In his latest column, Chapman University Chancellor Daniele Struppa raises interesting questions about whether scientists will ever be able to reliably predict earthquakes.

On May 2, 2008, Dr. Dimitar Ouzounov, currently a scientist from the Goddard Flight Center at NASA, issued an internal early warning earthquake alert for a very seismically active Sichuan region in southern China. Only seven days days later, a devastating earthquake (magnitude 7.8) hit a region slightly northwest of the predicted area.

How could Dr. Ouzounov have enough information to identify potential precursor signals for such an early warning?

To answer this question we need to go back to 2003, when Dr. Ramesh Singh co-authored a paper with S.Dey, where he identified a particular type of signal, namely surface latent heat as a possible indicator and precursor of a seismic phenomenon. For the purpose of this column, we can think of surface latent heat as an indication of how much the temperature of the surface of the earth is changing (above and beyond what can be explained by usual climatic causes).

In a word, the theory proposed by Singh and Dey tells us that a few days before a major earthquake, we should expect a sudden, measurable, and otherwise unexplainable change in the temperature of the surface of the earth. The change we expect is such that it can be easily detected by our current satellite capabilities. Interestingly enough, at least according to some scientists, the use of such an indicator does not seem to be well applicable to the Sichuan area, because of the special kind of earthquakes which occur there. Nevertheless, there are other potential precursor signals, which can be used for this area, and these are the ones used by Ouzounov.

Soon after the work of Singh, a team led by Professor Menas Kafatos (the leader of the team of scientists who are now moving to the new College of Science at Chapman University) used this concept to identify a novel way to attempt early warnings for earthquakes. Back in 2004, the team studied two earthquakes which had just taken place in Greece and developed accurate mathematical techniques to analyze the data received by satellites.

Under the constantly developing approach proposed by the team of scientists led by Kafatos, we can actually utilize multi-sensor satellites to analyze a variety of data, which appear to be precursors of seismic phenomena. Our satellites, for example, offer data on heat, wind, water vapor, sear surface temperature and chlorophyll concentrations, and all such data are freely available on the web. Using such NASA, NOAA and other data, along with data from a direct readout satellite facility (such a facility is being installed at Chapman, will help to speed up the verification and decision makings for earthquakes, obviously a major interest in California.)

In order to validate the approach, Kafatos and his colleagues have gone back and examined data in a variety of important earthquakes. In a paper published in 2007, for example, Singh, Kafatos and their colleagues have analyzed the data related to the disastrous earthquake (and tsunami) of December 26, 2004. Their results are quite stunning, as they show that in fact one can establish a strict connection between changes in measures of heat, vapor, and other atmospheric data, and the subsequent earthquake.

But the capability of the method developed by Kafatos and his colleagues has now graduated from the verification stage (after a quake, we verify that indeed the variations we should have expected were there) to the early warning stage, albeit with a lot of uncertainties and a lot more work that needs to be carried out (as demonstrated by Ouzounov in the recent case of the Schichuan earthquake). This method is capable of offering reliable and timely warning, which could be used at least to move civil and military resources to a high alert status.

The idea that remote sensing could provide a way to offer early warnings of such significant events is not new, but what has been lacking for a long time is the capability to measure atmospheric data in an almost continuous fashion. It is just not enough to take a look at what happens every week, we need much more frequent data. It is also the case that relatively recent advances in mathematics (especially the theory of wavelets, which was developed in the eighties for applications to signal processing) are now allowing scientists to detect what we call ’singularities’ and what a layperson would call a ’sudden change.’

The particular value of wavelets consists in being designed exactly to detect such sudden changes at the level of detail which is appropriate to a specific phenomenon. There is the additional difficulty that the physics of underlying earthquate processes that produce these precursors is not yet well understood or agreed upon by scientists.

From what the scientists understand, the reason for these variations in meteorological data have to do with the friction which occurs deep in the earth crust in advance of the actual seismic event, and it is estimated that the method we just described can be applied to earthquakes which are at least magnitude 5.5 in strength, and which originate no deeper than 25 to 50 kilometers underground.

While we are not yet at the stage where we can know when and where an earthquake will strike, we are certainly getting close to being able to determine both the when and where with sufficient precision to impact our readiness status. Additional validation and more work on basic science are needed to understand the physics of earthquakes, and to reduce the false alarms ratio or “finding the needle in the haystack of signals that would be associated with earthquake precursors” as Kafatos points out.

Learn more about quakes:

• Huge quake could kill 363, injure 9,000 in O.C.
• Three microquakes shake Anaheim Hills
• Earthquake Central: Check latest temblors 

Posted in: Earthquake news and research • Guest columns • Chapman • earthquake • NASA • quake • Struppa |

You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.