Listening for events…
TerraPulse Investigation

What's Really in the Sky?

We cross-referenced 80,000 UFO reports against fireballs, geomagnetic storms, and satellite passes. The answer isn't aliens — it's science, observer bias, and a 10,000-satellite constellation that didn't exist yet.

TerraPulse Data Lab April 2026 7 peer-reviewed analyses
80,331
UFO sightings analyzed
24.1%
attributed to known causes
10,119
Starlink satellites tracked
0
confirmed bolide matches

Every year, thousands of people report seeing something unexplained in the sky. Fireballs streak across social media. YouTube videos warn of imminent asteroid impacts. But what does the data actually say?

At TerraPulse, we don't speculate — we measure. Over the past month, we've ingested NaNM observations from NASA, NOAA, USGS, and a dozen other scientific agencies, then applied rigorous statistical analysis to answer the question: What's really up there?

80,000 Reports. Three Filters. One Answer.

The National UFO Reporting Center (NUFORC) has collected over 80,000 geocoded sighting reports from 1906 to 2014. We ran every single one through a cascading attribution filter:

1
Self-classified fireballs: 4,946 reports (6.2%) where the witness said "fireball." Removed.
2
Geomagnetic storms: 13,134 reports (16.3%) coinciding with Dst < -50 nT storms, when aurora is visible at mid-latitudes.
3
NASA fireball catalog: Zero matches. CNEOS detects ~20 kiloton-class impacts per year globally — too sparse to correlate with 6,000 reports per year.

After all filters: 75.9% remain unexplained. But that doesn't mean they're anomalous — it means our three filters are limited. The unexplained sightings show a massive summer bias (July has 2x February's reports), consistent with people being outdoors more, not with any physical phenomenon.

NUFORC attribution cascade
Cascading attribution filter applied to 80,331 NUFORC sightings. Click for interactive version.
Full attribution cascade Sankey diagram
Full attribution cascade including satellite proxy estimates. Click for interactive Sankey.
The chi-squared test: χ2(11) = 3,466, p < 10-15. The seasonal pattern is overwhelmingly driven by observer availability, not sky events.

Are Fireballs Increasing? No.

Social media periodically erupts with claims of unprecedented fireball activity. We checked the NASA CNEOS catalog — the definitive, instrument-grade record of bolide impacts detected by government sensors.

Year Fireballs vs. Average
202422normal
202522normal
2026 (Q1)4-0.1σ

The rate is 20-24 per year, every year, like clockwork. The apparent "surges" on social media are driven by increased reporting and attention feedback loops, not more rocks. When the American Meteor Society gets more reports, it's because more people are looking up and filing — not because the sky changed.

The Starlink Problem (That Can't Be Tested Yet)

SpaceX's Starlink constellation — 10,119 satellites and counting — is almost certainly generating UFO reports right now. But we can't prove it with the NUFORC data because of a hard temporal gap:

NUFORC data ends May 2014. Starlink launched May 2019.
Zero overlap. Direct attribution is impossible by construction.

But we can see what happened with the previous satellite constellation. When Iridium launched 66 satellites in 1997, "formation" sighting rates jumped 1.6x (OR = 1.60, 95% CI: 1.32-1.94, p = 1.6 × 10-6). Satellite keyword mentions increased 2.9x.

Starlink has 153x more satellites than Iridium, in orbits visible to 99.4% of NUFORC reporting locations. When post-2019 NUFORC data becomes available, we expect a substantial fraction of "unexplained" sightings to be attributable to satellite trains.

Satellite mention rate in NUFORC sightings over time
Satellite-like reports in NUFORC data, 1990-2014. Note the jump after Iridium's 1997 deployment. Click for interactive.
Starlink constellation orbital shells
10,119 Starlink satellites by orbital inclination and altitude. The 53 and 43 degree shells cover nearly all NUFORC locations.

What About Actual Asteroid Risk?

We track 51,000+ near-Earth object close approaches via NASA's JPL. The vast majority are house-sized rocks passing at 10-50x lunar distance — routine and harmless. Today's closest approaches:

Typical daily close approaches (April 2026):

  • 2026 DA1 — 4.96M km (0.033 AU), ~3m diameter
  • 2026 DZ7 — 5.82M km (0.039 AU), ~10m diameter
  • 2026 EA1 — 7.24M km (0.048 AU), ~5m diameter

For context, the Moon is 384,000 km away. These objects pass at 13-19x lunar distance.

Real asteroid risk is monitored by NASA's Planetary Defense Coordination Office. The named watchlist objects (Apophis, Bennu, 2024 YR4) are tracked daily in our system. None currently pose a threat.

The Invisible Sky: Space Weather and HF Radio

While people watch for lights in the sky, the real action is invisible. Geomagnetic storms from solar coronal mass ejections hammer the ionosphere, degrading radio propagation on polar paths by 0.29 dB — a measurable, statistically significant effect we detected using 21 years of WSPR ham radio data.

Polar paths
-0.29 dB
p < 10-5, N=200 storms
Equatorial paths (control)
-0.06 dB
p = 0.24, not significant

The equatorial null validates the method: storms hit polar paths (through the auroral zone) but spare equatorial paths. The effect increases monotonically with storm severity — the dose-response curve that good science demands.

WSPR corridor SNR during geomagnetic storms
Superposed epoch analysis: POLAR paths (purple) degrade during storms while EQUAT (green) is unaffected. Click for interactive.

The Bottom Line

The sky isn't falling. Fireball rates are normal. Asteroid flybys are routine. The "unexplained" UFO reports are dominated by observer bias, satellite constellations, and aurora from geomagnetic storms.

What IS real — and genuinely fascinating — is the invisible electromagnetic environment above our heads. Solar storms reshape the ionosphere, cosmic rays anti-correlate with solar activity, and the magnetic poles are drifting at 50 km/year. These are the signals worth watching. And we're watching all of them, 24/7, across NaNM observations and counting.

Research Papers Behind This Article

All data and analysis scripts are open source at github.com/isenbek/terrapulse

Live Feed