CNEOS fireball day-of-year vs. meteor-shower calendar (1998–2026)

Author: Claude (TerraPulse Lab) Status: Complete Created: 2026-05-20 GitHub Issue: #205

Hypothesis

Do CNEOS-cataloged atmospheric fireballs cluster at the calendar peaks of major meteor showers (Perseids, Geminids, Leonids, …), or is the kt-energy fireball flux dominated by the sporadic background?

• H1: At least one named shower shows a statistically significant rate elevation at Bonferroni-corrected α=0.05 across 12 tested showers.
• H2: The shower contribution to total fireball counts is small (< 25%).
• Null: Fireball day-of-year is consistent with a uniform Poisson process.

Data Sources

Source	Records (raw)	Records (deduped)	Span
CNEOS neo_fireball	1,416	355	1998-01-04 → 2026-05-10

The 1,416 → 355 reduction is fetcher re-ingest duplication (identical timestamp_utc). The pre-registration in #205 quoted N=1,394 from the raw count; the correct unique-event N is 355, confirmed by COUNT(DISTINCT timestamp_utc) = 355.

Annual counts span 1 (1998) to 28 (2024); median is 15/yr. The catalog grew once US Government Sensors program coverage improved post-2014.

Methodology

1. Deduplicate to unique (timestamp_utc, lat, lon) events.
2. Compute day-of-year for each event (1–366).
3. Schuster test on phase = (DOY − 1)/365.25 — global test for any periodicity.
4. Per-shower Poisson rate-ratio: observed count in circular ±5d window around each peak vs. expected under uniform null (N × 11 / 365.25). Two-sided exact Poisson p-value. Bonferroni correction across 12 showers (threshold p < 0.00417).
5. Sensitivity sweep at half-widths {3, 5, 7, 10} d.
6. Analytic detection floor: minimum rate-ratio that would clear Bonferroni at our N.
7. Sporadic fraction: 1 − excess events in flagged-shower-window days / total events.

Showers tested (peak DOY): Quadrantids (3), Lyrids (112), η Aquariids (125), Arietids (158), S. δ Aquariids (211), Perseids (224), Orionids (294), S. Taurids (309), N. Taurids (316), Leonids (321), Geminids (347), Ursids (356).

Findings

Null result is clean

• Schuster global test: D²=536.7, p=0.22, R=0.065, N=355. Day-of-year is consistent with a uniform distribution.
• All 12 per-shower tests fail Bonferroni at every window width (3, 5, 7, 10d). At ±5d, the best raw p-value is Perseids p=0.0904 — a DEFICIT, not an excess (5 observed vs. 10.7 expected, rate ratio 0.47).
• Geminids: 10 obs vs. 10.7 expected (ratio 0.94, p=0.88). The shower most often cited as the likeliest source of fireball clustering shows no detectable excess in the CNEOS kt-energy regime.
• Sporadic-fraction estimate: 100%. The 12 showers' 120 flagged days contain 112 fireballs vs. 116.7 expected under uniform — a small negative excess.

Detection floor (analytic)

At N=355 and 11-day windows, the minimum observable rate-ratio that clears Bonferroni-12 at α=0.05 is ≥1.96 (must observe ≥21 events in any 11-day window). The observed shower ratios range 0.47–1.31. The Geminid excess reported in optical bolide literature (~30–50%, i.e., ratio 1.3–1.5) is below our detection floor at this N. A non-detection here is consistent with the kt-energy population being mostly sporadic, AND with our N being too small to detect optical-scale shower contributions even if they exist in this energy regime.

Mechanism: CNEOS is not a meteor-shower sampler

The result is not surprising once the population is understood. CNEOS catalogues events bright enough to be detected by US Government Sensors — typically impact energies ≥0.1 kt and bolide masses ≥ tens of kg. Visual meteor showers are dominated by mm- to cm-scale cometary dust burning up near 100 km altitude; these particles produce visible meteors but almost never reach the kt-energy fireball regime. The Geminids (parent body 3200 Phaethon, asteroidal) do contribute occasional fireballs in visual surveys, but the rate is well below our Bonferroni detection floor at N=355. This population-mismatch is the same one discussed by Brown et al. (2013, 2016).

Limitations and V2 plan

• N=355 is fixed for this revision. The catalog accretes ~25/yr in the current era; by 2034 a V2 at N≈600 could probe rate-ratios as low as ~1.6 for the strongest candidates.
• The deduplication groups by minute-resolution timestamp + 0.1° coordinates. If two genuinely independent fireballs occurred in the same minute at the same 0.1°×0.1° cell, this would collapse them — but CNEOS rates make this collision rate negligible.

References

• Brown et al. (2013), "A 500-kiloton airburst over Chelyabinsk and an enhanced hazard from small impactors," Nature 503, 238–241.
• CNEOS Fireball Database: https://cneos.jpl.nasa.gov/fireballs/
• Schuster (1897), "On lunar and solar periodicities of earthquakes," Proc. R. Soc. 61, 455–465.
• IAU Meteor Data Center: meteor shower nomenclature and peak dates.