

Figure 11-6— THE UPPER MANTLE IN THE REGION OF FIJI-TONGA-RARATONGA 



INDIA PLATE 



FIJI 



TONGA TRENCH 



^* — 



PACIFIC PLATE 



LITHOSPHERE 



ASTHENOSPHERE 



RAROTONGA 



•j\' 



NEW ZEALAND 



This figure depicts an area where the Pacific plate (east of the Tonga trench) meets 

 with the India plate, pushing the lithospheric mass of the Pacific plate downward 

 forming the Tonga trench. Earthquakes take place all along this zone, closer to the 

 surface near the trench and at progressively greater depth beneath the continental 

 (India plate) mass. 



Changes in the rate of strain are 

 another potential premonitor. These 

 changes would be accompanied by an 

 increase in the rate of occurrence of 

 microearthquakes — i.e., very small 

 earthquakes that are indicators of 

 "creaking." The U.S. program for 

 earthquake prediction has a strong 

 component devoted to the problem of 

 detecting changes in rates of strain 

 along parts of the San Andreas fault 

 system, including triangulation and 

 leveling, tilt, distance measurements, 

 and microearthquake observations. 

 Some intermediate-sized earthquakes 

 have been preceded by observed in- 

 creases in rates of microearthquake 

 activity and by increases in the strain 

 rate, as measured by changes in the 

 lengths of reference lines drawn 

 across known faults and by changes 

 in the tilt rate. 



Other physical properties in the 

 vicinity of earthquake faults may 

 change prior to rupture. These in- 

 clude magnetic susceptibility, elec- 

 trical resistivity, and elastic-wave 

 velocities. There is one, as yet un- 



duplicated, example of a Japanese 

 earthquake preceded by major changes 

 in the local magnetic field. A minute 

 change in the magnetic field has also 

 been noted in the neighborhood of 

 one part of the San Andreas Fault 

 about one day before each of several 

 microearthquakes occurred. Changes 

 in the other properties have been ob- 

 served in laboratory experiments on 

 rock fracture but have not been veri- 

 fied in earthquake examples. 



Stress Measurement — There is 

 considerable debate about the values 

 of the critical stress required to cause 

 rupture. Seismological estimates place 

 the stress drop at about 10 to 100 

 atmospheres (bars). Laboratory ex- 

 periments show the stress drop to be 

 perhaps one-fourth the shear stress 

 across the frictional surface, although 

 there appears to be some seismologi- 

 cal evidence that the fractional stress 

 drop rises with increasing earthquake 

 magnitude. In any event, the overbur- 

 den pressure should be enough to 

 seal faults shut, and no earthquakes 

 should occur below about 2 kilome- 



ters. But earthquakes do occur below 

 this depth. Thus, one must find some 

 reason why friction at depth is re- 

 duced. One way of doing so is to 

 invoke the role of water as an impor- 

 tant lubricant: that is, rocks lose 

 some or all of their shear strength 

 when interstitial water is raised in 

 temperature. 



No major progress has yet been 

 made on in situ measurement of shear 

 stress and determination of pore 

 water pressure and temperature (to 

 determine critical shear rupture 

 stress). In principle, direct stress 

 measurement may be the simplest 

 way to predict earthquakes, but it 

 may also be the most difficult to 

 effect in practice. 



Historical Method — In this case, 

 we ignore the physics of the earth- 

 quake mechanism in large part, and 

 concentrate instead on the history of 

 earthquake occurrence (seismicity) as 

 a mathematical sequence. We can 

 then investigate this historical se- 

 quence for regularities — if any are 

 present. The search may take two 

 forms: (a) a search for triggering 

 effects — i.e., a tendency for earth- 

 quakes to occur at certain preferred 

 times; and (b) a search for organiza- 

 tion within a local catalog. 



Triggering is a cross-correlation 

 problem in which two time-series are 

 compared, one of which is the catalog 

 or compilation of the earthquake his- 

 tory for a particular region. No sig- 

 nificant triggering effects have yet 

 been found, although the earth tides 

 should be the most likely candidate. 

 In a number of cases, earthquake 

 activity at a distance from a given 

 region seems to be reduced following 

 a large shock. However, this effect 

 may be "psychoseismological": that 

 is, seismologists are more likely to 

 report aftershocks in an active area 

 and to neglect reporting for other 

 areas. Furthermore, the occurrence 

 of a large shock in one region will 

 reduce the tendency for another to 

 occur in the same region, and will 



37 



