Improving Earthquake Prediction

By V.I. Keilis-Borok

The grinding together of blocks of varying size in the Earth's lithosphere - a crust estimated to be perhaps 80 kilometres deep - triggers some one million earthquakes a year. An estimated 15,000 people are killed annually by these tremors. The relative motion of the shifting boundary zones sets up waves, with resulting instabilities which would appear to turn the lithosphere into a nonlinear system.

Earth scientists have now begun to investigate whether the mechanisms which create this condition of nonlinearity might feature deterministic chaos. If so, it could help unlock the secrets of earthquake prediction - through study of the premonitory patterns of seismic activity. The work that he and his colleagues have done on the possible interrelationship between chaos and earthquake occurrences was discussed by V. I. Keilis-Borok of the International Institute for Earthquake Prediction Theory and Mathematical Geophysics, Russian Academy of Sciences, Moscow. - Editor

At the lower end of the hierarchy, interface zones separate the grains of rocks. Earthquakes make up a significant part of the movement between the blocks which form the Earth's seemingly rigid crust. The largest of these blocks are massive tectonic plates as big as continents. Friction, cohesion and relative motion between the blocks, are controlled by many diverse processes: the generation and flow of fluids; petrochemical and other transformations; fracturing, buckling, rock fatigue, and so on. All of these mechanisms turn the lithosphere into a nonlinear dissipative system which may well exhibit the characteristics of deterministic chaos.

One is stress corrosion or what is known as the "rhebinder effects" (from "rhe," the unit of dynamic fluidity), in which solids and liquids can combine through a process of cracking and interpenetration. For example, when basalt and a sulphur solution meet, the basalt is permeated by a grid of cracks, and instabilities are created.

A competing mechanism is the more conventional filtration of fluids through pores or cracks. A slip (the geological term for the distance between two points that were next to each other before movement took place) can then occur, and, with friction lessening, the movement accelerates. Instability arises as tiny slips self-accelerate, grow and merge. The very diversity of these processes suggests that we need to look for an integrated, generalized description of their actions which will help us to more clearly see the lithosphere as a nonlinear dissipative system. The fundamental equations for this system are as yet unknown. Its study, therefore, is in a particularly challenging, "pre-equation" state of phenomenology and modelling.

One of the most consequential result of our explorations to date is the understanding that the basic regularities in earthquake occurrence can be explained by the general universal properties of nonlinear systems, without evoking processes intrinsic to the lithosphere itself (essentially the principle of universality fundamental to chaos that different systems can behave identically or, from another viewpoint, there are structures in nonlinear systems that are always the same.)

What Models Show

Models first developed in theoretical physics and astrophysics were introduced in earthquake studies as early as 1967. One such model has illustrated how small displacement of elementary blocks can self-organize into strong earthquakes. The elements show prominent collective behaviour. The most stressed become self-organized in linear configurations which eventually become unstable and trigger strong earthquakes. Our experience with models suggests that we should regard the earthquake-prone lithosphere as a system remaining permanently near a critical state. Key questions can now be explored with models: the connections between the internal and the external (the "observable") parameters of the lithosphere; the minimal set of parameters required to describe the essential features of the lithosphere's short-term dynamics; the types of instabilities and transitions to chaos, and so on.

Figure

Earthquake Predictions

Earthquake prediction consists basically in a step-by-step reduction of the time and space domain where a strong earthquake might be expected. But our forecasting ability is limited by the fact that the tremors are generated by a nonlinear system of earthquake-prone faults. Our hopes for improved accuracy may lie in two opposite directions.

One arises from the notion that a strong earthquake is not an abrupt instantaneous transition within a fault system, but part of some scenario slightly extended in time. The beginning of this scenario might be used as a short-term forecast - even though it is actually a fixed part of an earthquake which has already occurred.

A second possibility is that a non-linear system may become predicable after a process of smoothing or averaging. Different degrees of averaging may yield different stages of prediction: the bigger the averaging, the larger the time-space interval in which the earthquake may be expected. This would be relevant to intermediate or long-term forecasts.

Most of the premonitory seismicity patterns analysed thus far can be interpreted as different expressions of the same phenomenon: an increase in the response to excitation. This may be expressed by the increase in the following:

· clustering of the tremors;
· distance at which they interact;
· intensity of the earthquake flow;
· irregularities in time and space.

The approach of a strong earthquake is diagnosed when a sufficiently large number of these phenomena are observed. It must be emphasized that these, phenomena are defined over large areas; they must be roughly averaged, therefore, and not treated independently.

In the present pre-equation state of the art, one must be cautious about evoking too much from the enchanting vocabulary of chaos. Still, it is clear that a number of new approaches, rooted in this field, have yielded important results for our understanding of the dynamics of the lithosphere and earthquake prediction. There have been more attempts to explain major phenomena by their universal properties, not their details. There has been recognition of the partial similarity of structure and dynamics in a wide range of energy and neotectonic environments.

Such approaches have provided an important alternative to quite opposite, and until recently, dominant tendencies. The latter had tended to) focus on essential details first, in the hope of later integrating them into a general picture. The concern has been with differences, rather than similarities, and on specific mechanisms rather than their aggregation. This synthetic approach reflects the illusion that one may understand nonlinear systems by breaking them apart.

With the help of the new trends, our understanding of the lithosphere has been consolidated, the database was trimmed, and reproducible earthquake predictions emerged. This came none too soon in view of the critically growing need for earthquake prediction. The fact that this could be achieved by sheer force of new concepts even before they had been evolved into a theory illustrates how timely is the idea of this symposium.

Earthquake probabilities in California. With dark circles indicating an earthquake with a magnitude of M ³ 2.7 on the Richter Scale. The (Loma Prieta) earthquake of October 1989, measuring M ³ 6.9 was diagnosed in advance; the Imperial Valley diagnosis was based on data assembled after the earthquake.

15 October 1979 Imperial Valley

18 October 1989 Loma Prieta