Candidate precursors: pulse-like geoelectric signals possibly related to recent seismic activity in Japan

Transcription

1 (;cophys. J. Int. (1997) 131, SPECIAL SECTION-ASSESSMENT OF SCHEMES FOR EARTHQUAKE PREDICTION Candidate precursors: pulse-like geoelectric signals possibly related to recent seismic activity in Japan Yuji Enomoto,' Akito Tsutsumi,' Yukio Fujinawa,2 Minoru Kasahara3 and Hiroshi Hashimoto4 ' Mechanical Engineering Laboratory, Namiki 1-2, Tsukuba, Ibaraki 305, Japan. E-inail: National Research Institute for Earth Science and Disaster Prevention, Tennodai 3-1, Tsukuha 305, Japan. ; Hokkaido University, Faculty of Science, Kita-ku Kita 10-Nishi 8, Sapporo, Hokkaido 060, Japan. COMTEC Inc, Shimo-renjyaku , Mitaka, Tokyo 181, Japan Accepted 1997 August 29. Received 1997 August 29; in original form 1997 February 3 INTRODUCTION So far, various methods have been used to detect seismic electromagnetic signals over a wide frequency range from ULF to HF bands. Various types of electromagnetic anomalies preceded the Kobe-Awaji earthquake of magnitude 7.2, 1995 January 17 (Maeda & Tokimasa 1996), since when research on the detection of pre-seismic electromagnetic signals has intensified in Japan. On the other hand, there are many disputes about earthquake prediction based on observations of pre-seismic electromagnetic signals, especially VAN prediction (Lighthill 1996; Geller 1996), regarding the causality, correlation, precision, working hypothesis, etc. Such a situation may suggest that more detailed and reliable information is needed on the nature of pre-seismic electromagnetic signals, taking into account the effects of meteorological electric activity as well as industrial activity. SUMMARY Telemetric network observations of pulse-like geoelectric charge signals using a vertical dipole buried under the ground were undertaken at various observation sites over a wide area in Japan from April From continuous records of the signals during the six months following that, we attempted to select anomalous signals, possibly related to seismic electric activity. Special attention was paid to the elimination of noise resulting from industrial and meteorological electric activity, comparison with other electromagnetic signals in the VLF band and the relation between the precursor period and the distance from the eventual main shock that occurred in Japan. Four candidate precursor electric signals, which were not contaminated by industrial and meteorological electric activity, were then selected, of which the second appeared before the Akita-ken Nairiku-nanbu earthquake swarm, for which the maximum M = 5.9 occurred on 1996 August 11, and the third and fourth before the Chiba-ken Toho-oki earthquake, M = 6.6, on 1996 September 11. A tentative qualitative model explaining why the candidate precursory signal is related to stress building up before an earthquake is presented in terms of the electrification of gases released from fracturing rocks immediately prior to the main shock. Key words: earthquake prediction, electromagnetic signal, geoelectric current, seismic precursor. Observations of pulse-like electric-charge signals between a vertical dipole buried under the ground are being conducted at the Mechanical Engineering Laboratory using a specially designed charge amplifier, referred to as the EH method (Enomoto & Hashimoto 1994). These observations were started in 1992 at Tsukuba, 60km northeast of Tokyo. In April 1996, telemetric network observations were started at various observation sites across a wide area in Japan. In this paper, we attempt to select candidate pre-seismic electric signals from our continuous records, following, to the extent at present possible, the 'Guidelines of earthquake precursor candidates' provided by the IASPEI Subcommission on Earthquake Prediction (Wyss 1991). That is, we discuss the nature of geoelectric signals in terms of industrial and meteorological electromagnetic activity and by comparing them with signal observations in the VLF band, as well as in terms of their relationship with earthquake occurrences RAS 485

2 486 Y Enonioto et a1 (a) charge detector + source signal isolation transformer odem- I+&iJ@ Mechanical Engineering Laboratory Iti-observation sites personal computer isolation transformers Figure 1. Schematic views of (a) the Observation system and (h) the processed signals and the geoelectric-charge signal detector. r OBSERVATION METHOD Fig. 1 (a) shows a schematic view of the observation system. A polyvinyl chloride pipe with an outer diameter of 89 mm was buried at a depth of 100 m; one electrode was set at the bottom of the pipe and another was placed 50m deep outside the pipe. A second pipe with an electrode mounted at the bottom was buried at a depth of 5m close to the first pipe. When burying the pipe in the ground, sections of length 4 m were joined, sealed to make the inside watertight, and then filled with enough dry pebbles to offset the buoyancy of underground water. The electrodes were made of stainless steel. One of three combinations, that is 100 m-5 m, 50 m-5 m or 100 m-50 m, was used as a vertical dipole for the electric-charge signal measurement. In order to choose the electrode pair, we crosschecked to select the lowest electrical background signal during the initial setting-up of each system. The natural electric potential difference between the electrode at l00m and that at 5 m was typically about 100 mv. The measuring system is schematically illustrated in Fig. 1 (b), where the geoelectric signals with various frequency components are fed into an electric-charge amplifier through an isolation transformer, which cuts d.c. and lower-frequency electric noise less than about 100 Hz. The electric supply line for the amplifier and the personal computer was also electrically floated using isolation transformers to remove any d.c. and ultra-low-frequency electric noise. A specially designed electric-charge amplifier (COMTEC Inc, CSN-2000) was used for the measurement, where electric charges q that flowed into an electrode under the natural electric potential were fed to a pre-amplifier through an isolation transformer and converted IW ' Figure 2. Map of Observation sites. Open circles: the present system of geoelectric signals (EH method); open triangles: the system for measuring underground electromagnetic signals in the VLF hand (TAF method). The hatched area indicates the area where lightning can he detected by commercially available LPATS.

3 Geoelectric signals and seismic activity, Japun ee ' ! day z 5 29 day I _ 31 day Figure 3. Typical examples of CRT display of the one month geoelectric signal in August 1996 obtained by the EH method (a) Tsukuba, (b) Erimo and (c) correlation between (a) and (b).

4 488 Y. Enomoto et al I n L I i 2o [qugust (*) Figure 4. Geoelectric signals observed at Tsukuba, 60 km northeast of Tokyo, from April to September C indicates signals contaminated by lightning, I indicates signals possibly contaminated by localized industrial noise, T indicates signals contaminated by thunderclouds and S indicates the candidate seismic electric signals. Dotted lines indicate the tentative threshold signal intensity of 8. to an electric voltage signal C- q(t), where C is the capacitance of several pf. The system is sensitive in the nominal frequency range 10 khz-1 MHz. However, we paid attention to the detectioxof random pulse-like signals (electric surges) because of our working hypothesis, in which electrostatic charge may accumulate at ground level near the epicentre due to seismic electric activity, and then burst-like electromagnetic disturb- ances may occur due to discharge at ground level. To this end, the signals I( t) with a negative sign and a signal rise time higher than 1 ps were electronically selected, amplified, integrated over a sampling interval of 4 s, and then successively stored on a hard disk. The signals I(t) still include some groups appearing at a regular time interval, possibly attributable to higher components of industrial noise and their

5 beat-frequency noise. In order to eliminate such signals, the signal intensity I( t,) was differentiated using the computer algorithm I*(t,) = [I(t,)- I(tfl-,)], and the signal I*(t,) above the threshold value was picked up. Only random pulselike signals are then detected, which are displayed on the CRT. After carrying out the above procedure at an observation site, the processed signals were transmitted through a telephone line to the central observation station at the Mechanical Engineering Laboratory, Tsukuba, where the signals can be displayed on a CRT at any timescale desired. 12 stations, installed across a wide area of Japan, are operating well; these are shown by open circles in Fig. 2. In order to compare this method with other observation methods, the signal obtained by the EH method was compared with the subsurface signal of the electric field in the VLF band of 1-9 khz obtained using a vertical borehole antenna ranging from 100 to 800 m in depth; this is known as the TAF method (Fujinawa & Takahashi 1990, 1994). Open triangles in Fig. 2 indicate the observation sites of the TAF method In order to check if the signal was contaminated by meteorological lightning and thunderclouds, commercially available information on lightning and rain clouds is monitored in real time from Weathernews Inc. Lightning events are observed by the Lightning Positioning Tracking System (LPATS), in which the position of lightning is determined by the timeof-arrival method of electromagnetic emission in the range khz (Casper 1991). The hatched area in Fig. 2 indicates the detection area of lightning events. Meteorological information on radar echo maps showing the distribution of rain clouds was also monitored. RESULTS Selection of anomalous electric signal The geoelectric signal records of any two sites can be compared on a CRT or its printed copy. The one-month signals at the Tsukuba and Erimo sites for August 1996 are shown in Figs 3(a) and (b). From these data, correlation of the signals between the two can be obtained, as shown in Fig. 3(c), using the computer algorithm I**( t) = [I: (t)- I,* (t)], where 1: (t) and I,*(t) are the signals at two different sites. Fig. 4 shows all the geoelectric signals observed at Tsukuba from April 1996 to September Earthquakes that occurred in Japan during this period of magnitude greater than 5.5 and focal depth less than 99 km are shown on the timescale in Fig. 4 by the letters (c) and (d). Earthquakes of magnitude greater than 5 that occurred within a radius of 300 km from the Tsukuba site are also included in Fig. 4, shown by the letters (a), (b), etc. The epicentres of the earthquakes shown in Fig. 4 are plotted in Fig. 5. Earthquake information is reported by the Japan Meteorological Agency. The signals, shown in Fig. 4, might be contaminated by artificial electric noise from industrial electric activity or might be due to natural electromagnetic activity such as lightning and rainfall, geomagnetic storm and seismic electromagnetic activity. The amount of rainfall per day is included in the data at Tsukuba, shown as bars in Fig. 4. It is clear that the change in conductivity of the ground due to rainfall did not affect the signal. This is because the change in electrical conductivity associated with rainfall is essentially a low-frequency phenomenon that is diminished in the present detection system. It is RAS, GJI 131, Geoelectric signals and seismic activity, Japun 489 I m a) Iwate-ken Enganbu. M= km 13:08, April 23 b) Miyagi-ken Oki M=5.0.39km 48~36 May 23 c) Aklta-ken Nalriku-nanbu d) Miyagi-ken Hokubu o8:10, Aug. 11, respect. e) Torishima Kinkai, I I 3:15, Sap. 5 f) Chibir-ken Toho-oki k6.2.53km If :37, Sap. 1.1 Figure 5. A map showing the epicentres of earthquakes of M > 5.5 and depth less than 100 km that occurred from April to September The two earthquakes of M = 5.0 that occurred within a radius of 300 km of Tsukuba are also included. also noted that geomagnetic storms did not affect our signal, even though the activity was high on some days (Enomoto & Hashimoto 1994), probably because the geomagnetic storms induce an electric current in the Earth with a horizontal component, but little vertical component. All signals at the various observation sites were printed out in the same form as Fig. 4, and then checked to eliminate noise due to industrial activity and lightning and thunderclouds. To this end, we employed the criteria shown in Fig. 6. As mentioned above, we employed various ways to eliminate local industrial noise, such as using electronic filters, isolation transformers, etc. Nevertheless, some artificial or natural noise from unknown sources appeared at some stations. We then picked out the geoelectric signals with an intensity greater than a certain threshold signal intensity, tentatively 8 in Fig. 4. Electrical disturbances due to industrial noise might not appear simultaneously at more than one station, the stations being several tens of kilometres distant from one another, because they are limited to localized areas. There is a fear that the seismic electric signals, possibly related to regional shallow earthquakes of smaller magnitude, may appear at only one site and be eliminated. This study, however, focused on seismic electric signals associated with larger earthquakes, those of M > 5.5, in which we assume that the signals might appear over a wide area, and might be detected by more than one site. Therefore, the signals, labelled I in Fig. 4, which did not appear simultaneously at more than one site, were eliminated from the present consideration. This procedure was usually carried out by eye because this method is somewhat easier than using a computer procedure to pick up coincident signals as shown in Fig. 3(c). After these procedures, the signals that appeared at more than one site were examined in relation to the LPATS signal. Typical LPATS signals in August and September are shown in Fig. 7(b). The geoelectric signals labelled as L in Fig. 4 are

6 490 Y. Enornoto et a1 n o - r j Is the signal greater than threshold 7 no; labelled as I +,ye; Does the signal appear at more than one-site? Does the signal apper srmul with a VLF band s!gnal (Fulnawa and Takahash!)? yes: labelled as S Selected anomalous signal Earthuuake of Ab5.5 underclouds (Radar echo) 7 7 w!th l!ghtnmg (LPATS)? Figure 6. Selection criteria of the candidate seismic precursor signal from the pulse-like geoelectric signal records at multi-observation sites. c 1 I Sl I I I I Figure7. (a) Underground electric field signal in the VLF band using the TAF method and (b) the LPATS lightning signals for August and September S indicates the candidate seismic electromagnetic signal. Day attributed to lightning. It has been suggested that there is possibly a phyical relationship between regional lightning and shallow earthquakes (Oike & Yamada 1994). However, we excluded the signal L from the present consideration in order to confirm whether anomalous electric signals appear even on a fine day when there is no possibility of meteorological lightning. LPATS detects cloud-to-ground discharge activity, but less intracloud activity, because the latter information is not needed commercially. Sometimes, regional lightning was dominated

7 by intracloud discharge, which affected the signal. This could be checked by using the metorological information about rain clouds. For example, as shown in Fig. 8(a), the geoelectric signals on 1996 July 8 were contaminated by lightning. This fact is confirmed by the lightning events of Fig. 8(b), and also by the lightning map of that day shown in Fig. 8(c). By contrast, the geoelectric signals that appeared in the early morning of 1996 July 9, as seen in Fig. 8(a), did not correspond to lightning events, that is there was a lower LPATS signal on that day, as can be seen in Figs 8(b) and (d), but the signal might possibly have been contaminated by the intracloud activity that appeared offshore Honshu island. This was confirmed by commercially available meteorological information from radar echo maps (Fig. 8e) showing the regional distribution of rain clouds. In this case, the rain clouds, which extended from the lightning region to the observation sites, might have caused the anomalous signals, thus the geoelectric signals on July 9 were labelled T. Using this procedure, we selected the anomalous signals, labelled S. Next, we compared the geoelectric signals from the EH method with the VLF band signals of electric-field change from the TAF method (Fujinawa & Takahashi 1990, 1994). One signal peak, SO, that appeared on April 6 could not be checked against VLF band signals because there were no records at that time. As shown in the typical VLF band signals at Hasaki in August and September 1996 in Fig. 7(a), the anomalous signals S1, S2 and S3 also appeared in the VLF band signal. Time-extended two-day displays of the signals S1 and S2 at the sites where those signals appeared simultaneously are shown in Figs 9(a) and (b), respectively. The anomalous signal patterns are very similar to each other and the time coincidence is good. Both signals S1 and S2 appeared from the night to early the next morning, continuing for 12 and 9 hr, respectively, while signal S3 appeared during the day and its duration was 7 hr. t C C E m o July 8 0 July 9, 1996 Figure 8. (a) The geoelectric signals at Tsukuba, 1996 July 8 and 9, (b) the lightning events from LPATS, (c) a lightning map on July 8, (d) a lightning map at 2:OO JST on July 9, and (e) the corresponding radar echo map showing the distribution of rain clouds. Arrows in (c) and (d) indicate the lightning positions. Geoelectric signals and seismic activity, Japan 491 Figure 8. (Continued.) It is noted that the diurnal geoelectric signals, which appeared at midnight, became notable, especially at Tsukuba in September in Fig. 4. The source is as yet unknown, but their intensities were less than the tentative threshold value of 8 in Fig. 4. The intensities of the signals SO-S3 were higher than those of the diurnal signal. Relationship between anomalous signal S and earthquake occurrence A total of four anomalous signals, SOGS3, that might not have been contaminated by industrial and meteorological activity were selected as the candidate seismic signals to be checked in relation to earthquake occurrence. No earthquake occurred in the short term after the signal SO appeared, but 18 days

8 Y.Enonzoto et al. (a) August 8-9,1996 (b) September 5-6,1996 [TSUKUBA 1 I I I Date Figure 9. (a) The candidate seismic geoelectric signal S1 appeared simultaneously on 1996 August 8-9 at Erimo, Tsukuba and Kyonan, but not at Kobe. (b) The signal S2 on September 5-6 appeared simultaneously at Tsukuba and Kyonan, but not at Erimo and Kobe. Corresponding lightning events from LPATS are shown at the bottom. afterwards, on 1996 April 23, the Iwate-ken Enganbu earthquake of M = 5.0 at a focal depth of 74 km occurred on 1996 April 23. It is unlikely that SO was related to the earthquake, because we did not find any anomalous signal similar to SO for the 1996 May 23 M = 5.0 Miyagi-ken Oki earthquake at a focal depth of 39 km, which is comparable to the Iwate-ken Enganbu earthquake in terms of its magnitude and the location of the epicentre. It is noted that the Akita-ken Nairiku-nanbu earthquake swarm of maximum magnitude M=5.9 at 3:12 JST, 1996 August 11 occurred 66 hr after the signal Sl appeared. The focal depth of the earthquake was 7 km. After the main shock, a total of 77 earthquakes with magnitudes ranging from 2.9 to 5.7 were felt in the same day. Furthermore, the Chiba-ken Toho-oki earthquake of 1996 September 11, M = 6.2 at 11 : 37 JST occurred 131 and 96 hr, respectively, after the appearance of the signals S2 and S3. TWO maps of the observation sites, Figs lo(a) and (b), indicate the sites where the signals S1, S2 and S3 appeared simultaneously on August 8-9 and September , respectively. Solid circles and triangles indicate that the anomalous signal S appeared and open circles and triangles that it did not. The distance to the epicentre of the Akita-ken Nairiku-nanbu earthquake was 430 km from Erimo and 320 km from Tsukuba, whereas for the Chiba-ken Toho-oki earthquake the distance to those observation sites shown by filled circles in Fig. 10( b) was less than 100 km. The M 6.2 Torishima Kinkai earthquake was not accompanied by an anomalous signal S within the time-window of a week. DISCUSSION By cross-checking the continuous recording of geoelectric current signals against commercially available meteorological information, we selected anomalous signals SO33 which were not affected by meteorological lightning or thundercloud activity. The anomalous signals selected correlate closely with the signals observed using the TAF method in the VLF band, except in the case of the signal SO on April 6, when the latter

9 Geoelectric signals and seismic activity, Japan 493 C) Aklta-ken Nairiku-nanbu Y=5.9, 7km 3:12, Aug.11 ii /- I ii la d f) Chiba-ken M km Toho-oki Tnlr.srhima I 19137, Sep. 11 -I maf W ayama 7 Figure 10. Maps of observation sites. Filled symbols show that the candidate seismic signal simultaneously appeared (a) on August 8-9 and (b) on September ; it did not appear at observation sites marked by open symbols. Circles: EH method; triangles: TAF method. The epicentres of the earthquakes possibly related to the candidate signals are indicated by stars. was not operating. All three anomalous signals S1 to S3, selected using the criteria shown in Fig. 6, appeared a few days before two earthquakes of M > 5.5, that is the M 5.9 Akitaken Nairiku-Naubu earthquake swarm, 1996 August 11, and the M 6.2 Chiba-ken Toho-oki earthquake, 1996 September 11. The present results are summarized as follows. Three events, Sl-S3, out of four selected anomalous signals SO-S3 fell within a time window, tentatively determined as one week before an earthquake occurrence of M > 5.5 at a focal depth of less than 100 km, while two out of three earthquakes of M > 5.5 were accompanied by anomalous signals within a week before the earthquake occurrence. These facts might be statistically significant for the signals Sl-S3 to be identified as the candidate seismic precursor signal, although rigorous statistical arguments cannot be addressed at present because of the lack of enough events to carry out such testing. The Torishima Kinkai earthquake, marked (e) in Fig. 5, which was not accompanied by anomalous signals, occurred about 200 km offshore Japan island, where any seismic electromagnetic signals that might have appeared would have been dispersed by conducting sea water. Another reason why we identified the selected anomalous signals Sl-S3 as candidate precursors is that the signals appeared only at the observation sites within a limited area from the epicentre, as can be seen in Fig. 10. At present, therefore, there is no plausible explanation for these signals, except that we believe the signals Sl-S3 to be candidate seismic electromagnetic precursors. It is noted that a signal appeared simultaneously on August km distant from the epicentre. The maximum magnitude of the event, M = 5.9, might not be large enough to generate an observable signal at a distance of km, but the swarm might have involved much energy and released pre-seismic electromagnetic signals at the precursor stage of the main fault movement, causing an electromagnetic disturbance over a wide area. The IASPEI 'guidelines' ask for discussion on the observed precursory signals in relation to stress, strain or some other mechanisms leading to earthquakes. Why such signals appear before the earthquake is still a major problem to be solved. In order to explain the generally observed seismic electric signals, current levels of the order of ka are required at the hypocentral region (Park, Strauss & Aceves 1996), which are one to two orders of magnitude larger than those predicted from models proposed so far, such as piezo-stimulated current or electrokinetic flow models (Park et al. 1996). Furthermore, in the piezoelectric or piezo-stimulated current model, which is usually used to explain seismic electromagnetic phenomena, a coseismic signal should be observed (Wyss 1996). A new model is therefore necessary. A tentative model proposed by Enomoto (1994, 1996), although qualitative, focused on why the signal appears before an earthquake. The thermal exo-electron emission and/or fracto-electron emission from fracturing rocks during the precursor stage of the main shock may electrify gases, which are released at depth. Such electrified gases may generate electric potential difference along the fault zone, leading to electric discharge, which produces the seismic electromagnetic signal. This model may be supported by recent laboratory experiments showing that electric charge can be transported by the electrification of flowing gases over a mineral surface undergoing abrasion (Scudiero, Dickinson & Enomoto 1997), and by field observations of anomalous remanent magnetization of fault gouge due to earthquake lightning accompanying the 1995 Kobe earthquake of magnitude 7.2 (Enomoto & Zheng 1997). In summary, more evidence of the candidate anomalous signals is needed in our work to discuss the statistical analysis of the candidate precursor signal, the causality of earthquake RAS, GJI 131,

10 494 Y. Enomoto et al. occurrence and the possibility of predicting imminent earthquakes in a rigorous manner. The present finding of candidate seismic electric signals, free from any industrial and meteorological noise, may be promising for further research and development to pursue the possibility of imminent earthquake prediction based on the detection of precursor electromagnetic anomalies. REFERENCES Casper, P.W., Recent improvements to the LPATS time-ofarrival lightning tracking system, IUGG General Assembly, Symposium MI, Vienna. Enomoto, Y., Geotribology of electronic anomalies prior to earthquakes, Jpn J. Tribology, 39, Enomoto, Y., Notes on generation and propagation of seismic transient electric signals, A Critical Review of VAN, pp , ed. Lightill, J., World Scientific, Singapore. Enomoto, Y. & Hashimoto, H., Anomalous electric signals detected before recent earthquakes in Japan near Tsukuba, in Electromagnetic Phenomena Related to Earthquake Prediction, pp , eds Hayakawa, M. & Fujinawa, Y., Terra Sci., Tokyo. Enomoto, Y. & Zheng, Z., Geological evidence of seismic electrical activity accompanied by the mag. 7.2 Kobe earthquake, 29th General Assembly of the IASPEI, Thessaloniki (abstract). Fujinawa, Y. & Takahashi, K., Emission of electromagnetic radiation preceding the Ito seismic swarm of 1989, Nature, 347, Fujinawa, Y. & Takahashi, K., Anomalous VLF subsurface electric field changes preceding earthquakes, in Electromnclznetic Phenomena Related to Earthquake Prediction, pp , eds Hayakawa, M. & Fujinawa, Y., Terra Sci., Tokyo. Geller, R.J., Debate on evaluation of the VAN method: editor s introduction, Geophys. Res. Lett., 23, Lighthill, J., Preface of a Critical Review of VAN, pp. v-vi, World Scientific, Singapore. Maeda, K. & Tokimasa, N., Decametric radiation at the time of the Hyogo-ken Nanbu earthquake near Kobe in 1995, Geophy,?. Res. Lett., 23, Oike, K. & Yamada, T., Relationship between shallow earthquakes and electromagnetic noises in the LF and VLF ranges, in Electromagnetic Phenomena Related to Earthquake Prediction, pp , eds Hayakawa, M. & Fujinawa, Y., Terra Sci., Tokyo. Park, S.K., Straws, D.J. & Aceves, R.L., Some observations about the statistical significance and physical mechanisms of the VAN method of earthquake prediction, Greece, in A Critical Review of VAN, pp , ed. Lighthill, J., World Scientific, Singapore. Scudiero, L., Dickinson, J.T. & Enomoto, Y., The electrification of flowing gases by mechanical abrasion of mineral surfaces, Phys. Chem. Miner., submitted. Wyss, M. (ed.), Evaluation of Proposed Earthquake Precursors, AGU, Washington, DC. Wyss, M., Brief summary of some reasons why the VAN hypothesis for predicting earthquakes has to be rejected, in A Critical Review of VAN, pp , ed. Lighthill, J., World Scientific, Singapore RAS, GJl 131,