Stations such as this one deployed on the rugged hills of the Costa Rica’s Nicoya Peninsula use Global Positioning System (GPS) devices to help researchers detect slight shifts that indicate which portions of the subduction zone are locked. Courtesy of Andrew V. Newman
Analyzing GPS coordinates and geomorphic observations, a research team successfully identified the locked portion of a subduction zone in Costa Rica, forecasting the location and size of an earthquake.
February 18, 2014—Researchers analyzing the changing Global Positioning System (GPS) coordinates of survey markers and compiling geomorphic observations of the Nicoya Peninsula in Costa Rica were able to correctly identify the portion of the subduction zone that was locked and thus anticipate the location and approximate intensity of a magnitude 7.6 earthquake that shook the peninsula in September 2012.
The research, which was published as a letter to the journal Nature in December 2013, holds intriguing implications about the future of engineering building codes in seismic zones, according to one of the authors of the paper, Andrew V. Newman, Ph.D. an associate professor in the School of Earth and Atmospheric Sciences at the Georgia Institute of Technology. Marino Protti, Ph.D., a seismologist at Universidad Nacional de Costa Rica, Timothy Dixon, Ph.D., a professor at the University of South Florida, and Susan Schwartz, Ph.D., a professor at the University of California, Santa Cruz, were also key authors of the letter, “Nicoya Earthquake Rupture Anticipated by Geodetic Measurement of the Locked Plate Interface.”
“By mapping out the locked areas, and evaluating the true event potential in a region, we can give a much better set of parameters for a deterministic event,” Newman says. “This means, when designing for a specific region, you may be able to use such information to build for that particular event—an earthquake of X magnitude, occurring at Y location.”
Tourists prize the Nicoya Peninsula on Costa Rica’s Pacific coast for pristine beaches and nature reserves. The peninsula is approximately 75 mi long and varies in width from 19 to 37 mi. Seismic researchers value the peninsula as a location where one of the few subduction zones— the Cocos Plate subducts beneath the Caribbean Plate—is not resting deep beneath the ocean.
The subduction zone has been reliably active in the past, creating strong earthquakes in 1853, 1900, and 1950, prior to the temblor in 2012. The subduction zone’s history provided researchers with a strong framework to study the deformation and observe the build-up to and the results of the earthquake.
“We were able to forecast the location—including the spatial extent—and approximate size of a potential earthquake,” Newman says. “The model of locking could not tell when the future earthquake could occur, or whether the earthquake would rupture in a single large event, or several smaller events. However, this information is better constrained looking at the historic data.”
Newman says this is the first example of where researchers mapped out a locked area in a subduction zone with great fidelity. “With the subsequent earthquake, it also became the first place where we can detail the relationship between what we knew was locked before, and how the earthquake actually ruptured along the fault,” he adds.
The team used a combination of methods to monitor the subduction zone and determine where it was locked. The team drove survey markers into the ground and determined their precise locations with GPS devices. When the technology was first being tested for high-precision ground deformation, the GPS devices cost approximately $500,000. The devices now cost less than $10,000, and are usually left in place to record for three days, logging measurements every 30 seconds.
Because the plates are shifting by mere centimeters during these observations, the additional measurements help the team detect and adjust for discrepancies caused by a number of external factors, including large fluctuations in the GPS satellites’ orbits when they are buffeted by solar winds that can shift satellite positions by more than 10 ft in just minutes, Newman says.
Additionally, the team made detailed field observations of the site, noting such changes to coastline as shifts in the high tide lines. These observations provide valuable insight because they have a history that greatly predates GPS measurements.
Researchers in Japan are attempting to adapt similar techniques to record underwater measurements of subduction zones there. A magnitude 9.0 earthquake on a subduction zone unleashed a crippling tsunami on March 11, 2011, killing more than 15,000 people. Deploying the technology on the ocean floor greatly increases the cost and complexity. Researchers are deploying GPS monitors on land and creating an underwater cable network to record similar observations.
The techniques are also being utilized to measure such continental transform faults as the San Andreas Fault in California. “The results in such faults are a little trickier because these faults are near vertically oriented, and hence it becomes more difficult to differentiate shallow versus more deeply locked sections,” Newman notes. To help make that distinction, researchers are experimenting with strain meters installed in deeply bored holes along the fault line.
In the aftermath of the earthquake in Costa Rica, the team is monitoring how the crust and fault zone recover. “We are gaining new insight into the healing process of the fault, and how we again transition from a fault that slipped, to again one that locks. This process may take a decade or more,” Newman says.
“For most active plate-boundary faults, only a fraction of the fault's surface actually locks up to fail in future earthquakes. This fraction can be anything between 5 percent and 50 percent, depending on environment,” Newman says. “At this point we don't have a clear understanding why some sections lock, and others slip, and whether this changes with time through the seismic cycle.”
Normally this seismic cycle of locking, earthquake, recovery, and relocking occurs over periods that extend over several hundred to several thousand years. The team’s work in Costa Rica, however, observed almost the entire cycle.
“By analyzing when and how faults lock, we can truly move beyond mostly statistical assessments of earthquake recurrences to fundamentally physical models unique for each individual plate-boundary environment,” Newman says.