Annals 47, 1, 2004, 01/07def 199 ANNALS OF GEOPHYSICS, VOL. 47, N. 1, February 2004 Mailing address: Dr. Nikolai T. Tarasov, Schmidt Uni- ted Institute of Physics of the Earth, Russian Academy of Sciences, ul. Bolshaya Gruzinskaya 10, D-242, GSP-5, 123995 Moscow, Russia; e-mail: tarasov@ifz.ru Key words earthquake – initiated – seismicity – electromagnetic – pulse 1. Introduction The Bishkek site is situated in the junction area of North Tien Shan and Chuya Basin struc- tures. Its southern part consists of Paleozoic crystalline rocks and the northern, of Mesozoic/ Cenozoic sedimentary deposits. The Tien Shan is a highly seismic region of the Earth. The north Tien Shan earthquake-generating zone, where the Bishkek site is situated, contains deep crustal faults. Three earthquakes with magnitudes great - er then 8.0 occurred there during the past 100 years. Deep electrical sounding of the crust was car- ried out at the Bishkek test site of the Institute of High Temperatures RAS in 1983 to 1990. The source of energy was the MHD-generator, and load was electrical dipole of 0.4 Ω resistivity, whose electrodes were 4.5 km apart. When the generator was put into operation, the load current was 0.28-2.8 kA, the sounding pulses had dura- tions of 1.7 to 12.1 s, the energy being mostly in the range 1.2-23.1 MJ. The dipole was installed within those Paleozoic crystalline structures in the north Tien Shan (42.69°N, 74.68°E), which Spatial-temporal structure of seismicity of the North Tien Shan and its change under effect of high energy electromagnetic pulses Nikolai T. Tarasov and Nadezhda V. Tarasova Schmidt United Institute of Physics of the Earth, Russian Academy of Sciences, Moscow, Russia Abstract The effect of high-energy electromagnetic pulses emitted by a magnetohydrodynamic generator used as a source for deep electrical sounding of the crust on spatial-temporal structure of seismicity of the North Tien Shan is ex- plored. Five-six years periodicity of changes in spatial distribution of seismicity was revealed. The effect of elec- tromagnetic pulses increases the stability of the spatial distribution of seismicity over time and simultaneously speeds up cycles of its transformations, which develop on stabilization background. Increasing of seismic ener- gy release after electromagnetic impacts is observed basically in most active zones. Periodic variation of effi- ciency of earthquakes triggering on the distance to the MHD-generator was detected. It was shown that electro- magnetic pulses give rise to an appreciable increase in the rate of local earthquakes, occurring around 2-6 days after the pulses. Total earthquakes energy released after start-ups was by 2.03⋅1015 J greater than the energy re- leased before them. At the same time, the total energy transmitted by the MHD-generator was 1.1⋅109 J, i.e. six orders of magnitude smaller. Consequently, the electromagnetic pulses initiated the release of the energy that had been stored in the crust due to activity of natural tectonic processes in the form of comparatively small earth- quakes, which leads to an additional release of tectonic stresses. 200 Nikolai T. Tarasov and Nadezhda V. Tarasova were adjacent to the boundary of Mesozoic/ Cenozoic sedimentary deposits in the Chuya Basin. During the time of the MHD operation, 114 start-ups were carried out. A detailed de- scription of the MHD equipment and the sound- ing procedure can be found in Volykhin et al. (1993). The observation area was within coordinates 41.0° to 45.5°N and 74.0° to 81.5°E. By the cri- terion of geological structure, the area can roughly be divided by latitude 42.9° into a south- ern part (the Paleozoic features of North Tien Shan) and a northern part (Mesozoic/Cenozoic sedimentary deposits). Analysis of seismicity of the observation area was based on the earthquake catalog for the North Tien Shan and adjacent ar- eas of N. Mikhailova containing earthquake data for the period from 1975 to 1996 (Mikhailova, 1990). The catalog contained data on 14 114 earthquakes with magnitudes ML from 0 to 6.2. The recurrence diagram based on all events is linear in the ML ranging from 2.2 to 6.1. The trigged seismicity was identified upon the background of the natural variation by co- herent summation of earthquakes in time win- dows of ± 20 days about the time of each of 114 MHD-generator start-ups (i.e. we superposed all 114 windows in time). For this purpose, se- lection of earthquakes was made for each of 114 MHD-generator runs in the interval (tj + − 20, tj + 20), where tj is absolute times (in days) of the jth MHD-generator start-up. Ab- solute times Ti of all selected earthquakes (more than 4000) were replaced by the times tij = Ti + − tj measured from the time of the correspon- ding start up (here i is number of earthquake in the selection). Obtained selection was used to find Nt, which is the number of events occurring in the area during each day within time win- dows, i.e. those with t − 0.5 ≤ tij ≤ t + 0.5 for all integer t between − 20 and 20 days. Values Nt with t > 0 were analyzed to study the effect of electromagnetic pulses of the MHD- generator on seismicity over time. The events oc- curring during 20 days before the start-up were used to estimate the natural seismicity rate imme- diately before MHD-generator runs. We calculat- ed background level as averaged daily number of events and estimated the 99% confidence interval over all Nt with t < 0. For investigation of total seismic energy changes after MHD-generator runs, we calculated the ratio of the sums of the energy of earthquakes with tij > 0 and tij < 0, re- spectively, using the relationship for earthquake energy estimation presented in Rautian et al. (1981). Note, this technique allows one to in- crease the relation of signal/noise n times, where n is the number of the superposed time windows. This brings the relation up to 10.7 when 114 start ups of the MHD-generator are used. For study of spatial variations of total seismic energy changes after the MHD-generator runs, the observation area was divided into 1000 km2 cells of the grid. The ratio of the sums of the en- ergy of earthquakes with tij > 0 and tij < 0, re- spectively, was calculated for each cell. The map of distribution energy ratios was created using all cells with non-zero earthquake number. A simi- lar technique was used for creation of the map of the flow of total seismic energy over the area, but the total catalog was processed for this purpose instead of the selection made. For a quantitative estimation of variations of a spatial distribution of the earthquakes over time, the spatial correlation of earthquakes density for each two contiguous years rk as a function of the time from 1975 to 1995 was calculated. To do this, the region of observations was divided into square 1000 km2 cells of the grid too. nijk and nijk + 1 are numbers of the earthquakes occurred in- side each cell for kth and for k + 1th years, re- spectively, were calculated for all k varying from 75 to 95 year. The indices i, j denote the numbers of a grid cell for latitude and longitude. They are related with geographical coordinates X, Y as ,i X X dX j Y Y dYInt Int0 0= - = -^ ^h h where dX, dY are the length of sides of the cell on axes X, Y in degrees, and X0 = 41°N and Y0 = = 74°E are latitude of southern and longitude of western boundaries of the observation area. The function of spatial-temporal correlation of seis- micity rk was evaluated for all k = 75, 76, ..., 95 using the expression r S S C , k k k k k 1 1 2 $ = + + 201 Spatial-temporal changes of seismicity under effect of high energy electromagnetic pulses where C n n N M n n n N M n n N M n 1 1 1 ,k k ijk ijk j M i N ijk j M ijk j M i N i N k ijk j M i N ijk j M i N k ijk j M i N ijk j M i N 1 1 11 1 1 111 2 11 11 2 1 1 2 11 1 11 2 $ $ $ $ $ $ = -+ + == = + === == == + + == + == S = - $ S = - !! ! !!! !! !! !! !! d _ d _ d n i n i n N is number of cells along the axis X, and M is number of cells on the axis Y. 2. Results First, we examined the time variation in seis- micity after the MHD-generator runs. Figure 1a presents the distribution of daily rates of earth- quakes Nt for the entire area before and after MHD-generator runs. The values of Nt are seen to be appreciably higher one day after the start- up than before it. The maximum rate occurred on the second day after start-up with amplitude nearly 1.7 above background level. When only earthquakes in the north part are considered, no significant changes in Nt related to MHD-gener- ator runs can be detected (fig. 1b), whereas a well-pronounced maximum can be seen in the south (fig. 1c) occurring on the second day after electromagnetic action, nearly 2.2 above the mean background, to be followed by a gradual falloff in Nt during a few days. The triggering effect of electromagnetic puls- es of MHD-generator was most active in the upper 5 km layer. Figure 1d shows the distribution of the earthquakes occurring within the layer in the south part of the area. It is seen that during the first day after MHD-generator runs, the Nt values re- main at the background level and then a strong ac- tivization of the regime occurs that goes on for about 5 days. During this period, Nt exceeds back- ground level 2.7 times. The maximum is observed on the second day after MHD-generator start-up. Its amplitude exceeds 99% confidence level. The above results show that high energy elec- tromagnetic pulses emitted by MHD-generator give rise to an appreciable increase in the rate of 50 150 250 50 150 250 -15 -10 -5 0 5 10 15 50 150 150 250 350 N t, days Fig. 1a-d. Daily number of earthquakes in the area of the Northern Tien Shan and adjacent territories be- fore (t < 0) and after (t > 0) the MHD-generator runs on the Bishkek test site plotted against time: a) for the entire area, b) its northern and (c) southern (North Tien Shan) part, d) events occurring in the up- per 5 km layer of the southern part of the region. The vertical axis shows the daily rate of earthquakes; hor- izontal axis shows the time measured from the MHD start-ups (in days). The dashed lines indicate a mean background level (lower lines) and 99% confidence interval (upper lines). local earthquakes, occurring about 2-6 days after the pulses. It confirmed the initiating effect of ac- tion by electromagnetic pulses on the seismicity previously revealed (Tarasov, 1997; Tarasov et al., 1999) in the Garm and Bishkek areas. a b c d 202 Nikolai T. Tarasov and Nadezhda V. Tarasova 75 76 77 78 79 80 81 42 43 44 45 4 5 6 7 8 9 10 11 12 13 14 N E Lg E Fig. 2. Map of the flow of total seismic energy of earthquakes in Northern Tien Shan and adjacent territories obtained using all earthquakes of the catalog for 1000 km2 cells of grid. The values of energy are reduced to the flow for 1 year. 75 76 77 78 79 80 81 42 43 44 45 -4 -3 -2 -1 0 1 2 3 E N Lg Ea Eb Fig. 3. Map of total seismic energy ratio of earthquakes in Northern Tien Shan area and adjacent territories for 20 days after and 20 days before MHD-generator runs. 203 Spatial-temporal changes of seismicity under effect of high energy electromagnetic pulses On the one hand it can be assumed that this effect results from the extra energy loaded to the Earth by electromagnetic pulses, which is radiated in the form of earthquakes. Another ex- planation is that sounding pulses initiated the release of energy stored in the crust due to nat- ural geodynamic processes. To test these as- sumptions, we estimated the total seismic ener- gy released during 20 days before all MHD- generator start-ups Eb, total seismic energy re- leased during the same period after the start-ups Ea, and the post-start-up increase in total seis- mic energy release Ea-Eb. These estimates were based on all earthquakes of the area, on events occurring in the north and south, regardless of their focal depths, and the events from the up- per 5 km layer. An analysis of the obtained values shows that the total earthquake energy released in the area during 20 days after all start-ups was by 2.03⋅1015 J greater than the energy released for a period of the same duration before them. Most of the energy release was due to the earth- quakes occurring in the upper 5 km layer in the southern part of the area, which belongs to the crystalline rocks of the North Tien Shan. At the same time, the total energy transmitted by MHD-generator to the radiating dipole for all start-ups was 1.1⋅109 J, i.e. six orders of magni- tude smaller. Consequently, the high energy electromagnetic pulses radiated by the MHD- generator initiated the release of the energy that had been stored in the crust due to other sources. The mean energy of initiated earthquakes per one start-up was 2.2⋅1013 J. Tarasov (1997) investigated the initiating effect of MHD-gener- ator pulses on the seismicity of the Garm area in Tajikistan and concluded that the average in- crease in total energy release through local seis- micity was 1.1⋅1012 J per start-up. However it should be kept in mind that Garm site has an area much smaller than our area under study. When the estimates of energy were divided by the values of respective areas, the extra seismic energy release was 7.3⋅107 J/km2 per start-up in the Bishkek site and 6.6.107 J/km2 per start-up in the Garm site. These similar values indicate approximately similar responses of the Earth to electromagnetic pulses in different areas. Figure 2 presents the map of the flow of to- tal seismic energy of earthquakes of North Tien Shan and area of adjacent territories. The densi- ty of seismic energy flow is seen to vary wide- ly over the area. There are several zones of high energy flow in the middle and south, while en- ergy flow density in the northwest is signifi- cantly lower. Figure 3 shows the spatial variations in the changes of total seismic energy after MHD- generator start-ups. The map demonstrates that seismic energy release increasing takes place within a number relatively small zones located predominantly in southern and central parts of the observation area. One can see some correla- tion of maps in figs. 2 and 3. However, these maps are not identical. Therefore, electromag- netic effects cause the increase in seismic ener- gy release mainly in most active zones. Never- theless a triggering effect is determined not on- ly by a level of seismic activity, but also by oth- er phenomena. Using this map, it is difficult to recognize any regular dependence of efficiency of trigger effects of electromagnetic pulses on the dis- tance to location of the MHD-generator (42.69°N, 74.68°E). For detailed analysis, we calculated average Ea /Eb as function of the dis- tance from location of the MHD-generator to epicenters of earthquakes, irrespective of az- imuths. Using all MHD-generator runs, the summary seismic energy of earthquakes which occurred 20 days before (Eb) and 20 days after (Ea) electrical soundings were calculated for ring bands bounded by radii R and R + 50 km with the center in the site of location of the MHD-generator. Eb and Ea were computed in the distance range from 0 up to 600 km with 50 km step. Shown in fig. 4, the dependence of Eb on the distance to MHD-generator represents a natural change of the flow of seismic energy in the ob- servational region. Similar dependence of Ea shows summary effect of natural changes and trigged seismicity. Note, relatively low level of Eb is observed up to distances of ∼ 200 km. One can see in the figure that the total seismic ener- gy before MHD-generator runs was higher than after them for distances R < 110 km. Then, the interval where Ea is greater than Eb is observed. 204 Nikolai T. Tarasov and Nadezhda V. Tarasova -1.5 -0.5 0.5 1.5 2.5 0 100 200 300 400 500 R, km E E b aLg Further, the position of curves relative to one another changes, and so on. More clearly it is possible to see in fig. 5 where the dependence of Lg (Ea /Eb) being the ratio of the total seismic energy of earthquakes after and before MHD-generator start-ups on the distance is presented. The values Lg (Ea /Eb) change the sign periodically through every 80- 110 km. Hence, bands where the electrical ef- fects increases the flow of seismic energy and bands where the electrical effects diminishes the seismic flow, alternate periodically with in- creasing the distance to the MHD-generator. It is interesting that Lg (Ea /Eb) < 0 at the short distance (less than 100 km). Therefore, the electromagnetic pulses diminish seismic activi- ty inside near the field area of the electrical di- pole. At distances of 110-220 km, the seismici- Fig. 5. The ratio Ea /Eb of the total seismic energy of earthquakes of the North Tien Shan and adjacent areas occurring after and before 114 MHD-generator start-ups as a function of the distance to location of the MHD- generator on Bishkek test-site. The values Ea /Eb are plotted on a logarithmic scale. 10.5 11.5 12.5 13.5 14.5 0 100 200 300 400 500 R, km Lg E Fig. 4. The total seismic energy of earthquakes of the North Tien Shan and adjacent areas occurring before Eb (dashed line) and after Ea (solid line) 114 MHD-generator start-ups as a function of the distance to location of the MHD-generator on Bishkek test-site. The values Ea and Eb are plotted on a logarithmic scale. 205 Spatial-temporal changes of seismicity under effect of high energy electromagnetic pulses ty increases after MHD-generator start-ups, fur- ther, the curve goes to the negative domain again, etc. As a whole, the sign-variable de- pendence is observed, whose amplitudes in- crease up to a distance of 375 km to electrical dipole of the MHD-generator and, then, the am- plitudes begin to decrease. It is important to note that the amplitudes of positive half-periods of the dependence are higher than the amplitudes of negative half-periods. This confirms that the effect of the electromagnetic pulses causes sub- stantial growth of seismic energy of the region in the whole. For a quantitative estimation of variations of a spatial distribution of the earthquakes over time, the spatial correlation (rk) of earthquakes density for each two contiguous years as function of the time from 1975 to 1995 year was calculated us- ing the technique discussed in the first part of the paper. This function is presented in fig. 6. First of all, it is interesting that for the period of 1975- 1983 the periodic changes of the function are clearly observed. For this time interval, value rk has rather low values. The correlation regularly varies from 0.1 to 0.45 with a period of 5-6 years. It shows that there are periods of «fast» changes in spatial distribution of seismicity, which are re- placed by periods of its relative stability every other 2.5-3 years and vice versa. One can clearly see in fig. 6 that the func- tion abruptly changes the shape since 1983. The 5-6 year periods of variations in correlation are changed by periods of 3 year duration. At the same time, a significant increase in correlation factors rk up to 0.3-0.9 is observed. Thus, the spatial distribution of earthquakes becomes more stable after 1983. As the time of change in frequency of periodicity and in increase in val- ues rk coincides with the beginning of experi- ments with MHD-generator, it is possible to as- sume that the action of electromagnetic pulses causes an increase in stability of spatial distri- bution of seismicity over time and simultane- ously speeds up cycles of its transformations, which develop on stabilization background. Interestingly, the variation of values of frac- tal dimension shows that the earthquakes tempo- ral distribution becomes more random for time period of experiments with the MHD-generator (from 1983 to 1990) (Chelidze et al., 2002). Figure 7 shows a similar dependence based on the selection of earthquakes occurring in the upper 5 km layer of the crust. Just at the begin- ning of observations time interval, the rather high values of a correlation function are ob- served on the curve. Further, within 6 years from 1977 to 1983, the values of rk do not ex- ceed 0.23. At the beginning of experiments with 0 0.2 0.4 0.6 0.8 75 80 85 90 t, year r k Fig. 6. Spatial-temporal correlation of density of earthquakes of the North Tien Shan and adjacent areas as function of time. The vertical axis indicates the values of a correlation function rk; the horizontal axis indicates time in years. The vertical points are time of beginning and termination of a series of MHD-generator start-ups on the Bishkek test-site. 206 Nikolai T. Tarasov and Nadezhda V. Tarasova the MHD-generator, the level of spatial-tempo- ral correlation sharply increases. Thus, in 1983, rk reaches 0.35, and, further, the values oscillate within the range 0.42-0.97. The 3-4 year perio- dicity is observed for this time again. These re- sults confirm that the most appreciable changes in spatial-temporal structure of seismicity after the beginning of experiments with a MHD-gen- erator are observed in the upper 5 km layer. Previously, we showed that the most inten- sive increase in total seismic energy of earth- quakes after MHD-generator start-ups was ob- served at the same depths not only in the North Tien Shan, but also in the Garm region of Tad- jikistan (Tarasov, 1997; Tarasov et al., 1999). Because the initiating action was most effective in the depth range 0-5 km which is the hydro- static zone of the crust containing free water that fills cracks, pores and cavities, it was con- cluded that fluids are largely responsible for the initiation of seismicity by electromagnetic puls- es. The presence of fluids in rocks gives rise to the seismoelectric and electroseismic effects. To test this assumption, we studied the changes in efficiency of the earthquakes trig- gering by electromagnetic pulses depending on a level of atmospheric precipitates. First, the values Ea /Eb obtained on each of 114 start-ups were compared to an average level of precipi- tates before each of them. To average the pre- cipitate level before the start-ups, the time win- dows of different duration from 1 to 20 day were applied. However, a significant correlation between the flow of seismic energy and level of precipitates was not found. Comparison of long-period changes of Ea/Eb with average annual level of precipitates was more successful. To do this, the values of average annual level of precipitates were calcu- lated for each year of experiments. For the same period, we computed the ratios Ea/Eb of total seismic energy of the earthquakes which oc- curred in layer 0-5 km after and before the MHD-generator start-ups within each year. As the intervals between start-ups varied for differ- ent years, the time windows for Ea/Eb calcula- tion were limited by 5 days. Tarasov et al. (1999) have shown that it reduces the noise lev- el which arises because of partial superposition of time windows of different start-ups when they are executed too often. Figure 8 represents the variations Ea/Eb de- pending on time for 1983-1990 years. The changes in the annual level of atmospheric pre- cipitation in Bishkek are shown for the same period in the fig. 9. One can see that these curves are correlated. The correlation factor in this case was equal to 0.845, though this esti- 0 0.2 0.4 0.6 0.8 75 80 85 90 r k t, year Fig. 7. Function of a spatial-temporal correlation of density of earthquakes occurringin the upper 5 km layer of crust of the North Tien Shan and adjacent areas. The vertical axis indicates the values of a correlation function rk; the horizontal axis indicates time in years. The vertical points are the time of beginning and termination of a series of MHD-generator start-ups on the Bishkek test-site. 207 Spatial-temporal changes of seismicity under effect of high energy electromagnetic pulses mation should be assumed as preliminary, be- cause we have only very small statistics. More clearly it could be seen in fig. 10 where points represents the ratio of total seis- mic energy after and before all MHD-generator start-ups for each of 8 years as a function of the annual level of precipitates for the related year. The fitted line obtained by the mean square method is also given. It is possible to see the values Ea /Eb increase depending on the level of precipitates. These results confirm the assump- tion that fluid is largely responsible for initia- tion of earthquakes by electromagnetic pulses. The dependence of efficiency of earthquakes triggering by electromagnetic pulses at a hypo- center depth shows that this can depend on pulse duration, as this parameter determines the thick- ness of the skin layer. Figure 11 represents Ea /Eb as a function of sounding pulse duration. To cal- culate the diagram, all start-ups of the MHD-gen- erator were divided into a number of subse- quences with respect to pulse duration. The total seismic energy of the earthquakes that occurred before and after pulses of some duration was evaluated using all start-ups of MHD-generator included in the relevant subsequence. In the figure we do not observe any regular increase or decrease in efficiency of electromag- netic effects with the pulse duration. One can see that the dependence has sign-variable char- acter. The shortest (less than 1.5 s) and longest (more than 12 s) pulses cause a decrease in total -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 83 84 85 86 87 88 89 90 L g (E a /E b ) t, year 0.7 0.9 1.1 1.3 1.5 1.7 83 84 85 86 87 88 89 90 L , m m /d ay t, year Fig. 8. Average annual values of Ea / Eb being the ratio of total seismic energy of earthquakes of the North Tien Shan occurring 5 days after and 5 days before MHD-generator start-ups in the upper 5 km layer as a function of time. The values Ea / Eb are plotted on a logarithmic scale. Fig. 9. Averaged annual level of atmospheric precipitation in Bishkek as a function of the time. 208 Nikolai T. Tarasov and Nadezhda V. Tarasova seismic energy of earthquakes. The 6.75-9.1 s duration pulses cause a similar effect. The high efficiency is observed for pulses of duration ∼ 2 s, 3.5-6.7 s and 9.4-11.8 s. The sign-variable character of the curve suggest that layers where the electrical effects increases the flow of seis- mic energy and layers where the electrical im- pact decreases them alternate with depth. 3. Discussion Changes in seismicity could be connected with the influence of some external factors. It is shown for the Central Asian region (Nikolaev and Vereshagina, 1991a,b; Tarasov and Taraso- va, 1995) that mb > 5 earthquakes and under- ground nuclear explosions significantly influ- Lg(E a /E b) = 3.708 L – 3.5766 r = 0.845 -1 0 1 2 3 4 0.8 1 1.2 1.4 1.6 1.8 L g( E a /E b ) L, mm/day -4 -3 -2 -1 0 1 2 1 3 5 7 9 11 x, s E E b aLg Fig. 10. Averaged annual values of Ea / Eb being the ratio of total seismic energy of earthquakes of the North Tien Shan and adjacent areas occurring 5 days after and 5 days before MHD-generator start-ups in the upper 5 km layer plotted as function of average annual level of atmospheric precipitation L in Bishkek site (dots) and fit- ted line. The values Ea / Eb are plotted on a logarithmic scale. Fig. 11. Ratio Ea /Eb of total seismic energy of earthquakes of the North Tien Shan and adjacent areas occur- ring after and before MHD-generator start-ups as a function of the electromagnetic pulses duration. The values Ea/ Eb are plotted on a logarithmic scale. 209 Spatial-temporal changes of seismicity under effect of high energy electromagnetic pulses ence seismicity at a distance up to 1500 km. The influence of geomagnetic storms on seis- micity of the observation area is shown in (Tarasov and Tarasova, 2002; Sobolev and Po- nomarev, 2003). Thus, activation of remote earthquakes, nuclear tests or increasing geo- magnetic activity before MHD-generator start- ups could be the reason for the observed in- creasing in seismicity. In order to check this assumption, the selec- tion of mb > 5 earthquakes, which occurred within the time window ± 30 days about the time of each of 114 MHD-generator runs at dis- tances up to 1500 km from the center the region was made from the World Earthquake Cata- logue of ISC. As in a previous study, obtained selection was used to find Nt, which is the total daily number of remote earthquakes before and after all 114 MHD-generator start-ups (fig. 12a). Estimations of the average background level and 99% confidence interval were made for all earthquakes with mb > 5, which occurred from 1983 to 1990 within the distance range of 1500 km. In the same way we obtained daily numbers of underground nuclear explosions (fig. 12b). For analysis of geomagnetic activity we drew up the curves of changes in average daily values of Dst index (fig. 12c) and Kp index (fig. 12d), which were averaged for all of 114 time windows. Figure 12a-d shows that before MHD-generator start-ups (t < 0) there is no anomalous increase in the number of remote earthquakes or nuclear explosions, as well as anomalous changes of geomagnetic activity. Therefore the observed effect is not connected with the influence of these factors. Some articles, for example (Oike and Ya- mada, 1994), detected a correlation between shallow seismicity and atmospheric thunder- storm activity. Unfortunately we have no infor- mation on thunderstorms in the region. In order to estimate this factor all MHD-generator runs were divided into two groups according to sea- sons: from March to August and from Septem- ber to February. In the observation area, the first period is characterized by high thunder- storm activity, in the second period there are no thunderstorms. However, a significant increase in seismic activity after MHD-generator start- ups was observed for both groups. It means 0 2 4 6 8 t, day -20 -15 -20 -10 0 10 20 2 3 0 5 10 15 20 N Dst Kp Fig. 12a-d. Daily number of mb > 5 remote earth- quakes (up to 1500 km) in the area (a) of the North- ern Tien Shan and adjacent territories, b) of semi- palatinsk underground nuclear explosions, averaged (c) Dst and (d) Kp geomagnetic indices before (t < 0) and after (t > 0) 114 MHD-generator runs on the Bishkek test site plotted against time (in days). The dashed lines indicate a mean background level and 99% confidence interval. that the effect is not connected with thunder- storm activity. One more reason for activation of seismic- ity could be the mechanical impact of the MHD-generator on the ground surface. Unfor- tunately it was impossible to check this hy- pothesis in this region. However, such estima- a b d c 210 Nikolai T. Tarasov and Nadezhda V. Tarasova tion has been made for analogous experiments in Garm region of Tadjikistan (Tarasov, 1997). Along with the deep electrical sounding seis- mic soundings of the crust were executed by explosions of 400 kg of explosive (Gamburt- sev et al., 1983) (within the period from 1979 to 1984 there were executed 276 explosions). Figure 13a,b show changes in the daily number of earthquakes before and after MHD- generator start-ups in Garm region (from Tarasov, 1997), and fig. 13c shows changes of Nt before and after explosions in the same area. It is clearly seen that the flow of seismic events grows after the MHD-generator start- ups. However, unlike nuclear explosions, rela- tively weak chemical explosions do not cause noticeable changes in seismicity. While their impact on the ground surface is much stronger than the mechanical impact of the MHD-gen- erator, it is possible to conclude that the ob- served effect is caused by its electromagnetic, but not mechanical impact. It is interesting that in Garm region after electromagnetic impacts the most long-stand- ing activation of seismicity was observed (fig. 13a-c), beginnning 6 days after experiments, but not after 2 days as in fig. 1a-d. Such delay of activation after the time of impact shows that electromagnetic pulses initiate some process, which causes an increase in seismicity, and the time of its development depends on rock prop- erties. A similar dependence was observed in Garm region after remote nuclear explosions (Tarasov and Tarasova, 1995). It is possible that rock properties determine not only the time of delay, but influence the ef- ficiency of effect. It was shown above that in Paleozoic structures of the southern part of the region a strong activization arises after MHD- generator runs, while in Mesozoic and Ceno- zoic structures of its northern part there is prac- tically no increase in seismicity. This fact could have the another explanation. In connection with the geoelectric cross-sections of the obser- vation area presented in Volykhin et al. (1993), it is possible to assume the existence of a high conductive layer, which plunges from the MHD-generator site to the south and is a kind of channel which directs most of the pulse en- ergy in that direction. However, the efficiency of the initiating ef- fect of electromagnetic pulses is influenced by other, more complicated factors too. Periodic changes in Ea/Eb with an increased distance from the MHD-generator, shown in fig. 5, could be connected with the block structure of the crust. Estimations of prevailing dimensions of the crust blocks for Tien Shan give 70 and 140 km (Sadovsky and Pisarenko, 1989), which is close enough to periods of change in Ea/Eb in fig. 5. The other explanation of such periodicity could be in the long scale anticorrelation of rock properties (Sahimi et al., 1993). At this stage, the physical mechanism of the process of initiation of seismicity by electro- 0 5 10 15 -20 -10 0 10 20 1400 1600 1800 2000 200 300 400 N t, day Fig. 13a-c. Daily number of earthquakes in the Garm region of Tadjikistan before (t < 0) and after (t > 0) 34 MHD-generator runs on the Garm test site plotted against time for the entire area (a) and for upper 5- km layer of the Tadjik Depression (b), daily number of earthquakes in the same area of Tadjikistan before (t < 0) and after (t > 0) 276 local chemical explosions of 400 kg explosive (c). The dashed lines indicate a mean background level (lower lines) and 99% confi- dence interval (upper lines). a b c 211 Spatial-temporal changes of seismicity under effect of high energy electromagnetic pulses magnetic pulses is not clear. Now it is difficult to explain with confidence the observed spatial- temporal characteristics of this effect. However, observed phenomena show new properties of triggered seismicity and could be useful for in- vestigation of a process of earthquake initiation. 4. Conclusions It has been shown that 2-6 days after effect of high energy electromagnetic pulses emitted by the MHD-generator there is activation of rel- atively small earthquakes in the region under study. The total seismic energy of initiated earthquakes is six orders higher than the energy transmitted by the MHD-generator to the radi- ating dipole. Efficiency of the initiating effect of elec- tromagnetic pulses depends on the natural level of seismic activity over the region, electromechanical characteristics of rocks, depth and distance to dipole. A 5-6 years periodicity of spatial-temporal correlation of earthquake density was detect- ed indicating the existence of periods of «fast» changes in spatial distribution of seis- micity over time, which alternate every 2.5-3 years by periods of relative stabilization. It is shown that the effect of electromagnetic puls- es increases the stability of spatial distribution of seismicity over time and simultaneously speeds up cycles of its transformations, which develop on stabilization background. The results of this study show that the ef- fect of high energy electromagnetic pulses ra- diated by MHD-generators causes substantial spatial-temporal changes in the seismicity of earthquake source zones and accelerates the release of energy stored in the crust due to the activity of natural tectonic processes, thus, serving as a kind a trigger. Based on energy balance assumption for the crust, it can be concluded that a man-made in- crease in part of the seismic energy radiated in the form of flow of relatively small earthquakes leads to an additional release of tectonic stress- es, thereby diminishing the likelihood of cata- strophic events (or at any rate, reduces the en- ergy of such events). If such assumption is confirmed, the detected effect could be used to develop a technique of re- duction of seismic hazard by artificial discharge of tectonic stress by electromagnetic pulses. Acknowledgements We are grateful to Kostas Eftaxias for useful advice and discussion. This work was support- ed by INTAS-99-0064 grant. REFERENCES CHELIDZE, T., V. DE RUBEIS, T. MATCHARASHVILI and P. TOSI (2002): Influence of strong electromagnetic discharges on the dynamics of earthquakes time distribution at the Bishkek test area (Central Asia), in Proceedings of ESC XXVIII General Assembly, Genova 2002. GAMBURTSEV, A.G., N.G. GAMBURTSEVA, S.P. STARO- DUBROVSKAJA and V.N. FRIDMAN (1983): Recording waves and seismic model of the Earthcrust on the Dushanbinskiy test site, in Seismic Sounding of Source Zones, edited by A.V. NIKOLAEV (Nauka, Moscow), 109-122. MIKHAILOVA, N.N. (1990): Catalog of Earthquakes in North Tien Shan and Adjacent Area (Nauka, Alma- Ata), pp. 224. NIKOLAEV, A.V. and G.M. VERESHAGINA (1991a): About triggering of earthquakes by earthquakes, Dokl. Acad. Nauk SSSR, 318 (2), 320-324. NIKOLAEV, A.V. and G.M. VERESHAGINA (1991b): About triggering of earthquakes by underground nuclear ex- plosions, Dokl. Acad. Nauk SSSR, 319 (2), 333-336. OIKE, K. and T. YAMADA (1994): Relationship between shal- low earthquakes and electromagnetic noises in LF and VLF ranges, in Electromagnetic Phenomena Related to Earthquake Prediction, edited by M. HAYAKAWA and Y. FUJINAWA (Terra Scientific Publishing Co., Tokyo), 115-130. RAUTIAN, T.G., V.I. KHALTURIN, M.S. ZAKIROV, A.G. ZEMTSOVA, A.P. PROSKURIN, B.G. PUSTOVITENKO, A.N. PUSTOVITENKO, L.G. SINELNIKOVA, A.G. FILINA and I.S. SHENGELIA (1981): Experimental Exploration of Seis- mic Coda (Nauka, Moscow), pp. 141. SADOVSKY, M.A. and V.F. PISARENKO (1989): Discontinuous hierarchical model of the geophysical medium, in Mul- tidisciplinary Studies in Terrestrial Physics, edited by S. NEGMATULAEV and V. FREMD (Nauka, Moscow), 9-26. SAHIMI, M., M. ROBERTSON and C. SAMMIS (1993): Fractal distribution of earthquake hypocenters and its relation to fault patterns and percolation, Phys. Rev. Lett., 70, 2186-2189. SOBOLEV, G.A. and A.V. PONOMAREV (2003): Physics of Earthquakes and Precursors (Nauka, Moscow), pp. 270. TARASOV, N.T. (1997): Crustal seismicity variation under elec- tric action, Dokl. Acad. Nauk SSSR, 353 (4), 542-545. TARASOV, N.T. and N.V. TARASOVA (1995): Response of 212 Nikolai T. Tarasov and Nadezhda V. Tarasova seismoactive medium to nuclear explosions, in Earth- quakes Induced by Underground Nuclear Explosions, edited by R. CONSOLE and A. NIKOLAEV, NATO ASI Se- ries 2. Environment (Sprunger-Verlag, Berlin Heidel- berg), 4, 215-223. TARASOV, N.T. and N.V. TARASOVA (2002): The effect of ge- omagnetic storms on the seismicity, in Proceedings of Third International Workshop on Magnetic, Electric and Electromagnetic Methods in Seismology and Vol- canology, Moscow, Russia, 3-6 September 2002 (Insti- tute of Physics of the Earth, Moscow), 206-208. TARASOV, N.T., N.V. TARASOVA, A.A. AVAGIMOV and V.A. ZEIGARNIK (1999): The effect of high energy electro- magnetic pulses on seismicity in Central Asia and Kazakhstan, Volc. Seismol., 4-5, 152-160. VOLYKHIN, A.M., V.D. BRAGIN, A.P. ZUBOVICH, N.L. AITOV, J.P. BABAKOV, V.J. BATALEV, A.G. BLOKH, E.L. GELLER, V.A. ZEIGARNIK, E.I. IVANOV, J.I. ISAEV, N.A. KOSHKIN, R.K. KUZMIN, T.I. KULEVTSOVA, A.I. MATIX, N.N. OR- LENKO, A.V. PISAKIN, A.N. REVTOV, M.F. RETINSKY, A.R. RUNG, A.K. RIBIN, B.G. TKATCHENKO, J.A. TRA- PEZNIKOV, P.S. TUROVSKY, V.V. FINIAKIN, E.V. SHARONO- VA and A.P. SHISHKINA (1993): Geodynamic Processes in Geophysical Fields (Nauka, Moscow), pp. 158.