The whisper before the roar
The solid ground beneath our feet is in constant, silent motion. Tectonic plates grind against each other, building stress that must eventually be released. Sometimes, before the violent rupture of a major earthquake, the fault shows subtle signs of the immense forces at play. These are precursory slips: small, often imperceptible movements that can precede a larger event. They come in several forms, from tiny foreshocks—small quakes that occur before the mainshock—to slow slip events, which are gradual movements along a fault that unfold over days, weeks, or even months.
One of the most intensely studied regions for these phenomena is the Cascadia Subduction Zone, stretching 1,000 kilometers from Northern California to Vancouver Island. Here, the Juan de Fuca plate slides beneath the North American plate. Instead of locking completely until a massive magnitude 9.0+ earthquake occurs (the last one was in 1700), parts of the fault undergo regular slow slip events. These events, also known as episodic tremor and slip (ETS), can release the same amount of energy as a magnitude 6.8 earthquake, but over a period of weeks instead of seconds. They are detected not by shaking, but by networks of high-precision GPS stations that measure millimeter-scale movements of the Earth's surface. A recent global analysis of GPS data from 90 earthquakes of magnitude 7.0 or greater revealed a subtle but consistent signal: an exponential acceleration of fault slip beginning about two hours before the main rupture.
Other potential precursors include changes in groundwater levels or chemistry. As stress builds in the crust, it can open or close tiny fractures in the rock, altering how water and gases move underground. This can lead to measurable changes in the concentration of dissolved radon gas, a radioactive element released from rocks. Spikes in radon concentration have been documented before some earthquakes, as the stressed rock releases more of the trapped gas.
The prediction problem
The existence of these precursors led to a period of optimism in seismology. The most famous case is the 1975 Haicheng earthquake in China. Based on a long-term forecast, months of unusual ground tilt, changes in water wells, and a dramatic increase in foreshock activity, authorities ordered the evacuation of the city of one million people just hours before a magnitude 7.3 earthquake struck. This action is credited with preventing an estimated 150,000 fatalities and injuries.
However, the Haicheng success is a unique event. The following year, the devastating Tangshan earthquake, which caused an estimated 250,000 fatalities, occurred with no warning or observable precursors. This showed the central challenge: precursory signals are not consistent. A study of Southern California earthquakes found that 72% of magnitude 4.0 or greater events were preceded by foreshocks, but only about 5% of all small earthquakes are actually foreshocks that lead to a larger event.
The Parkfield Earthquake Prediction Experiment was a major scientific effort to capture an earthquake in the act. Based on a series of magnitude 6 earthquakes that occurred on the San Andreas Fault near Parkfield, California, with quasi-regularity (about every 22 years), the USGS predicted the next one would occur between 1985 and 1993. An extensive network of instruments was deployed to catch any precursory slip. The predicted earthquake did not arrive until 2004, failing to satisfy the specific predictions and showing no clear short-term precursors. The experiment demonstrated that even on a well-understood fault segment with a history of characteristic earthquakes, prediction remains immensely difficult.