen Probes era.We should emphasize that this measurement was made during the largest magnetic storm, in terms of Dst, since September 2017, comparable to the most intense storms seen by the Van Allen Probes.It should be noted that measurements for the outer belt were taken in the quasi-trapped region as shown in Figure 1, where these measured electrons are supposed to precipitate into the atmosphere within one drift period.In other words, the drift echo features are not expected to be maintained or observed in the quasi-trapped region.These measured drift-echo features suggest that trapped electrons had been continuously scattered into the drift loss cone (e.g., Xiang et al., 2019) and measured by REPTile-2 before their loss, indicating that the drift-echo features observed in the quasi-trapped region were also present in the trapped population not measurable by REPTile-2.
Figure 4d presents the detrended flux rearranged as a function of time (and L) and drift frequency.With the drift periods at all energies calculated, the energy of the electron flux in Figure 4c is converted to drift frequency.To obtain detailed time information of the drift echoes, we follow the theory of Lejosne and Roederer (2016) and Lejosne and Mozer (2020), assuming difference in the angular portions that electrons at the central energy of the consecutive peaks or troughs have swept is: Δφ = 2π.The duration of the drift echo can be inferred from the drift frequency difference between the consecutive peaks or troughs: Δt|ω i ω i+1 | = 2π, where Δt is the estimated backward onset time of the drift echoes, ω i and ω i+1 are the drift frequencies of two consecutive peaks or troughs.Based on this, for the model of a single onset time, we would expect, ideally, the gaps between consecutive peaks and troughs to be evenly spaced when plotted against frequency and time as in Figure 4d, which indeed shows rather straight "stripes" through the entire inner belt.However, the "stripes" in the outer belt seem less straight and with bigger gaps between them.The onset time for the drift echoes in the inner belt was estimated 6 hr ago while the onset time for the outer belt drift echoes was about 2 hr ago.
It is possible that all the drift echoes shown in Figure 4 were caused by a single onset, such as a prompt emergence of an electric field (e.g., Lejosne & Mozer, 2020).However, stronger, and more frequent disturbances, for example, transient electric fields and ultra-low frequency (ULF) waves, in the outer belt could result in some of the drift echoes merged and appeared to be formed ∼2 hr ago.Fluxes for the previous 3 and subsequent 2 passes through, or partially through, the trapped regions are included in Supporting Information S1 and show that drift echo features are indeed measured over multiple CIRBE orbits with differing spacing, and more persistent in the inner belt.There are more uncertainties associated with the calculated drift frequencies for the outer belt electrons.One involves the assumption of the pitch angle of the trapped electrons as discussed in Supporting Information S1, the other is that we use the IGRF model to determine the L for the outer belt as well; during such an intense storm, the influence of external currents becomes more significant in the outer belt region.Detailed modeling efforts will be required to fully understand these measured drift-echo features across such wide ranges in L, energy, and time.

Summary
REPTile-2 onboard CIRBE has been collecting data since 19 April 2023.With its high energy resolution (60 channels for electrons: 0.25-6 MeV) and time resolution of one sec, various detailed features of radiation belt electrons have been observed.Among them are (a) variations of relativistic electrons over a wide energy range, 0.25-6 MeV, in the Earth's magnetosphere, especially those associated with the largest magnetic storm (23-24 April 2023) in the past 6 years; (b) multiple wisp features associated with human-made VLF waves in the inner belt; (c) "drift echoes" or "zebra stripes" of 0.25-1.4MeV electrons across the entire inner belt and part of the outer belt.These new observations provide opportunities to test existing theories and to better understand the dynamics of radiation belt electrons.obtaining and renewing the radio licenses of CIRBE.We owe our deepest gratitude to Rick Kohnert of LASP, who was the project manager for CIRBE and guided engineers and students since the "cradle" phase of CIRBE.Sorrowfully, Rick passed away before the launch of CIRBE, but we are certain that he would be proud of the contributions that CIRBE has made and will continue to make to the scientific community.His legacy lives on.This work was supported by NASA Grants 80NSSC19K0995 and 80NSSC21K0583, and NSF Grant AGS 1834971.

Figure 1 .
Figure 1.CIRBE orbital tracks from 19 April to 30 May 2023 plotted over a world map, color-coded for regions corresponding to different populations of the electrons mirroring at the satellite: untrapped (red), quasi-trapped (green), and stably trapped (different color-contours indicating different margins of stably trapping).The detailed method and the definition of the trapped status margin are described in Selesnick (2015).L contours, calculated McIlwain L based on the IGFR model, are overplotted as the black solid lines and the white region is for L > 10. Specific orbit tracks, 3a, 3b, and 4, are for Figures 3 and 4, respectively.

Figure 2 .
Figure 2. Dst and AE indices on top-left panel, where the five vertical bars (b-f) indicate the time for the five "snapshots," panels (b)-(f), of the first 50 channels of REPTile-2 electron measurements, which are color-coded and sorted in L.All measurements were taken when CIRBE was mostly above the SAA region (trapped areas).The time duration for each "snapshot" is listed on the top right corner of each panel.Count rates ≤1 are excluded to reduce the noise.The Dst index is provisional from WDC for Geomagnetism, Kyoto (http://wdc.kugi.kyoto-u.ac.jp/wdc/Sec3.html)and the AE index is from https://lasp.colorado.edu/space_weather/dsttemerin/dsttemerin.html(X.Li et al., 2007;Luo et al., 2013).
observations do not show clear wisp structures.Cunningham et al. (2020) explained this phenomenon by resonance at distinct wave normal angles of NWC transmitter signals.The multiple wisps can also be generated by transmitter signals with different frequencies or by harmonic resonances with the same wave frequency.Detailed modeling efforts will be required to quantitatively understand such multiple wisp phenomena.

Figure 3 .
Figure 3. Like Figure 2, x-axis is for time (UT) with L values shown below, and the measurements were taken when CIRBE was in the quasi-trapping region, as indicated by the highlighted orbital tracks in Figure 1, labeled 3a for panel (a) in northern hemisphere on 2023/05/07 and labeled 3b for panel (b) in southern hemisphere on 2023/ 04/30.The wisp features are visible at L ∼ 1.7 in panel (a); and panel (b) shows multiple wisps in L ∼ 1.7-1.9, with the strongest one at lower L. Count rates ≤1 are excluded to reduce noise.

Figure 4 .
Figure 4. (a) Electron fluxes derived from the least-squares method for the pass labeled "4" in Figure 1.Black curves overplotted on top of the color-coded electron fluxes are selected contours of electron drift period in hr.(b) Electron fluxes using a moving average window of ±19% in energy.(c) Detrended electron fluxes, calculated by subtracting the moving averaged flux from measured flux, in the same format as (a).(d) Detrended electron fluxes rearranged as a function of electron drift frequency.Drift frequencies for all energies are calculated.All fluxes plotted over time use a moving average of count rate with 10 times of the measurement cadence (1 s) window, that is, a 10 s moving average.Count rates ≤1 after smoothing are set to 0 to remove noise.Time periods with zero count rates after filtering are displayed as blank space.