Earth's Bombardment History: The Complete Impact Catalog

Start before the catalog, before there was a stable Earth to catalog. The dominant scientific model for how the Moon came to exist is the Giant Impact Hypothesis, and it is a violent one. Around 4.51 billion years ago a Mars-sized protoplanet, nicknamed Theia and roughly 6,000 kilometers across, slammed into the proto-Earth at high speed, releasing energy on the order of 10^32 joules and hurling a disk of vaporized and molten rock into orbit. The Moon condensed out of that disk. The idea was first floated by Hartmann and Davis in 1975 and by Cameron and Ward in 1976, became the field's preferred model after a 1984 conference in Kona, and is held up today by the isotopic chemistry of the Apollo lunar samples, by computer simulation, and by the peculiar angular momentum of the Earth-Moon system. Read the tier with care. This is Tier 1 as the consensus best-available explanation, and the research file is scrupulous about the next words: consensus, not proven. Real problems remain, and the very next claim is one of them.

The model keeps getting refined, and the refinements keep arriving. In 2022 a high-resolution simulation study (Kegerreis and colleagues, in The Astrophysical Journal Letters) ran the collision with more than 100 million particles and found the Moon may have assembled within hours of the impact rather than over the months or years earlier models assumed, producing a simulated body with an iron-poor exterior and a partly molten interior that match the real Moon's properties well. It is a genuine, published result and it is widely discussed, but the research file is honest about its standing: it is still being tested at higher resolution, promising rather than proven.

The most startling recent idea is that pieces of Theia may still be inside us. In 2023 a team led by Qian Yuan published a study in Nature proposing that two continent-sized regions of dense, anomalous rock deep in Earth's lowest mantle, the Large Low-Velocity Provinces that seismologists map sitting on the core beneath Africa and the Pacific, each about twice the mass of the Moon, are surviving iron-rich remnants of Theia's own mantle. Rather than mixing evenly into Earth, the idea goes, that foreign material sank and solidified down there. The team's simulations produced two blobs whose size and position roughly match the real seismic anomalies. If it holds, identifiable fragments of the world that made the Moon are still buried inside this one. It is a single peer-reviewed paper with a specific mechanism, and those deep-mantle blobs have several competing explanations, so it sits at Tier 2, not higher.

And the impactor itself is getting a home address. A study published on November 20, 2025 in Science, credited to researchers at the Max Planck Institute for Solar System Research and the University of Chicago, used chemical modeling to reconstruct Theia's likely original composition and concluded it probably formed in the inner solar system, closer to the Sun than Earth's own orbit. That narrows where the Moon-forming body came from without disturbing the core impact model. We flag it at Tier 2 for its recency, and note plainly that this file could pin the study down only through its science-news coverage, not the full primary paper.

For the first half-billion years after the Moon formed, the inner solar system was a shooting gallery, and here the honesty has to be careful. The research file describes a Late Heavy Bombardment, a roughly 300-million-year stretch from about 4.1 to 3.8 billion years ago when Earth and its neighbors were struck by a dramatically elevated rate of impactors, some models proposing more than 22,000 craters larger than 20 kilometers on Earth alone, driven by the migrating orbits of the giant planets and possibly sterilizing the surface more than once. That the early Earth took a far heavier pounding than the modern Earth is not in dispute. Whether it came as a single sharp cataclysm spiking around 3.9 billion years ago, which is how the older framing reads, or as a longer and more gradual decline across hundreds of millions of years with no distinct spike, is a live and unresolved argument in current planetary science, complicated by how genuinely hard the individual lunar basins are to date. The elevated bombardment is Tier 1. The cataclysm shape of it is Tier 2, and openly contested.

The oldest direct evidence of those early strikes on Earth's own surface, rather than inferred from the Moon's cratering, is written in South Africa. The Barberton Greenstone Belt preserves three distinct spherule beds, layers of tiny glass droplets that formed when impact-vaporized rock condensed and rained back down, dated to roughly 3.47, 3.26, and 3.24 billion years ago. The largest, the S4 layer, points to an impactor estimated between 20 and 58 kilometers across, a body that would make anything in the human-era catalog look like a pebble. This is Tier 1.

The oldest confirmed crater still readable on Earth is Yarrabubba, in Western Australia, radiometrically dated by Erickson and colleagues in a 2020 Nature Communications study to 2,229 million years ago, roughly 30 kilometers across. Its date is more than a curiosity. The impact struck during a global glaciation, a Snowball Earth, and may have helped end it, blasting enough water vapor and carbon dioxide into the atmosphere to warm the planet back out of the ice. It is the first solid evidence of an impact tied to a climate recovery rather than a climate catastrophe: a strike that may have thawed the world instead of freezing it.

Around 470 million years ago the sky itself changed for a while. A roughly 200-kilometer asteroid broke apart out in the asteroid belt, and for several million years afterward the debris raised the meteorite flux reaching Earth roughly a hundredfold. We can hold the proof in hand: fossil meteorites from that very breakup are found embedded in Swedish limestone. The event, the breakup and the elevated flux, is Tier 1. The tempting further claim, that the resulting dust cooled the climate and helped trigger the Great Ordovician Biodiversification Event, a major expansion of marine life, is proposed but unproven, and sits at Tier 3.

One deep-time crater is worth visiting for what it made. The Popigai crater in Siberia, about 100 kilometers across and 35.7 million years old, turned the graphite in its target rock into industrial diamonds under the impact's heat and pressure, across an area so wide the deposit is estimated in the trillions of carats, reportedly more than every other known diamond deposit on Earth combined. The stones are too small and flawed for jewelry but ideal for cutting and grinding, and the Soviet Union kept the whole deposit secret for decades. An asteroid strike left behind the largest diamond field on the planet.

Sixty-six million years ago the catalog produced its most consequential entry. The Chicxulub impact struck the Yucatan Peninsula of Mexico 66.043 million years ago, give or take about 11,000 years, gouging a multi-ring crater about 180 kilometers across, now buried under 600 to 1,100 meters of younger limestone. The impactor was a carbonaceous chondrite asteroid roughly 10 to 15 kilometers wide, confirmed by chromium isotope analysis, and it came in steep, at about 60 degrees from horizontal (Collins and colleagues, 2020) and around 20 kilometers per second, releasing energy on the order of 4.2 x 10^23 joules. It was found almost by accident: the geophysicists Antonio Camargo and Glen Penfield spotted the buried structure during a 1978 PEMEX oil survey, and Alan Hildebrand tied it to the extinction boundary in 1991. You can still trace its rim today as a ring of cenotes, water-filled sinkholes, curving across the Yucatan.

The killing came in stages, on wildly different clocks. Within minutes, seismic shaking equivalent to a magnitude 9 to 11 earthquake and vast landslides. Within hours, a pulse of heat from re-entering ejecta and global wildfires, recorded now as a worldwide layer of soot and charcoal, and megatsunamis over 100 meters high across the Gulf of Mexico. And then the long killer: an impact winter, as sulfur aerosols and dust blocked an estimated 80 to 90 percent of sunlight for months to years, dropping global temperatures by roughly 10 degrees Celsius and starving the food web from its photosynthetic base upward. One detail of location made it far worse. The Yucatan bedrock was unusually rich in sulfates, so the impact vaporized much more climate-forcing sulfur than a strike into granite or basalt would have. The staged mechanisms and the global soot layer are Tier 1; the exact percentage of sunlight lost and the exact degrees dropped are debated within a credible range.

We can read the first hours of the aftermath almost like a diary, because in 2016 a drilling expedition (IODP Expedition 364) cored straight into the crater's peak ring, and the 2019 analysis (Gulick and colleagues, in PNAS) recovered a single preserved day. The core runs from impact melt rock, up through sediment dumped by tsunami backwash, up into fine ejecta that settled out of the sky. Within minutes, rock from more than 10 kilometers down had been excavated and flung out. Within hours, tsunami waves over 100 meters high struck the Gulf and re-entering debris heated the surface to hundreds of degrees. Within a day, soot from global fires and a heavy load of dust and sulfur had reached the stratosphere worldwide. It is, in the study's own phrase, the first day of the Cenozoic, held in a few meters of rock.

There is a genuine complication, and the file does not hide it. As Chicxulub hit, India was already on fire. The Deccan Traps, a flood-basalt province covering some 500,000 square kilometers, erupted from roughly 250,000 years before the impact to about 500,000 years after it, venting its own climate-forcing carbon dioxide and sulfur dioxide. The mainstream press-pulse model (Arens and West, 2008) casts the Deccan volcanism as a long ecological press that had already stressed the world's ecosystems, with the sudden impact as the acute pulse that pushed them over; 2019 studies (Schoene and colleagues; Sprain and colleagues) even found the eruptions accelerated just after the impact, as if its seismic energy stoked the volcanism. A minority position, most associated with the geologist Gerta Keller, argues the volcanism was the main killer and the impact secondary. The research file names this a minority view against the weight of the evidence, with the consensus holding that both mattered but the impact pulled the trigger. It is Tier 2, honestly contested.

Move forward to the window where humans and impacts overlap, and the catalog gets vivid. The best-preserved crater on Earth is Barringer Crater in Arizona, widely called Meteor Crater, about 1.2 kilometers across and roughly 50,000 years old, punched out by an iron-nickel meteorite about 50 meters wide. There is no scientific dispute about its origin at all, which is exactly why it became the textbook crater against which more eroded, more arguable structures are measured.

Barringer has company, all of it confirmed, some of it recent enough that people were certainly alive nearby. Odessa in Texas, about 49,000 years ago, a 170-meter crater. Lonar Lake in India, about 39,000 years ago, a 1.8-kilometer crater blasted into Deccan basalt. Henbury in Australia, about 5,000 years ago, a field of 13 or 14 craters. Morasko in Poland, about 4,700 years ago, seven craters. Campo del Cielo in Argentina, about 4,500 years ago, more than 26 craters and a recovered iron mass of 37 tonnes, among the largest single meteorite masses ever found. Wabar in Saudi Arabia, about 3,000 years ago, craters partly filled with impact glass. And Whitecourt in Canada, only 36 meters across and only discovered in 2007, dated to around 1,000 CE. These are Tier 1, every one.

One of these deserves its own line, because it is where a real crater collides with a famous story. Kaali, on the Estonian island of Saaremaa, is a genuine, confirmed field of nine craters, the largest about 110 meters wide and 22 meters deep with a lake in its floor, formed when a small meteorite, 4 to 8 meters across, fragmented on the way down. The crater is Tier 1. The date has quietly moved, though: where the research file places the impact around 4,500 years ago, the best current radiocarbon dating of the crater's own formation puts it between about 1530 and 1450 BCE, roughly 3,500 years ago, close to a thousand years more recent. Hold that corrected date, because the story built on Kaali is the doorway into the trickiest section of this whole catalog.

Start with the real thing. There is a legitimate, peer-reviewed field of cultural astronomy studying cases where Indigenous Australian oral tradition preserves verifiable knowledge of actual impact craters. Duane Hamacher, now Director of the Burrin Program for Cultural and Indigenous Astronomy at the University of Melbourne, and his colleagues have documented traditions, art, and named sites tied to at least four real craters: Gosses Bluff, Henbury, Liverpool, and Wolfe Creek. The Wolfe Creek crater, about 880 meters across, is known in the Djaru language as Kandimalal and described in living tradition as the work of a falling star or a serpent; one recorded account from the Djaru elder Jack Jugarie tells of the evening star coming down to the ground in an explosion, a flash, a dust cloud, and a great noise, exactly what a witnessed or long-transmitted memory of an airburst would sound like. And Hamacher's own work carries a crucial discipline built in: he found that widely repeated Hopi and Navajo oral traditions about Barringer Crater were actually an early-twentieth-century media invention, not authentic Indigenous knowledge, a warning against swallowing every claimed ancient crater myth whole. This is Tier 2, credible, precisely because the field polices itself this way.

Now the popular guess, and note how the tier drops. The Estonian scholar and former president Lennart Meri proposed, in his book Hobevalge, that the Kaali impact was witnessed and encoded into regional oral tradition, including the fire-from-heaven imagery in runes 47 through 49 of the Finnish national epic, the Kalevala. It is a genuinely lovely idea and it has real cultural staying power. It also has no independent evidence linking the meteorite to those specific verses, and current scholarship says so directly. The corrected dating actually helps the general plausibility, since 1530 to 1450 BCE places the impact within human memory in the region, but plausibility is not evidence, and the specific Kalevala connection stays an unproven correlate at Tier 4, not the equal of the Australian work above.

A larger version of the same temptation runs across the whole sky. The Taurid Complex is a real, debated stream of debris, including Comet 2P/Encke, which at 3.3 years has the shortest orbital period of any known comet, and the twice-yearly Taurid meteor showers Earth crosses each June and November. The astronomers Victor Clube and Bill Napier proposed, in their 1982 book The Cosmic Serpent and in work extended by Asher and Steel in 1998, that all of it came from a single giant comet, 100 to 200 kilometers across, that entered the inner solar system 20,000 to 30,000 years ago and has been fragmenting ever since. That much is Tier 2, a published and argued hypothesis in cometary astronomy. But Clube and Napier went further, suggesting the world's sky-serpent and dragon traditions, Chinese flying dragons, the Mesoamerican feathered serpent, the Norse Midgard Serpent, Babylonian Tiamat, might all encode ancestral memories of a fragmenting comet overhead. That last leap is Tier 5, mythological: the research file itself offers it only as a possible reading, never a demonstrated link, and it belongs in a different column from the astronomy that carries it.

And now the claim that has to be named as withdrawn, not merely doubted. In 2021 a high-profile paper (Bunch and colleagues, in Scientific Reports) argued that a Tunguska-scale cosmic airburst destroyed the Middle Bronze Age city of Tall el-Hammam in the Jordan Valley around 1650 BCE, citing shocked quartz, melted pottery and mudbrick, and surface temperatures above 2,000 degrees Celsius, and framed it explicitly as a possible physical basis for the biblical destruction of Sodom. On April 24, 2025, Scientific Reports formally retracted the paper, after critics (Jaret and Harris; Boslough and Bruno) identified errors in its mineralogical and geochemical analyses and showed that its comparison to Tunguska leaned on outdated, overstated figures for Tunguska's own blast. Several of the original authors disagreed with the retraction and say they intend to republish with new data, but as this file is written the claim has no standing peer-reviewed publication behind it. That is what an honest Tier 4 looks like: not the site or its destruction in question, but this specific cosmic-airburst explanation pulled from the record by the journal that printed it.

How does anyone confirm that a battered ring in the ground was made by the sky and not by a volcano or a fault? Impacts leave diagnostic fingerprints that nothing else on Earth can fake. Moldavites, the green tektite glass thrown from the Ries crater in Germany 14.8 million years ago and scattered more than 200 kilometers, have been worn as gemstones for centuries. Libyan Desert Glass, a 98-percent-silica glass about 29 million years old from a source crater still not pinned down, was carved into the centerpiece scarab of a pectoral buried with Tutankhamun. Shocked quartz and maskelynite form only above pressures of 10 gigapascals, reached in nature only by hypervelocity impact, and shatter cones, distinctive nested fracture patterns, are similarly unique to impacts. These, together with the iridium anomaly behind the Chicxulub story, are the evidence that turns a suspicious circle in the landscape into a confirmed scar. This is standard, uncontested Tier 1 geology, and it is the reason the rest of the catalog can be trusted at all.

Step back and the honest scale of the whole catalog comes into focus. The Earth Impact Database at the University of New Brunswick currently lists over 200 confirmed craters worldwide, and they are wildly unevenly spread: 59 in North America, 44 in Europe, 27 in Australia, 20 in Africa, 19 in Asia. That distribution is not a map of where impacts happened. It is a map of where stable, well-studied, well-exposed old rock has survived for craters to be found in. Erosion, ocean-floor recycling, ice, and vegetation have erased or hidden the rest, which is almost all of it. Over the planet's history Earth was struck many thousands of times; a few hundred scars are what is left to read.

Several proposed giants remain unconfirmed, and the file keeps them at arm's length. Wilkes Land in Antarctica is a proposed 480-kilometer structure known only from a gravity anomaly under the ice sheet, which if it were ever confirmed would be the largest impact structure on Earth. Bedout, off Western Australia, is a proposed 250-kilometer structure once tied to the Permian-Triassic extinction, a link widely criticized since. Shiva, off the Mumbai coast, is a proposed 500-by-400-kilometer feature once floated as a second K-Pg impact, which most geophysicists now read as volcanic or tectonic instead. And the Burckle Crater, a proposed 29-kilometer submarine structure in the Indian Ocean dated to about 3,000 BCE and offered as the source of a Holocene mega-tsunami, is likewise unconfirmed, held by the research file at Tier 2 to 3, not the Tier 1 of the real craters above. All of these are speculative until the diagnostic evidence of the previous section turns up. They are candidates, not entries.

Finally, the catalog does not stop at what has hit; it includes what has come close and what is still out there. The full story of the rocks at human scale, the ones we have filmed, the missions that now track them, the asteroid we deliberately shoved onto a new path, and the Apophis scare that flared and vanished, lives in the sibling file next door and is not repeated here. What belongs to the deep-time record is the wider tally of close passes and long-odds risk. In 1989 the asteroid 4581 Asclepius crossed within 690,000 kilometers of Earth, through the very point in space Earth had occupied only about six hours earlier. On January 26, 2023, the object 2023 BU passed just 3,600 kilometers above the surface, closer than the geostationary satellites and the nearest non-impacting approach ever recorded. And 3122 Florence, about 4.4 kilometers across, is the largest asteroid ever tracked passing this near, about 7,066,000 kilometers out in September 2017. For the long horizon, 101955 Bennu carries an estimated cumulative impact chance of about 1 in 2,700 through the year 2182, the highest of any known object, and (29075) 1950 DA, about 1.1 kilometers wide, about 1 in 8,300 for the year 2880, a figure that hinges on the Yarkovsky effect, the faint thermal nudge that can shift an asteroid's orbit over centuries. These are Tier 1: published NASA and JPL assessments and observed records.