A titanium-housed device on the seafloor off the Oregon coast is measuring the weight of the water above it with an accuracy that would be astounding even in a controlled laboratory setting, at a depth where sunlight hasn’t reached in geological time. It continuously records data via a fiber-optic cable that travels back to shore, where the readings are added to a stream of data that geodesists and seismologists use to watch one of North America’s most dangerous fault systems breathe in slow motion.
Stretching approximately 1,000 kilometers along the Pacific Northwest coast, the Cascadia Subduction Zone extends from Northern California thru Oregon, Washington, and British Columbia. It is the point where the Juan de Fuca oceanic plate plunges beneath the continent of North America. In certain areas, this process is currently locked, which means the plates remain trapped rather than moving past one another smoothly, creating tension that will eventually be released. According to Japanese historical records, the last time the strain was fully released throughout the entire length of the fault was in January 1700, which resulted in a megathrust earthquake with an estimated magnitude of 9.0 or higher and a wave that spanned the Pacific. According to the geologic record, these occurrences occur every 200 to 500 years. The Pacific Northwest has been silently coping with that interval’s math for a while.
Before the accumulated strain manifests as ground motion, the pressure sensors being placed along the Cascadia fault are attempting to interpret it. The idea is simple: the weight of the water column above a segment of the fault floor changes proportionately if it rises or decreases by even a fraction of a millimeter. These pressure variations can be detected with exceptional sensitivity by a piezoelectric sensor, which is based on quartz crystals that, when squeezed, produce a detectable electrical charge. This information can then be used to determine whether, in which direction, and at what rate the seafloor is deforming. Slow-slip events appear in the pressure record as a sequence of minute movements that would be missed by a surface seismograph. These events occur when the locked fault gradually and silently releases strain over days or weeks.
Doing this at depth presents substantial and unresolved engineering hurdles. Sensor drift is a recurring issue brought on by the same pressure that enables the measurements. Even the most reliable equipment eventually have systematic mistakes in their baseline readings at the pressures seen several kilometers deep; these inaccuracies can add up to about 10 centimeters annually. That may seem insignificant, but keep in mind that the tectonic signals that scientists are attempting to identify are measured in millimeters. The signal is smaller than the noise. In order to maintain the integrity of the long-term baseline, engineers are currently exploring pressure-referencing devices that use internal atmospheric standards in order to separate actual seafloor movement from instrumental drift.
Thru cooperation between USGS, American university research organizations, and the Ocean Networks Canada cabled observatory system—which operates sensors anchored to the seafloor and connected to onshore data centers via undersea cable—the Cascadia monitoring network has been expanding. The cable connection is important because wired connections provide near-real-time data at higher resolution, whereas acoustic data transfer thru seawater contributes its own flaws and latency. With nodes placed along important Juan de Fuca plate boundary segments, the NEPTUNE observatory is one of the most advanced seabed monitoring systems in use today worldwide.
The fact that seismologists explicitly acknowledge that precise earthquake prediction—a precise date, time, and magnitude—remains outside the purview of earth science is an honest constraint that remains unchanged by technological advancements. The pressure sensors can identify slow-slip events that some researchers think may precede larger ruptures, and they also enhance the image of where strain is building up and how quickly.
However, it is still unclear how slow-slip and megathrust earthquakes are related, and the data history is too small to confidently identify trustworthy precursor trends. The sensors can provide a more accurate picture of the fault’s current condition, including how locked, how strained, and how close it is to the circumstances that have historically preceded significant catastrophes. In a deterministic sense, that is not prediction. However, it’s far superior to the alternative, which is functioning without it.
