The cave didn’t look that promising from the outside, Charlie Rubin remembers. But when the earthquake geologist and his colleagues walked in and started digging, “our jaws dropped,” he said. The researchers noticed that a depression in the floor of the cave—near Banda Aceh, Indonesia—contained distinct stratigraphy: dark layers of organic material separated by clearly defined layers of lighter-colored sand.
“We looked at each other and wondered if the sand was tsunami sand,” Rubin said. After closer examination, the team members realized they had found a natural record of tsunamis sweeping sand repeatedly into the cave over thousands of years. By radiocarbon dating the sandy layers, the researchers were able to achieve what’s often thought of as a holy grail in tsunami science: a reconstruction of when previous tsunamis occurred thousands of years in the past.
Rubin and his colleagues showed that at least 11 tsunamis had swept over the region over a span of about 5,000 years. But the massive waves were not regular in time: Periods of calm ranged from millennia to merely decades. This finding—that tsunami recurrence intervals are highly variable—is proof that regional hazard mitigation plans should be based on the high likelihood of future destructive tsunamis rather than estimates of recurrence intervals, the team suggests. That’s particularly important in the Indian Ocean, a region that’s prone to megathrust earthquakes and, accordingly, large tsunamis. Those massive waves include the deadliest tsunami in history, which was unleashed in 2004 not far offshore from where the cave is located and which killed more than 200,000 people.
The geological record contained within the Banda Aceh cave is “extraordinary,” said Brian McAdoo, a tsunami scientist at Yale-NUS College in Singapore. This study also represents the first time that cave data have been used to measure tsunami recurrence intervals, McAdoo said.
A Layer Cake
In 2011 and 2012, Rubin and his colleagues excavated six trenches at the rear of the 120-meter-long coastal cave. Beneath a crust of sand topped with bat guano they dug into alternating layers of sand and organic material that reached depths of 2 meters in some places. The scientists carefully collected tiny pieces of charcoal and shells from the layers and radiocarbon dated the material in the laboratory. Using these radiocarbon measurements, the team calculated the most likely age of each of the 11 buried layers of sand and therefore the approximate date of each tsunami.
The researchers found that the 11 sand layers spanned roughly 4,500 years, from about 7,400 to 2,900 years ago. However, the guano-encrusted 12th and uppermost sand layer—which contained shreds of clothing, suggesting it was deposited very recently—differed from the stack of alternating deposits beneath it: Its bottom face was jagged and irregular, unlike the smooth boundaries between the deeper layers.
The scientists suspect that this irregularity resulted from powerful waves from the 2004 tsunami triggered by the Sumatra-Andaman earthquake sweeping into the cave, scraping away previously deposited material, and literally erasing the geological record laid down after 900 BCE. The older layers of sand were probably never disrupted in a similar way because they’re located in a natural depression in the cave, Rubin said. “They’re packed down and they’re protected.”
The research team reported its findings last month in Nature Communications.
Far from Constant
Rubin and his team showed that the time span between successive tsunamis is far from constant: Although 10 intervals within 4,500 years breaks down to an average of 450 years between the events, the researchers found evidence of one 2,000-year period free of tsunamis and also a single century that saw four tsunamis. “This study provides new evidence that tsunami recurrence can be highly variable,” said Katrin Monecke, a geoscientist at Wellesley College in Wellesley, Mass.
The researchers, who included experts in earthquake science, hypothesized that the thickness of each sand layer reflects the magnitude of the tsunami-causing earthquake because a larger earthquake would produce a larger tsunami and therefore plausibly transport more sand into the cave. According to this theory, the thickest sand layer, measuring roughly 25 centimeters, should correspond to the strongest earthquake that occurred within the nearly 5,000 years of history recorded in the cave.
The scientists inferred that no tsunamis occurred for more than 2,100 years after this thickest layer of sand was laid down. This extremely long interseismic gap is consistent with a period of reduced stress along faults—and therefore of a lower probability of another quake—after a massive temblor released a large amount of energy, the team suggests. Conversely, the researchers found that the four sand layers corresponding to the four tsunamis that occurred within 100 years of each other were all thin (fewer than 10 centimeters), which makes sense, they argued in their paper, because short interseismic periods are consistent with weaker earthquakes.
Rubin said he and his colleagues hope to find additional caves containing evidence of past subduction zone earthquakes. Although other complementary techniques exist for determining that tsunamis occurred in the past, for example, oral histories and chemical analysis, Rubin and his team are excited to literally dig into the past. “The only way to get at tsunami older than historical records is with geology,” Rubin said.