Spotlight on Research: Mason Center Close to Achieving Earthquake Prediction System

Posted: February 10, 2005 at 1:00 am, Last Updated: November 30, -0001 at 12:00 am

By Robin Herron

Within 30 minutes of an earthquake occurring somewhere in the world, the computer program Guido Cervone spent two years designing automatically generates an analysis, complete with color charts and graphs. Now Cervone, who works for Mason’s Center for Earth Observing and Space Research(CEOSR), has turned his attention from retrospective analysis to forecasting, an area of research expected to intensify in the wake of the deadly Dec. 26 Sumatran earthquake and tsunami.

Forecasting, however, is a tricky business. “The problem with earthquake prediction,” says Menas Kafatos, CEOSR director and dean of the School of Computational Sciences, “is that there have been many false alarms, and that has caused people to be very cautious, including ourselves. To have the scientific community convinced, we need to be able to predict things, and we’re not there yet. But I believe that within six months we’ll have an operational system here at George Mason.”

In fact, the center’s ongoing monitoring of earthquakes led fellow CEOSR scientist Ramesh Singh to take the bold step of issuing a warning to India shortly after the Dec. 26 quake. Data from NASA satellites equipped with remote sensors, which the center uses in its analysis, indicated that strong aftershocks in earthquake-prone areas of India were occurring, and are still occurring to this day. There was a possibility these aftershocks could trigger a large quake like the one in Sumatra. The story was reported in the Washington Post,Fairfax section, on Feb. 3. The same day, a 5.5 magnitude earthquake hit the Assam region as Singh had warned.

“Dr. Singh’s warning about the fault lines in India was received with a little bit of caution, but I think it was received well,” says Kafatos. “The governor of the Assam region said, ‘Yes, we should be monitoring these things.’ And the president of India said, ‘We’d like our scientists to build a system that can predict earthquakes.’ This is our approach here as well.”

Cervone, who earned his PhD in Computational Sciences and Informatics and his MS in Computer Science from Mason, began working with Singh, a visiting professor, as a student a couple of years ago. “With his background in earthquakes and my background in data mining, we made a good team,” Cervone says. They developed their analysis system using five years of retrospective data from 300 earthquakes. The system became the basis for Cervone’s PhD thesis.

The massive earthquake in the Indian Ocean last December, measuring 8.9 on the Richter scale and causing the devastating tsunami that claimed some 300,000 lives, has provided mounds of data that can help refine the system even further. What Cervone has been doing and will continue to do is monitor atmospheric parameters, observed through remote sensing, that are correlated to earthquakes. The Center has developed a web page that provides numerous data and links, including an analysis of parameters that show anomalies as precursors to the deadly Sumatra event.

Guido Cervone with earthquake data
Guido Cervone with a poster showing anomalies as precursors to earthquakes in coastal Japan.
Photo by Evan Cantwell

“What we believe is that in the case of coastal earthquakes, there is a very strong interaction between the lithosphere [the outer surface of the earth], the hydrosphere, [bodies of water and vapor in the atmosphere] and the atmosphere,” Cervone says. “So even though earthquakes can only occur either in the ocean or on land, their effects can be seen in the atmosphere. With our remote sensing sensors, we can study the atmosphere very well with our global coverage every day.” The data CEOSR receives daily from the U.S. Geological Survey shows that one or more earthquakes occur nearly every day somewhere in the world.

Since most coastal earthquakes occur along known continental boundaries, these areas get special attention. “If we see anomalies that have geometrical characteristics similar to one of the continental boundaries, right on top of the boundaries, then we know that they are most likely related to earthquakes and not just to other atmospheric phenomena. We look for very large anomalies that have a very particular geometrical shape—these can be a line, a U-shape, or an inverse U,” Cervone explains.

Other indicators of an earthquake are thermal anomalies, electromagnetic anomalies, release of gases, and sound—something that animals are particularly attuned to. CEOSR’s observations also show that there is often a chlorophyll blooming before a coastal earthquake.

But what interests Singh, Cervone, and Kafatos primarily is a parameter called surface latent heat flux. Latent heat is heat that changes only the structure or phase of a material, not its temperature. When coastal earthquakes occur, increases in surface temperature due to the earthquake’s motion cause water to evaporate. And when evaporation is taking place, there is a positive surface latent heat flux transfer, indicating that the surface is losing energy to the air. “When the surface of the water gets warmer, there is more evaporation, and we can monitor it,” Cervone says.

Diagram of SLHF before the Sumatran earthquake
This image shows a surface latent heat flux anomaly in the vicinity of the Sumatran earthquake several weeks before the earthquake occurred.

The CEOSR scientists believe surface latent heat flux may be the key to successful forecasting. The signs are promising. On Jan. 1, Cervone forecasted an earthquake off the coast of Japan. Eight days later, the earthquake, registering 5.3 on the Richter scale, occurred where he had predicted. His system has shown that the weaker the earthquake, the shorter the time span between when an anomaly is spotted and when the quake occurs. A massive quake has a much longer time span between the appearance of an anomaly and the actual quake, providing more time for preparation or evacuation.

To refine his quake prediction system, Cervone is trying to integrate the different parameters. “If we just see one anomaly, latent heat flux, as high as it can be, it’s very hard to determine whether it’s really an earthquake or something else. But if we start seeing an anomaly in heat flux, maybe in chlorophyll, some atmospheric anomalies—then we can be more sure about our prediction. Our work in the past few months has been in trying to combine our methods with other groups around the world, to really find out what are the signals that predate earthquakes.”

Cervone also plans to narrow his focus to certain regions of the world, because anomalies change from region to region. “Then we’ll analyze those regions individually, maybe with some local experts in those regions.”

Kafatos adds that CEOSR expects to begin working with scientists at the University of Athens and the Istanbul Technical University in Turkey to monitor the earthquake-prone Eastern Mediterranean with high-resolution data.

With the tsunami fresh in people’s minds, Kafatos feels that a sense of urgency will propel research and action to increase the accuracy of forecasting. “I don’t think it’s business as usual,” he says. “Seventy-five percent of the world’s population lives near the coast. People are aware, for example, that nuclear power plants near coastlines pose dangers.”

CEOSR’s work on earthquakes is part of a larger research effort that comes under the Virginia Access and Mid-Atlantic Geospatial Information Consortium and also focuses on other natural disasters, such as hurricanes, forest fires, dust storms, floods, and desertification.

Write to at