Volume 126, Issue 3 e2020JA028637
Technical Reports: Data
Free Access

RBSP-ECT Combined Pitch Angle Resolved Electron Flux Data Product

A. J. Boyd

Corresponding Author

A. J. Boyd

Space Science Applications Laboratory, The Aerospace Corporation, El Segundo, CA, USA

Correspondence to:

A. J. Boyd,

[email protected]

Search for more papers by this author
H. E. Spence

H. E. Spence

Institute for the Study of Earth, Oceans and Space, University of New Hampshire, Durham, NH, USA

Search for more papers by this author
G. D. Reeves

G. D. Reeves

Space Science and Applications Group, Los Alamos National Lab, Los Alamos, NM, USA

The New Mexico Consortium, Los Alamos, NM, USA

Search for more papers by this author
H. O. Funsten

H. O. Funsten

Space Science and Applications Group, Los Alamos National Lab, Los Alamos, NM, USA

Search for more papers by this author
R. M. Skoug

R. M. Skoug

Space Science and Applications Group, Los Alamos National Lab, Los Alamos, NM, USA

Search for more papers by this author
B. A. Larsen

B. A. Larsen

Space Science and Applications Group, Los Alamos National Lab, Los Alamos, NM, USA

The New Mexico Consortium, Los Alamos, NM, USA

Search for more papers by this author
J. B. Blake

J. B. Blake

Space Science Applications Laboratory, The Aerospace Corporation, El Segundo, CA, USA

Search for more papers by this author
J. F. Fennell

J. F. Fennell

Space Science Applications Laboratory, The Aerospace Corporation, El Segundo, CA, USA

Search for more papers by this author
S. G. Claudepierre

S. G. Claudepierre

Space Science Applications Laboratory, The Aerospace Corporation, El Segundo, CA, USA

Department of Atmospheric and Oceanic Sciences, University of California Los Angeles, Los Angeles, CA, USA

Search for more papers by this author
D. N. Baker

D. N. Baker

Laboratory for Atmospheric and Space Sciences, University of Colorado Boulder, Boulder, CO, USA

Search for more papers by this author
S. G. Kanekal

S. G. Kanekal

Heliophysics Science Division, NASA Goddard Space Flight Center, Greenbelt, MD, USA

Search for more papers by this author
A. N. Jaynes

A. N. Jaynes

Department of Physics and Astronomy, University of Iowa, Iowa City, IA, USA

Search for more papers by this author
First published: 11 February 2021
Citations: 12

Abstract

We describe a new data product combining pitch angle resolved electron flux measurements from the Radiation Belt Storm Probes (RBSP) Energetic Particle Composition and Thermal Plasma (ECT) suite on the National Aeronautics and Space Administration’s Van Allen Probes. We describe the methodology used to combine each of the data sets and produce a consistent set of pitch-angle-resolved spectra for the entire Van Allen Probes mission. Three-minute-averaged flux spectra are provided spanning energies from 15 eV up to 20 MeV. This new data product offers researchers a consistent cross calibrated data set to explore the particle dynamics of the inner magnetosphere across a wide range of energies.

Key Points

  • A new combined pitch angle resolved electron flux data product, spanning 15 eV up to 20 MeV, for the Van Allen Probes mission is described

  • Results from cross-calibration of angular distributions of the RBSP-ECT data are presented

  • Criteria used to determine data quality flags for each energy channel are also described

1 Introduction

Over of the course of its 7-year mission, the Van Allen Probes has provided unprecedented observations of the inner and outer radiation belts. This is due in large part to the Van Allen Probes’ comprehensive suite of instruments that can measure the particles, fields and waves. A key part of this instrumentation was the Energetic Particle, Composition and Thermal Plasma Suite (RBSP-ECT; Spence et al., 2013), which provided measurements of electrons from a few eV up to tens of MeV. Ions are also measured over a similar energy range, but they are not discussed here.

In this paper, we describe a new data product that combines unidirectional electron fluxes from all of the ECT instruments. This is the second in a series of combined ECT data products. The first was the combined spin-averaged electron flux (Boyd et al., 2019). Unlike spin-averaged fluxes, unidirectional flux refers to a specific, limited, directional field of view. The third and final data set will be a combined electron phase space density. All of these datasets use a 3-min time cadence and are meant to compliment the currently available ECT data products. Moving forward, the intent is to have these combined data files serve as the starting point for any analysis of the Van Allen Probes electron data.

The basic steps for generating this data set are as follows: (1) do a 3-min time average of each of the input datasets removing low count and out-of-family measurements, (2) at each time, fit the pitch angle distributions to get a consistent set of combined spectra, (3) perform a spline fit of the energy spectra at each pitch angle value. The rest of the paper describes each of these steps in more detail: In Section 2, we describe the input data sets, Section 6 discusses the pitch angle fitting, Section 10 compares adjacent energy channels between the different ECT instruments, Section 13 describes the spline fitting methodology, Section 14 discusses known caveats, Section 18 discusses potential scientific applications and Section 19 gives a summary.

2 Data

The ECT instrument suite is composed of three instruments: The Helium Oxygen Proton Electron Mass Spectrometer (HOPE; Funsten et al., 2013), the Magnetic Electron Ion Spectrometer (MagEIS; Blake et al., 2013) and the Relativistic Electron Proton Telescope (REPT; Baker et al., 2013). In this section, we describe the unidirectional flux measurements made by each of the instruments and a brief summary of the instrument-specific data processing that goes in to the generation of the combined ECT data set. For all the instruments, we are utilizing the latest available data which is release 04 for HOPE and MagEIS and release 03 for REPT. A final reprocessing for the combined electron data product will be produced using the final HOPE, MagEIS, and REPT archival data products.

For each of the datasets we do a 3-min time average. This places all the datasets on the same time base and helps to improve the counting statistics. In order to avoid the influence of outlier fluxes, we use a 10% trimmed mean to get the value in each time bin. The time label is placed at the center of the bin, so for example data labeled 1:01:30 UT will cover 1:00 to 1:03 UT. In each bin, we require at least five nonzero fluxes in order to get an accurate average. If there are fewer points than this, the bin is marked as invalid (i.e., set to a fill value of −1.0E31).

2.1 HOPE

The lowest energy instrument in the ECT suite, HOPE uses a time of flight measurement and channel electron multiplier detectors to observe of electrons from ∼15 eV up to 51 keV (Funsten et al., 2013). HOPE measures electrons and ions on alternating spacecraft spins, so unlike the other ECT instruments, electron measurements are only available every ∼22 s (approximately twice the Van Allen Probes spin period). The instrument features five pixels that are coplanar with the spacecraft spin axis. Each of these pixels use 72 energy steps with 16 azimuthal sectors. The measurements in each of these sectors are subsequently binned into 11 local pitch angle values. HOPE has three modes of operation that influence what energies are measured in the steps: the first is apogee mode where the instrument measures the full energy range, the second is perigee mode where all but the lowest energies (<25 eV) are measured and the final is burst support mode, where a small subset of energies are measured at greater time resolution (for more information see Funsten et al., 2013). For the data product described in this paper, only apogee mode data are used.

The ECT combined spin-averaged data (Boyd et al., 2019) featured several data processing steps that were applied to the HOPE data that we also include in this pitch-angle-resolved data product. The first is identifying and removing times when the highest HOPE energy channels measure low count rates. These times are identified by comparing the corresponding HOPE and MagEIS channels in the overlap energy region of 30–50 keV. The HOPE electron detection efficiencies fall off rapidly for energies greater than 1 keV. Therefore, the channels in the overlap region have low detection efficiencies and in the event of disagreements we defer to the MagEIS measurements. Figure 1 shows a comparison between the 32/33 keV electron flux from HOPE and MagEIS at 90° local pitch angle. The points are color coded based on the total HOPE counts recorded in the 3-min window. To avoid the scatter at lower fluxes, we exclude all HOPE observations where there are fewer than 10 counts (shown in red-to-yellow hues). This cutoff is applied to all energies above 10 keV for all pitch angle bins. This 10-count threshold is comparable to the 125 omni-directional count cutoff that was used in the spin-averaged data.

Details are in the caption following the image

Comparison of the 90° local pitch angle 32 keV unidirectional electron flux (FEDU) from MagEIS and HOPE. The color of the points shows the total number of HOPE counts in each 3-min bin. In order to remove points with poor agreement, only times with more than 10 counts (shown in green) are included.

The second processing step is to apply some corrections to the HOPE data. The first is an efficiency correction for days when there is a poor agreement with the MagEIS data. The same daily correction factor that was applied to the spin-averaged data is applied here at all pitch angles. The second is a correction at low L-shells (L < 2.5). In the inner zone, HOPE measures a significant background due to the population of high energy protons. To correct for this, we use the ratio of HOPE to the 90° MagEIS fluxes at 32 keV to estimate the number of background counts and subtract those counts from the HOPE measurements at each energy channel. This combined correction factor is recorded in the data files as HOPE_FACTOR variable.

2.2 MagEIS

The medium-energy instrument in the ECT suite, MagEIS uses a strong magnet to steer particles into a set of solid state detectors that measure electrons from ∼30 keV up to ∼4 MeV (Blake et al., 2013). MagEIS actually consists of 4 separate instruments that cover different energy ranges: a LOW unit that cover 30–200 keV, two medium units (M35 and M75) that cover ∼200 keV–1 MeV and a HIGH unit that covers 1–4 MeV. The LOW, M75 and HIGH units each provide identical pitch angle coverage centered at 75° from spacecraft spin axis. The M35 unit was originally intended to provide additional pitch angle coverage, but the final configuration of the spacecraft caused this improvement to only be marginal. Given the challenges that come with integrating the M35 observations, we chose to remain consistent with the MagEIS Level 3 data and only include observations from the LOW, M75 and HIGH units. The MagEIS fluxes are reported at the spacecraft spin cadence (∼11 s). Each of the spacecraft spins is divided into up to 64 sectors which each measuring part of the pitch angle distribution. The number of sectors used has changed several times over the course of the mission. The measurements in these sectors are ultimately binned to 11 pitch angle values at 21 energy values. While both HOPE and MagEIS have the same number of pitch angle bins (11), the values of pitch angle bins for each of the instruments are different.

A highly effective background correction algorithm has been developed for the MagEIS data (Claudepierre et al., 2015). For this data set, the background corrected data is utilized whenever it is available. However, there are periods of time when the instrument is operating a mode that prevents the background calculations and makes the corrected fluxes are unavailable. During these times, the uncorrected fluxes for the LOW and M75 unit are used. These times can be identified with the unidirectional electron flux quality flag (FEDU_Quality). The uncorrected fluxes for the HIGH unit are not used in this data set.

Similar to HOPE, there were several data processing steps included in the spin-average combined data that we now use here. In particular, we remove data points from the HIGH unit when the average FEDU_ERROR in the 3-min time window is above 75%. As described in Claudepierre et al. (2015), the FEDU_ERROR variable for the background corrected data incorporates errors from both Poisson counting statistics and the measured background contamination. Applying that condition would remove nearly all of the observations from the highest two energies channels that measure >3 MeV electrons. Therefore, for consistency, we do not include any of the measurements from these two channels.

2.3 REPT

The highest energy instrument in the ECT suite, REPT uses a well-shielded stack of solid state detectors to measure electrons from 2–20 MeV (Baker et al., 2013). REPT points perpendicular to the spin-axis. Measurements are taken every spacecraft spin (∼11 s). Each of these spacecraft spins is divided in 36 sectors which each measure part of the pitch angle distribution. In the REPT Level 3 data, these sectors are binned into 17 pitch angles across 12 energy channels.

Once again, we include several processing steps that were introduced in the spin-averaged combined data product. First, in both the spin-averaged and unidirectional fluxes, the lowest REPT energy channel (1.8 MeV) tends to be out-of-family with the adjacent MagEIS energy channels. Following the methodology of Boyd et al. (2019), we do not include data from this REPT channel. Additionally, the higher REPT energy channels have a background due in large part to galactic cosmic rays. Boyd et al. (2019) describe a method to estimate this background. Here, we directly implement that technique: for a given energy, if the spin-averaged data was below the estimated background, the flux at all corresponding pitch angle bins are set to fill values.

3 Pitch Angle Fitting

3.1 Legendre Polynomials

The above methodology places all fluxes on a uniform time-base, but not a uniform set of pitch angle bins. Therefore, we first fit the pitch angle distributions before combining the HOPE, MagEIS and REPT data into a single spectrum. While there are a number of techniques available to fit angular distributions, here we use Legendre polynomials. This technique has been shown to be effective in reproducing a large variety of pitch angle distributions in the ECT data (Chen et al., 2014; Zhao et al., 2018). Using Legendre polynomials, the flux is fit as:
urn:x-wiley:21699380:media:jgra56302:jgra56302-math-0001
where j is the flux as a function of local pitch angle α, Pn is the nth order Legendre polynomial and Cn is the corresponding coefficient. The goal of the fitting routine is to determine the coefficients that produce an accurate representation of the observed flux. Here, we use a least-squares fitting routine that is part of the python NumPy package. Following the results of Chen et al. (2014) and Zhao et al. (2018) we do not include terms higher than P10. For trapped electrons, we expect the pitch angle distribution to be symmetric about 90°. Therefore, for most of the fits, only the even terms are needed. However, there are times, particularly in the HOPE energy range, where the observed angular distribution is clearly not symmetric. For these times, the odd terms are included in order to give a better fit of the observations.

The fit is done on the log of the flux. This serves the dual purpose of reducing the range of values and removing bad/missing points (negative fluxes), giving a more stable result. At least five nonzero flux values are required to perform the fit. The fitted flux is output at 35 local pitch angle values covering every 5° between 5 and 175°. In addition, the coefficients (Cn) are included in the data files, allowing the user to reconstruct the fits and output to any additional desired pitch angle values.

The first step in the fitting procedure is to determine if the measured pitch angle distribution is asymmetric and the odd terms need to be included. To determine this, we take the mean ratio of second, third and fourth pitch angle pairs as defined below:
urn:x-wiley:21699380:media:jgra56302:jgra56302-math-0002
Where j is the flux and α2, α3, α4 are the second, third and fourth pitch angles values. For example, for HOPE these values would correspond to the 18°, 36° and 54° bins. The first pitch angle bin (4.5° on HOPE) is not included here because it rarely contains a nonzero flux. Any points where the flux is ≤0 are not included in the calculation of the asymmetry. If there are not at least two valid pitch angle pairs, A it is set to 1.0. If A<0.75 or A>1.3, then the odd terms are included. Figures 3c and 3d show examples of asymmetric fits and Figure 2 shows how often these asymmetric fits are needed as a function of energy. They are most commonly needed for the HOPE channels below 100 eV, where less than half the distributions require asymmetric fits. At higher energies, these comprise a much smaller percentage. As shown with the light-colored bars, in most cases the odd terms represent a small perturbation relative to the even terms.
Details are in the caption following the image

Histogram of fraction of asymmetric pitch angle fits (defined by nonzero P1) as a function of energy. The lighter portion of each bar shows points where the odd coefficients have a small contribution (|P1| < 0.2*|P2|).

Next, there are a few steps to prepare the data before fitting. First, if a symmetric fit is being used and there are missing values in any of the four most field aligned bins (i.e., the two bins closest to 0° and the two bins closest to 180°), the corresponding value is mirrored. An example of this is shown in Figure 3e. Second, since the fit is done on the log (flux) any values where the flux is equal to 0 (indicating no measured counts or below background) are problematic. To ensure the fit is properly constrained, the flux at these points is set to 75% of the minimum nonzero flux. These values are only included to ensure the fit is well behaved and are later removed. An example of this process is shown in Figure 3f.

Details are in the caption following the image

Examples of the pitch angle fitting routine. Panels a and b show typical fit examples for REPT. Panels c and d show asymmetric fits from HOPE. Panels e and f show MagEIS examples where mirrored and artificial points are added to compensate for missing values at smallest two pitch angles and zero fluxes near 90°.

The next step is to fit the data. Each fit is performed up to 4 times. The first fit is done using only terms up to P4, the second fit including only terms up to P6, third fit up to P8 and fourth fit up to P10. Each fit is then evaluated using the R2 value. For each pitch angle distribution, we use the fit where R2 ≥ 0.85 and where the R2 value of the next highest fit does not show significant improvement (ΔR2 < 0.01). Often, the first fit (terms up to P4) is good enough to reproduce the angular distribution, as shown in Figure 3a.

However, sometimes additional terms are needed to get an accurate fit, as shown in Figure 3b. If the R2 value is still too small after including terms up to P10, the fit is considered invalid. The reasoning behind this process is to produce an accurate representation of the observed pitch angle distribution and avoid overfitting. Adding more terms to the fit can often result in the introduction of unphysical features as demonstrated in Figure 3d. Therefore, we try and use the simplest form (i.e., fewest terms) possible. Finally, if the fit using the full pitch angle range is invalid, the entire process is repeated using a limited range of 30–150°.

3.2 Omni Directional Fluxes

In addition to the unidirectional fluxes, we also calculate an omni-directional flux (FEDO) from the pitch angle fits. We follow the convention of reporting omni-directional fluxes per-steradian:
urn:x-wiley:21699380:media:jgra56302:jgra56302-math-0003
Where j is the unidirectional flux and α is the local pitch angle. Comparing these omni-directional fluxes with the combined spin-averaged fluxes shows excellent agreement across all energies, with nearly all points within a factor of 2. An example of this comparison at 2.2 MeV is shown in Figure 4. The comparison is even better for the HOPE energies, which were already included as an omni-directional flux in the combined spin-averaged data. In addition to providing the combined omni-directional spectra (FEDO), we also provide a set of spline fit spectra (FEDO_FIT) following the method outlined in Section 13.
Details are in the caption following the image

Comparison between the calculated omni-directional flux and the spin-averaged flux from the combined ECT data product at 2,280 keV. The red, green and blue diagonal lines show the factor of 1, 2 and 10 differences respectively.

3.3 Equatorial Pitch Angles

The fluxes discussed up this point (and the corresponding fits) were done in terms of the local pitch angle. In these data files, we also provide the combined spectrum in terms of equatorial pitch angles (FEDU_Eq). The structure of these variables is the same and the equatorial pitch angles are output at the same static grid as the local pitch angles. Since the fits are in terms of local pitch angle, we first use the following equation to map each output equatorial angle to the corresponding local pitch angle:
urn:x-wiley:21699380:media:jgra56302:jgra56302-math-0004
Where α is the local pitch angle, αeq is equatorial pitch angle and B/Beq is the ratio of the local and equatorial magnetic field strengths taken from the Magnetic Ephemeris (MagEphem) files (Spence et al., 2013) that utilize the Olson and Pfitzer (1977) magnetic field model. For equatorial pitch angles greater than 90, 180-α is used. Each of these values is then used as input to the Legendre polynomial fits to get the equatorial fluxes. For equatorial pitch angles that are not observed when the spacecraft is too far off the magnetic equator, the fluxes are set to fill values.

4 Cross Calibration

In this section, we compare the overlapping energy values between the ECT instruments to evaluate how well they are cross-calibrated. In this section, we use the fit fluxes described in Section 6. These fits do not significantly alter the results presented here, but for completeness, comparisons of the measured fluxes at nearby pitch angle bins are available in the supporting information S1.

4.1 HOPE-MagEIS

The first comparison is between HOPE and MagEIS, which have overlapping energy channels at 33/32 and 51/54 keV. Figure 5 shows a comparison of these two channels on both spacecraft. For 90° pitch angle, the results are consistent with what was observed with the spin-averaged data (Boyd et al., 2019). The mean ratios are slightly smaller (closer to 1), but there is a bit more spread in the data, particularly near lower fluxes where the Poisson error becomes larger. However, more than 80% of the values are within a factor of 2.

Details are in the caption following the image

Comparison of the adjacent channels on HOPE and MagEIS for 90° pitch angle. The color indicates the number of points in each bin. In each panel, the mean, median, and quartiles refer to the HOPE/MagEIS ratio.

At pitch angles away from 90°, the agreement does start to get slightly worse. At 60°, there is still very good agreement, as the mean ratios at 32 keV are 1.02/0.8 for spacecraft A/B. At 30°, there starts to be more spread in the data, and the mean ratio moves to 1.48/1.2 at 32 keV, likely influenced by poor counting statistics at these lower pitch angles. However, even at the 30° more than 60% of the points are within a factor of 2. Figures showing these additional comparisons can be found in the supporting information S1.

One noticeable feature at the highest fluxes (above 106) is a clear flattening as HOPE continues to increase while MagEIS stays constant. This is due to saturation in these two MagEIS-LOW channels. This saturation was only observed near 90° during a small number of intervals during the entire Van Allen Probes mission. We do not apply a correction to the saturated MagEIS data, but these times are flagged with the FEDU_QUALITY flag and are given less weighting in the spline fit described in Section 13.

4.2 MagEIS-REPT

Next, we compare the adjacent energy channels on MagEIS and REPT. Nominally, there are five overlapping energy channels. However, as described in Section 2, the lowest REPT channel at 1.8 MeV and the highest two MagEIS channels >3 MeV are typically out of family with the surrounding channels and so are excluded here. This leaves two remaining energy channel pairs for comparison at 2.1 and 2.6 MeV. Figure 6 shows these comparisons for 90° pitch angles. In the figure, the expected mean takes into account the energy difference of the two channels, assuming a power law spectrum with a slope of −8.

Details are in the caption following the image

Comparison of adjacent energy channels from MagEIS and REPT for 90° pitch angles. The color indicates the number of points in each bin. In each panel, the mean, median, and quartiles refer to the MagEIS/REPT ratio.

Overall, the agreement is very good and is nearly identical to what was observed for the spin-averaged fluxes. Nearly all of the points agree to within a factor of 2, and the mean ratio is close to what is expected. The agreement is particularly good at 2.6 MeV, where the two instruments have very similar energy values. At pitch angles of 30 and 60° (shown in the supporting information S1), the results are very similar, with mean shifting by <0.05, indicating that the angular distributions are consistent between the two instruments.

5 Spline Fitting

The HOPE, MagEIS and REPT measurements are combined in the FEDU spectrum, which contains the electron flux at the 102 measured energy channels (72 from HOPE, 19 from MagEIS and 11 from REPT) and 35 fit local pitch angles. The flux at each of these pitch angle values is then fit using a univariate cubic spline to give a consistent set of spectra. By default, the spline fit is output at 127 logarithmically spaced energy channels and the knots and coefficients are provided in the data files if users want to specify their own energy values. The methodology for each of these spline fitting routine is briefly summarized here, but more details can be found in Boyd et al. (2019).

To calculate the spline, we use the SciPy software package. The routine uses two inputs: the weight of the points and a smoothing parameter s, which restricts the residual of the fit:
urn:x-wiley:21699380:media:jgra56302:jgra56302-math-0005
where w are the weights, y is the log of the input fluxes, z is the spline fit, and s is the smoothing factor. Nominally, s is set to 0.25 for all fits. The knots and coefficients are then automatically determined. Points that are considered less reliable due to poor counting statistics or potential saturation have their weight reduced from 1.0 to 0.5. These weights were chosen to include as much data as possible while allowing the fit to place more weight on reliable values. These reduced-weight points can be identified with the FEDU_Quality data flag. Additionally, the fit is not used to extrapolate, so the lower bound is set to the lowest valid energy and the upper bound is set to the highest valid flux for L > 2.5 or 1 MeV for L < 2.5 (see Section 14).

Each fit undergoes a number of validation checks. The first is a simple check that the residual is less than the smoothing parameter s. If not, the smoothing factor is increased to 0.5. Next, the fit is checked near any significant gaps in the input spectrum. If there are more than 18 missing energy values below 100 keV or more than six missing values above 100 keV, the gap is included in the fit spectrum and any affected fluxes are set to fill values. For smaller gaps, the spline is compared to a linear interpolation and if the root-mean-square deviation is above 0.4, the smoothing is increased to accommodate these. Next is an examination of the second derivative. Any points where the absolute value of second derivative is exceptionally large (indicating a nonphysical change in spectral slope) the smoothing factor is increased and a data gap is added if necessary. Finally, at the highest energies, we expect to consistently observe a negative spectral slope. Therefore, any points above 5 MeV where the first derivative of the spline is positive are considered invalid and set to fill values in the FEDU_FIT variable.

This fit routine is independently applied to each of the 35 of the pitch angle values. An example of one of these fits at 90° is shown in Figure 7. The advantages of these spectra over the currently available instrument-specific data (shown in black) are readily apparent. This is particularly important near the instrument boundaries, where cross-calibration offsets can exist leading to inaccurate spectral shapes and features. Therefore, we strongly encourage any users interested in utilizing data from multiple ECT instruments to use these combined data products.

Details are in the caption following the image

Example of the spline fitting for 90° fluxes on RBSP-B. The black points show a simple 3-min average of the data with no additional processing. The blue, green and red points show the FEDU spectra, and the orange line show the spline fit.

6 Known Caveats

In this section, we describe some of the major known caveats of this data product at the time of initial release. For the most upto-date list of known caveats, particularly as the ECT data approaches its final data release, users should consult the accompanying documentation available on the RBSP-ECT website (http://www.rbsp-ect.lanl.gov/).

6.1 Inner Zone

As noted in Section 2, the MagEIS-HIGH and REPT fluxes are all set to 0 in the inner zone. This is done based on the results from (Fennell et al., 2015; Li et al., 2015) which showed there were no observations of >900 keV electrons in the inner zone. Recent results from Claudepierre et al. (2019) have shown limited intervals where low fluxes of >1 MeV electrons are observed, but these require special data processing and are therefore not included in the combined ECT electron product.

6.2 Pitch Angle Distributions

As described in Sections 2 and 13, the data is first fit in pitch angle using Legendre polynomials, then is fit in energy using a spline fit. These splines are created independently for each of the 35 pitch angle values. While they generally do a good job of maintaining the shape and magnitude of the pitch angle distribution, some small artifacts can be introduced. Typically, these artifacts are very small, but in rare cases they can be significant. Figure 8 shows an example of this. Therefore, studies that are interested in the details of the shape of the pitch angle distribution should not rely solely on the FEDU_FIT data and should also consult the FEDU spectra.

Details are in the caption following the image

Example of pitch angle shape differences that can exist between FEDU_FIT and FEDU.

6.3 Early Mission

The results shown in the previous sections are for observations made after September 2013, when most of the major changes to the instruments’ energy channels and operational modes were complete. Prior to this date, changes in the energy channel definitions, flux conversion factors and pitch angle sectoring can produce nonphysical features in the data. Therefore, data from prior to October 2013 should be used with caution.

7 Scientific Applications

The goal of creating this data set is to provide additional utility to the RBSP-ECT measurements. By providing a set of fit pitch angle distributions and pitch-angle resolved spectra, these data can be easily integrated with other datasets including Arase (Miyoshi et al., 2017), GPS (Morley et al., 2017), MMS-FEEPS (Blake et al., 2016) (LANL-GEO; Bame et al., 1993; Belian et al., 1992; Meier et al., 1996) and THEMIS (Angelopoulos, 2008). This easy integration is more important than ever as more studies require multipoint, multisatellite observations.

One particular science topic these data will contribute to is studies of wave-particle interactions in the radiation belts. Multiple types of waves including VLF Chorus (Thorne, 2010) and ULF waves (Ukhorskiy et al., 2005) operate on particular parts of the pitch angle distribution and influence electrons across wide energy ranges. This data set could be used to study how pitch angle distributions of different energy populations including the source, seed and core populations evolve in response to wave-particle interactions (e.g., Jaynes et al., 2015).

8 Summary

This paper describes a new data product combining the pitch angle resolved electron fluxes from all three of the ECT instruments. The goal is to produce a consistent, reliable and cross-calibrated data set, offering comprehensive electron spectra throughout the inner magnetosphere over a wide range of energies. The combined data set will bring additional utility to the ECT data and will be especially useful for researchers interested in studying effects which span two or more of the ECT instrument’s energy ranges. As the ECT data approaches its final data release, the combined data products will be updated and refined. Any updates to the techniques described here will be available on the ECT web page.

This was the second in a series of combined ECT data products that use the methodology and techniques described here. This data set will lead directly to the final data product in the series, the official release of ECT combined phase space density.

Acknowledgments

Processing and analysis of the ECT data was supported by Energetic Particle, Composition, and Thermal Plasma (RBSP-ECT) investigation funded under NASA’s Prime contract no. NAS5-01072. Work at the Aerospace Corporation was supported by the Sustained Experimentation and Research for Program Applications (SERPA) program.

    Data Availability Statement

    All RBSP-ECT data are publicly available at the Web site http://www.RBSP-ect.lanl.gov/