Lightning-induced relativistic electron precipitation from the inner radiation belt

Energetic particles trapped in Earth’s magnetosphere form the inner and outer radiation belts. The outer belt commonly hosts MeV electrons, while the inner belt typically contains 10s–100s keV electrons and energetic protons; the upper limit of inner-belt electron energies has been debated, with little MeV presence observed during parts of solar cycle 24 unless strong geomagnetic disturbances occur. Microbursts are rapid (sub-second) precipitation events that drive radiation belt losses and are usually associated with chorus waves in the outer belt. Lightning-generated very low frequency (VLF) whistler waves can drive lightning-induced electron precipitation (LEP) in the inner belt and slot region, but prior evidence at MeV energies has been largely indirect. The research question addressed here is whether lightning can directly drive precipitation of MeV electrons in the inner radiation belt and what the spatiotemporal characteristics of such events are. The study aims to identify and characterize MeV-energy bouncing microburst packets at low L-shells and to establish their connection to lightning activity and geomagnetic conditions.

Prior work has established microbursts as an important loss mechanism for outer-belt electrons across keV–MeV energies, often driven by interactions with whistler-mode chorus. LEP has been observed directly at 10s–100s keV and inferred at MeV energies via subionospheric VLF perturbations, though ionospheric recombination times limit temporal resolution, masking the fine structure of bouncing packets. Bouncing microburst packets have been modeled and occasionally observed as decaying sequences consistent with repeated magnetospheric bounces and progressive precipitation. Previous direct precipitation near L≈2 from ground VLF transmitters has shown energies up to ~700 keV, leaving a gap in direct evidence for MeV LEP. The spatial dependence of LEP, nightside preference due to reduced ionospheric attenuation, and longitudinal and hemispheric asymmetries tied to Earth’s magnetic field have been reported, but the temporal morphologies beyond simple decay signatures remain under-characterized.

Data source: 20 ms count rate data from the SAMPEX/HILT instrument were analyzed spanning Aug 7, 1996 to Aug 7, 2006. Event detection used two sequential algorithms in MATLAB. First, an adapted O’Brien et al. microburst detector was applied using the threshold (N100 − A500)/(1 + √A500) > 5, where N100 is the 100 ms count rate and A500 is the centered 500 ms running average; flagged indices within 200 ms were merged into one interval. Second, a custom bouncing-packet finder searched each candidate interval (extended by +0.2 s at the start and +1.0 s at the end) for local maxima (findpeaks) with prominence ≥ 0.25 of the count-rate range and minimum peak separation ≥ 0.06 s. Intervals were retained if they contained runs of ≥ 4 consecutive peaks with consistent spacing (variation < 0.15 of mean spacing), total duration < 15 s, and ≥ 10 unique count-rate values to avoid telemetry quantization artifacts. Events were limited to L-shell ≤ 3 (inner belt and slot region). Identified events were reviewed visually to exclude noisy or ambiguous cases. Event characterization: Background trends were removed by identifying local minima and interpolating between them; this background was subtracted to reveal the microburst packet shape, which was classified as decreasing, crown (increase then decrease), increasing, or other. Minimum peak spacing per event was used to estimate a characteristic electron energy via a dipole-bounce model (per Schulz & Lanzerotti), acknowledging uncertainty from the 20 ms instrument resolution. Example: at L=2 and equatorial pitch angle 15°, a 200 ms bounce implies ~1.4 MeV whereas 220 ms implies ~550 keV. Geomagnetic context: Hourly Dst (OMNI) values during events were compared to the full-decade distribution via a two-population t-test. The relationship between microbursts and geomagnetic storms (Dstmin ≤ −50 nT) was examined by histogramming delays between storms and microbursts versus inter-storm intervals. Lightning correlation: For events with footprints or conjugate mirror points over the contiguous U.S., National Lightning Detection Network (NLDN) cloud-to-ground data were analyzed. A success was defined as a lightning strike with peak current ≥ 100 kA occurring 0–3 s prior to microburst onset, consistent with expected ~1 s LEP delays from whistler and electron propagation. Conjugate points were computed with IRBEM. To test chance coincidence, the success proportion for U.S.-region events was compared with events over Africa (no NLDN coverage) using a two-population t-test. Spatial mapping: Geographic locations and magnetic field strengths (IGRF-13) were used to assess clustering and hemispheric preferences. Distributions of L-shell, MLT, number of peaks, and peak spacings were compiled.

• Identified 45 clear bouncing microburst events at L ≤ 3 over a decade of SAMPEX/HILT data; periodicities (~200 ms) matched expected 1 MeV electron bounce periods at L≈2 within 20 ms resolution limits.
• Temporal morphologies: approximately one-third crown-shaped, one-third decreasing, and the remainder increasing or other.
• Durations: detected packets lasted at least 0.96–2.9 s (lower limits based on detected peaks).
• Peak spacings: minimum spacing per event distributed approximately normally, centered around ~0.212 s.
• Energies: characteristic energies inferred from bounce periods spanned ~323 keV to 7.81 MeV; energy estimates are sensitive to small changes in observed period given 20 ms resolution (e.g., at L=2, αeq=15°, 200 ms ≈ 1.4 MeV vs 220 ms ≈ 550 keV).
• Spatial and local time distributions: events cluster near L≈2 and predominantly on the nightside; geographic clustering around the southern tip of South America and the eastern coast of South Africa, with two events in the northern hemisphere over the continental U.S.
• Hemispheric preference: predominance in the southern hemisphere is consistent with weaker magnetic field strengths there, allowing lower mirror altitudes (within/near the SAA), enabling detection at SAMPEX altitude.
• Geomagnetic context: events occurred during more disturbed conditions; mean Dst during events was −26.73 nT versus −16.40 nT overall, two-population t-test p = 0.0046. Kp and AE showed similar trends.
• Storm relation: microburst events tended to occur during or shortly after geomagnetic storms (Dstmin ≤ −50 nT) when the slot region was temporarily filled with MeV electrons; delay times between storms and microbursts were significantly shorter than inter-storm intervals.
• Lightning correlation: of seven events with U.S. or conjugate coverage, three showed nearby high-amplitude (≥100 kA) cloud-to-ground strokes within a few seconds prior to microburst onset, consistent with expected ~1 s delay for LEP. Statistical comparison of U.S. events to Africa events (no NLDN coverage) yielded a highly significant difference in success proportions (two-population t-test p = 1.16×10⁻⁶), supporting lightning as the driver.
• Related occurrences: multiple instances of paired events separated by ≤1 minute suggest clustered lightning sources (e.g., strokes within the same storm).

The observations constitute the first direct in situ evidence that lightning-generated whistlers can precipitate MeV electrons from the inner radiation belt near L≈2. This directly links terrestrial lightning activity to relativistic electron loss processes, bridging meteorological phenomena with space weather dynamics. The nightside occurrence and geographic/hemispheric distributions are consistent with LEP expectations given ionospheric attenuation and the asymmetric geomagnetic field, while the timing relative to geomagnetic storms confirms that slot-filling events provide the transient MeV source population at low L-shells. The diversity of temporal morphologies—beyond the classically observed decaying sequences—indicates significant spatial and temporal variability of precipitation patches and the spacecraft’s traversal relative to patch structure, with scales from hundreds to thousands of kilometers. Compared to prior direct VLF-transmitter-induced precipitation limited to ≲700 keV, these MeV observations extend the energy range for wave-driven precipitation near L≈2. The findings inform models of wave-particle interactions and radiation belt lifetimes and suggest that extreme lightning (superbolts) may drive particularly strong LEP.

This study provides the first direct observations of lightning-induced precipitation of MeV electrons from the inner radiation belt, identifying 45 bouncing microburst packets near L≈2 with nightside preference, southern hemisphere predominance, occurrence during geomagnetically disturbed periods, and statistically significant associations with high-amplitude lightning over the U.S. The results reveal a direct coupling between lightning and relativistic electron losses, expand the known energy range of LEP, and expose previously uncharacterized variability in microburst temporal morphologies. Future work should refine detection algorithms to improve event capture, leverage global lightning datasets for comprehensive correlation studies, and utilize forthcoming high time-resolution missions (e.g., IMPAX, RADICALS) to map the global occurrence, spatial structure, and energy dependence of LEP microbursts and quantify their contribution to radiation belt loss.

• Lightning data coverage was limited to the contiguous United States (NLDN), excluding global correlations; in-cloud discharges and very large strokes with complex waveforms may be missed or undercounted.
• Instrument time resolution (20 ms) introduces uncertainty in bounce period and inferred energy, making energy estimates sensitive to small timing errors.
• Strict event selection criteria and visual vetting likely led to undercounting of true events; short durations and localized precipitation patches further reduce detection probability.
• Hemispheric and geographic detection biases exist due to magnetic field asymmetries (e.g., SAA) and low-Earth-orbit sampling; dayside ionospheric attenuation reduces LEP wave penetration, skewing local time occurrence.
• The study period predated widespread availability of global lightning networks, limiting comprehensive source attribution.