Standing on the South Jetty at the mouth of the Columbia River on a stormy day in January, a person would have to squint to make out details along the far shore. The Cape Disappointment lighthouse would flash a smeary streak of white through sheets of gray, horizontal rain. Then, with another gust, the winter gloom would close in again, making it difficult to see the state of Washington from the Oregon side. At this point, the river is four miles (6.4 km) wide.
A few hardy souls, storm watchers who love nothing more than the wet sting of salty wind on their cheeks, can nearly always be found here clinging to the guardrail around the viewing platform at Fort Stevens State Park, their exhilarated or foolish companions leaning into the gusts and grinning at the ferocity of it all. The Columbia River’s freshet fighting an incoming Pacific tide is an impressive sight.
Unlike most other big rivers, the Columbia has no delta to speak of, no meandering Mississippi mud flats, no maze of trackless swamps. Hemmed in by hard rock mountains that were pushed up by plate tectonics, the river chews like a chainsaw through steep, stony canyons and then gushes headlong into the open Pacific. Columbia’s estuary is more compact than a river this size would normally be. Much of the mud that might have formed a sprawling delta in less vertical terrain gets dumped almost directly into the sea.
Suspended in its swollen belly, the river carries silt, sand, and forest debris from 258,000 square miles (668,000 km2) of western mountains that are slowly dissolving under millions of years of rain, rock-splitting ice, and heavy snow. Along the way a host of smaller streams and rivers—the Kootenay, the Pend Oreille, the Snake, the Kicking Horse, the Kettle, the Cowlitz, and many others—add their contributions to the flow and the silt load. At Portland the Columbia inhales the Willamette River, swings north to find a gap between the Coast Range and Olympic Mountains, and then turns west one last time for its final plunge to the sea. The big river’s muddy torrent sweeps past the outport town of Astoria, dumping about 265,000 cubic feet (7,500 m3) of fresh water into the snarly North Pacific every second.
The ocean fights back with tremendous swells, pounding surf, and a tidal surge that will not be denied. At the point roughly five miles (8 km) offshore where these two irresistible forces meet and do battle—the infamous Columbia Bar—a never-ending churn between outgoing river and incoming tide, amplified by howling winds—creates standing waves as high as thirty feet (9 m). Having reached the horizontal plane of the sea after more than a thousand downhill miles (1,600 km), the river slows and begins dumping its cargo of suspended debris in a constantly shifting barricade of sandbars and shoals that make the mouth of the river one of the most hazardous places in the world for mariners. Some say nearly two thousand ships and boats have been wrecked along this stretch of the coast—nicknamed the Graveyard of the Pacific—over the past two centuries.
Once the Columbia has forced its way over the bar and out to sea, it veers north, pushed by the Davidson Current and huge Pacific storms. Along the southwestern shore of Washington State, the river dumps more of its load, giving birth to long spits of sand and narrow barrier islands parallel to the mainland with miles and miles of beautifully isolated beaches.
In satellite photos you can see a muddy plume swirling across the edge of the continental shelf before apparently disappearing into the ocean depths. Losing momentum, the Columbia finally drops what’s left of that pulverized mountain debris onto the sea floor, piling it up precariously at the head of a steep canyon. The finer grains of silt keep moving farther out to sea.
Reading two 1970 papers by LaVerne Kulm and Gary Griggs (an OSU graduate student when this story began), I learned why those canyons and channels cutting across the flat surface of the ocean floor looked so much like a big river and its tributaries meandering across the prairies. It’s because that’s exactly what they are. This spidery web of channels is basically an extension of the Columbia River system across the sea floor—through the mud on top of the Juan de Fuca plate.
When massive glaciers filled the mountain valleys and covered most of the Pacific Northwest more than ten thousand years ago, the level of the sea was roughly 425 feet (130 m) lower than it is now. With so much water trapped in glacial ice on land, the ocean shrank and the continental shelf was exposed as dry land cliffs. In those days the Columbia had to flow much farther west to reach the sea. It ran across the drying continental shelf, cutting a groove all the way to the outer edge. Then it sliced a canyon down the steep slope to the flat oceanic plate below.
As ice sheets came and went over millions of years and the level of the sea rose and fell, so the Columbia periodically changed course. When ice was thick in the mountains and the sea level was low, the river’s main current ran directly west across the exposed shelf and down to the sea through what is now known as Astoria Canyon. When the glaciers started melting again, gradually raising sea level and redrown-ing the canyon, the river’s current swung northwest, pushed by strong ocean currents and storms, and meandered across the reflooded continental shelf until it reached the edge again farther north, where it cut another groove down the continental slope at what is now called Willapa Canyon, off the coast of Washington.
During several of these big melting cycles, broken slabs of ice plugged a narrow valley in the upper Columbia basin near the Idaho–Montana border, creating an ice dam 2,500 feet (760 m) high that backed up an inland sea called Glacial Lake Missoula, as big as Lake Erie and Lake Ontario combined. Eventually the dam broke—twelve thousand years ago—releasing a catastrophic torrent of water and debris that roared down the channel cut by the river, spilling out to sea across the continental shelf, then gushing down Astoria and Willapa Canyons before spreading like thick mud soup across the Juan de Fuca plate.
That’s why the Cascadia basin—the spreading section of ocean floor otherwise known as the Juan de Fuca plate—came to look so much like an underwater prairie. It got buried repeatedly by glacial outwash and periodic floods until the ocean floor was 1.9 to 2.5 miles (3–4 km) deep in silt, sand, and debris.
When I read about this, I recalled the “no trench” part of Walter Sullivan’s story in the New York Times. Evidently the subduction zone was so full of Columbia sediment and Missoula mud that the first wave of geologists and oceanographers to explore the Cascadia basin with echo sounders could not see much of a trench where they thought one should have been. The early technology used for mapping the sea floor could not penetrate the mud to see the real basement: the boundary zone where the Juan de Fuca plate was dipping down and thrusting its way underneath North America.
With no trench obvious, Cascadia became “a special case,” unlike most other subduction zones. Add to this the absence of big temblors and it must have seemed like all the proof any reasonable person might need to be convinced Cascadia’s fault was aseismic—quake free and essentially harmless. But when Griggs and Kulm and the team at OSU began to investigate the web of riverlike channels crossing the deep ocean floor, the hypothesis of aseismic subduction started to unravel.
They used heavy piston rigs to gouge core samples that showed a long series of landslides of silt and sand (those turbidity currents that Goldfinger explained years later aboard the Roger Revelle) that had flowed down the canyons and been deposited on top of the older Missoula mud. One of those layers was the infamous Mazama ash, which allowed them to radiocarbon date all the other turbidites and calculate the average amount of time between landslides—which turned out to be roughly 550 years. The timing between slides seemed unusually consistent. Another coincidence? Not likely, yet nobody knew how to explain what appeared to be a recurring cycle. Thirteen turbidite landslides—roughly 500 to 600 years apart. Why?
The core samples also painted a vivid picture of what happened once these debris flows began to move downhill. Griggs and Kulm had calculated that each year about one million cubic feet (28,000 m3) of muck was being carried across the continental shelf by the Columbia and dumped at the head of Willapa Canyon. So a million cubic feet of this stuff p
iles up every year for five centuries—and then something makes it tumble.
The core samples traced the downhill flow of these currents and showed that some were so high and fast they had splashed over the walls of the main deep-sea channel and spread out sideways as much as 10 miles (17 km). The biggest flows were more than 325 feet (100 m) high and ran more than 400 miles (650 km) down the channel. At approximately 30 feet or more (10 m) per second, one of these swirling plumes would take nearly two full days to run its course—to run out of “downhillness,” as Chris Goldfinger had put it. All of which was amazing enough, although the key question—what had triggered the landslides so regularly for thousands of years—remained unanswered.
Griggs and Kulm offered two possible causes, “periodic earthquakes or severe storms.” They drew no conclusion of their own. As Goldfinger told their story years later to a newer generation of graduate students off the coast of Sumatra, nobody in 1970 would have believed the earthquake hypothesis because there had never been a big subduction shake in the Pacific Northwest in all of recorded history.
The logic sounded straightforward: if these kinds of megathrust events were possible, we would have seen one by now. Surely in 150 years of recorded history one of these monsters would have attacked. There was a well-accepted principle in geology called uniformitarianism, which held that “the present is the key to the past.” Geologic processes that we see happening now are the same processes that happened long ago. Therefore, if we see no great earthquakes in Cascadia now, this subduction zone has probably always been quiet.
Without more data, it was simply easier to believe that some howling great winter storms had triggered all those offshore landslides. But every 550 years? How could anything in nature be so apparently punctual? That part still rankled for those who were suspicious of coincidence. And there was one doubter in particular who just wouldn’t let it go.
Seismologist John Adams, whom I’d met at the Pacific Geoscience Centre on Vancouver Island in 1985 while filming my first earthquake documentary, already knew that plate boundaries could take several centuries to build up enough strain to rupture. Before moving to Cornell University in New York in the late 1970s to work as a postdoctoral research associate, he had completed a study of the Alpine fault, along the southwest coast in his home country of New Zealand. There, instead of subducting, or diving underneath, the Pacific plate was obducting—being forced over the top of the Indo-Australian plate.
Like Cascadia, there had been no major ruptures of the Alpine fault system in all of recorded history, which in this case amounted to roughly 150 years. There were plenty of signs, however, that land along the mountainfront had been folded and bent and was under extreme stress. Beaches on the Pacific plate had been pushed up into terraces as they had in Cascadia. The Southern Alps mountain chain on New Zealand’s southern islands had been uplifted as a result of compression between the plates, and in places Adams was able to study both vertical and horizontal displacements caused by earthquakes that happened centuries ago.
He noticed several other important things about the timing and the amount of movement along the Alpine system. Some—not all—of the built-up strain had been relieved during big ruptures that happened in the not-too-distant past, but there appeared to be “seismic gaps.” It was pretty obvious that parts of the fault were moving spasmodically in earthquakes. Other segments of the fault, however, showed no evidence of rupture and were either sliding along smoothly or had been stuck together by friction, building up stress for a long time, and they were probably ready to slip again in a big quake. Seismologists call this a “stick–slip” scenario.
Just as he would find years later in Cascadia, Adams learned that the science community was divided about the risk posed by the Alpine fault. Even though it was a “San Andreas–scale” crack in the crust, few seismologists had paid it much attention. “They didn’t see earthquakes,” said Adams. “Their seismic hazard analysis actually ignored it, basically. Whereas the geologists said, ‘This thing has moved recently. You can see the offsets and the other characteristics. And therefore, it has to be an active plate boundary and will generate great earthquakes.’”
The long gaps between rock- and landslides triggered by Alpine earthquakes were eerily similar to the intervals between the deep-sea landslides that Griggs and Kulm had found in the mud off the Oregon coast, so it’s easy to see why Adams was intrigued by their papers when he finally came across them. While Griggs and Kulm weren’t really looking for quakes (they had set out to study the structure and evolution of the deep-sea channel system), serendipity gave them data that would later play a pivotal role in the debate about Cascadia’s fault.
Adams, on the other hand, was definitely searching for seismic fingerprints—earthquake history—which is why he would eventually write to Oregon State University asking permission to examine the mud cores, data logs, and timelines compiled by Griggs and Kulm, Hans Nelson (who would later team up with Chris Goldfinger on a series of follow-up studies), and other members of the original OSU team. Adams wanted a closer look at the patterns. In the meantime, he drew a chilling conclusion about the Alpine fault zone in New Zealand.
While there was ample evidence of seismic activity to the north and south, there had never in recorded history been a major rupture along the central part of the fault zone. Now, with physical evidence from a series of dated landslides, Adams felt confident more of the same would occur. He wrote that large quakes with “a rupture length of 270 kilometres [168 miles], a maximum displacement of 9 metres [30 feet], and magnitudes of approximately 8 are indicated for the central part of the Alpine fault.” New Zealand, like Cascadia, was apparently locked and loaded for a major shockwave.
By the time his Alpine paper was in its final draft and being peer reviewed in the winter of 1979, Adams was already working in North America with a keen interest in the Cascadia Subduction Zone. The Alpine paper was published in a scientific journal called Geology only two months before the eruption of Mount St. Helens, and I suspect its significance—especially the parallels to Cascadia—may have been lost in that spectacular volcanic dust cloud.
Unfazed, Adams continued to work on other evidence that Cascadia was an active threat. He had ongoing battles to fight against conventional wisdom and the principle of uniformitarianism. While at Cornell he had begun working with a researcher named Robert Reilinger on a study of how much the Coast Range mountains of Washington and Oregon were tilting to the east. Like the work of Ando and Balazs, which had come out in 1979, this Cornell project involved new data from highway survey crews that showed a significant upheaval: a change of elevation along the entire western side of the mountain range.
In 1982 Reilinger and Adams took the Washington highway data from the earlier study and extended it southward by adding new measurements from five more resurveyed east–west highways crossing through the mountains in Oregon. They showed that survey markers located near the coast had been lifted up a noticeable amount in the less than eighty years since the last set of surveys. Tide gauge data along the coast showed pretty much the same thing; the beaches had been lifted as well. Put it all together and you got a picture of a mountain range about 370 miles long and 37 miles wide (600 km by 60 km) being hoisted up along its western edge and tilted, en masse, toward the east.
In this new mountain-tilting paper Adams and Reilinger drew attention to another worrisome study published only six months earlier by Jim Savage and a team at the USGS that showed land in the Puget Sound lowlands around Seattle apparently being compressed: squished together in a northeasterly direction. That was the same direction the Juan de Fuca plate was supposed to be moving. If the ocean floor was actively sliding underneath the continent and if the two plates were locked together by friction, this kind of compression, or “crustal shortening,” near Seattle was exactly what you’d expect to find. It was also exactly contrary to what Robert Crosson and Ando and Balazs had said earlier.
When Ando looked at the vert
ical shift of outer coast survey markers and the eastward tilting of the Coast Range, he and Balazs reasoned that the long-term, apparently quake-free uplift meant the two tectonic plates were not locked. That’s why Cascadia was seismically quiet. So how could one explain Savage’s new compression data? How does the ground squeeze together if the plates are not locked?
A series of measurements of the distances between geodetic survey monuments on opposite sides of Puget Sound, spaced six to eighteen miles (10–30 km) apart across the sound, revealed a surprising and somewhat baffling trend. Between 1972 and 1979, Savage and his colleagues used a Geodolite, a powerful and precise distance-measuring instrument that fired a laser beam from a survey marker on one side of Puget Sound across to a similar marker on the other side. There, a bank of highly polished mirrors bounced the laser back to the Geodolite, which measured how long the beam took to make the round trip.
If it took less time in 1979 than it did in 1972 the two markers had to be closer together, and that’s exactly what they found. They figured the accuracy of the Geodolite was within 0.2 inches (5 mm) and that the amount of squeezing of the valley floor was statistically significant. Savage and his coauthors (Mike Lisowski and Bill Prescott) recognized that their new data were “not easily reconciled” with Ando’s aseismic subduction concept, but they published them anyway. They concluded that the laser measurements were evidence of strain building up, probably caused by the subduction of the Juan de Fuca plate underneath North America.
Cascadia's Fault Page 12