D e t e r m i n i n g E n e r g y o f Elastic w a v e s Caused by E a r t h q u a k e (*) V . L . B E L O T E L O V , N . "V. K O N D O R S K A Y A , E . T H . S A V A R E N S K Y Rieevuto il 7 dicembre 1960 The recent achievements in seismology and the development of the USSR seismic station network make it possible to determine the absolute value of the energy of elastic oscillations radiated by a source, on the basis of the method suggested by B. B. Galitzin in 1915 (l) and developed by H. Jeffreys (2). To solve this problem, it is necessary to determine the changes in the density of elastic wave energy with distance, and to work out methods for determining the energy of non- stationary oscillations. The present communication represents the contents of two ar- ticles (3'4), connected with the question of determination the absolute value of the energy of longitudinal and transverse waves. The value of the energy radiated by the source is that of the energy of body waves in the period range of 2 to 10 sec. where standard equip- ment of D. P. Kirnos type has constant magnification. In estimating elastic energy radiated by a source we shall proceed from the following assumptions: 1. The flux of the energy transmitted from the source to the earth surface is directed along the rays. 2. The variations in the energy flux due to reflection and refrac- tion at the intermediate underground boundaries are of the same order as the errors in determining the oscillation energy, and only the effect of the bottom of the Earth's crust and the Earth surface is taken into account. 3. The duration of elastic oscillations in P and S phases radiated by the source does not change in propagating towards the Earth surface. (*) P a p e r read a t the Helsinky Assembly of the I . U . G . G . , 1960. 5 6 V . L. B E L O T E L O V , N . V . K O N D O R S K A Y A , E . TIL. S A V A R E N S K Y In this paper the term " earthquake energy " will signify the energy of elastic body waves radiated by the earthquake source. This energy we represent as the sum of energies of longitudinal (Ep) and transverse (Es) waves radiated from the source. E = Ep Es The calculation of the energy of P or S waves from the data of individual stations has been carried out by means of the following formulas: 4 n R2 sin 6 sin e0 ( E0 \ E = —— - e ) • P o f(e,a) cos e de dd QC E0 = oc d AN Y / DA,: dt ) + \ dt K 2 v dAjV dt ) K/ dt, where 0 = epicentral distance, e = angle of emergence of the seismic ray from the earthquake source, k — coefficient of energy absorption in the Earth, c = velocity of the incident P or 8 wave near the Earth surface, o = rock density in the vicinity of the seismic stations, E0 = density of oscillation energy in the incident wave at the point of obser- vation, e0 = angle of emergence of the seismic ray to the earth surface, AK, AE, AZ = components of true ground displacement, at the earth surface iTv, Kz = coefficients of reflection at the free interface for the horizontal and the vertical component respectively, / (e, a) function taking into account the mechanism of the earthquake source. Energy determinations were carried out for 11 earthquakes recorded by seismic stations of the USSE. The co-ordinates of their epicentres were determined on the basis of the time of arrival of 8 and P waves at Soviet and foreign seismic stations. The basic data on the sources are given in Table 1. The geographical distribution of the epicentres is presented in Fig. 1. 112 seismograms with the most distinct recordings were selected, which served as basic material for the determination of P and 8 wave energy. These seismograms represented the records obtained by Soviet instruments " CTK " and " CBK " (records made by B. B. Galitzin- type instrument were used only in two cases). In calculating the oscillation energy we first selected the oscillation groups corresponding to P and 8 waves in the seismograms. We cal- D E T E R M I N I N G E N E R G Y OK E L A S T I C W A V E S C A U S E D B Y E A R T H Q U A K E 5 7 ciliated the energy at the point of observation for the time interval corresponding to these selected groups. The arrival of the respective wave was assumed to be the beginning of the group, its duration being determined by the time interval from the arrival to the appearance of new visible arrivals. In our cases this average duration was equal to T a b l e 1 NN D a t a Time of origin 7t m s Koordinates of epicenter M Number station of determining NN D a t a Time of origin 7t m s hkm M M IgE 1 26-X-1952 19 19 18 39.4 143 3 30 6 2 15 13 2 27-X-1952 03 17 15 39.6 143 3 30 6 2 28 12 3 28-III-1954 20 36 24 50.9 175 9 50 6 4 22 11 4 1-IV-1954 18 18 47 46.8 153 5 60 6 0 24 9 5 18-VII-1954 09 07 41 36.0 141 0 40 6 2 15 10 6 9-VI1I-1954 19 16 51 53.3 160 7 60 6 2 26 8 7 30-VIII-1954 07 57 24 43.9 147 4 60 5 9 18 9 8 6-IX-1954 18 30 52 51.7 157 9 60 6 2 30 10 9 23-IX-1954 21 43 37 48.6 156 8 60 6 3 21 13 10 18-VIII-1957 21 42 36 50.0 156 5 40 6 4 15 13 11 3-1-1957 12 48 29 44.0 130 0 560 6 4 17 34 the doubled time of amplitude increment after arrival. I t proved to be equal on the average to about 20 sec for P and 10 sec for S waves. The abovementioned intervals were selected after a detailed consideration of many records. I n calculating the energy of the selected P and S groups, we as- sume t h a t we determine the energy of P and 8 waves within the given frequency range. Since waves of different frequencies have different attenuation with distance, it was necessary to estimate the frequency spectrum of P and S waves as a function of the cbstance. The average periods in the P and S phases were measured. The dependence of the average period on the distance in the se- lected groups is shown in Fig. 2. As may be seen, t h e average periods in the groups investigated are very similar and do not change with a 5 8 V . L . B E L O T E L O V , N . V . K O N D O R S K A Y A , E . TIL. S A V A R E N S K Y variation in the epieentral distance (on the average, TP is 4-5 sec and TS — 6-8 sec.). One may assume that starting with d = 20° and up P and S waves propagate without substantial distortions or shape variations. Our data on energy values are somewhat understated owing to the limited range of the periods under consideration. r / ( d A \2 I — j-j dt we departed from the usual o scheme representing oscillations as sinusoid curves. The calculation of these integrals was carried out by means of a special device. The operating principle of this device consists in the following. The seismogram is placed on the drum of a station recorder which ro- tates at a constant speed. A rheostat having an indicator on its slide is fixed parallel to the drum generatrix. A permanent voltage is applied at the ends of the rheostat, which is proportional to the amplitude on the seismogram at each moment of time. Further, the electronic device differentiates the oscillation curve, squares and then integrates it (5). The integrals were measured repeatedly, the results being continuously D E T E R M I N I N G E N E R G Y OK E L A S T I C W A V E S C A U S E D B Y E A R T H Q U A K E 5 9 compared with the calibration sinusoid curves for which the integral value was determined numerically. The average error of individual measurements turned out to be about 20%. For calculation the amplitudes of true displacement on the earth surface, the normal magnification of the instruments was used as a divider; it is characterized by the magnification of stationary sinusoidal oscillations. The paper by D. P. Kirnos and N. V. Kondorskaya (6) shows t h a t in the case of non-stationary oscillations the instrument amplification may differ considerably from normal. An estimate was made of the cal- fox PWUUVJ foi 3 u/xx irej 10 T s e c * < S • • •50 ' j o ® Pig. 2. - Dependence of the average period in the P a n d S phases on t h e epicentral distance (6). . . . — for P waves xxx - for 8 waves culation error due to the adoption of stationary normal magnification. I t proved to be up to 20%. The reflection coefficients at the free interface and the refraction coefficients at the intermediate boundaries have been taken from papers by B. Gutenberg (') and S. D. Kogan (8). sin 0 ' sin. o The function — 0 was calculated on the basis of Hodgson's cos e de dO tables (9) and the numerical differentiation with respect to the tabulated values of e. Values of this function for different depths from 0 to 600 km were plotted. We defined / (e, a) as the amount of energy per unit solid angle in the direction e, a as related to the average energy of the source in a unit solid angle. To calculate this, it is necessary to known the direction of the forces in the source, and the position of the seismic stations which 6 0 V . L. B E L O T E L O V , N . V . K O N D O R S K A Y A , E . TIL. S A V A R E N S K Y is determined by the angles e and a. For the calculations of / (e, a) we employed the equations for displacements in P and S waves suggested by Keibs-Borok (10). The source mechanism was studied specially for one of the earthquakes of the series under review, namely t h a t of J a - nuary 3, 1957, on the basis of distribution of displacement signs upon the arrival of P.SV, SH and PCP waves at the seismic stations. The force system of this earthquake is modelled by a double force with a T a b l e 2 D a t a hkm IgEp igEs lg Ep+S M igEui) A IgE (IgEp+s- IgE(M)) 1 26-X-52 197; 30 20 9 20.7 21 1 6 2 21 1 0 2 27-X-52 03/i 30 20 8 20.6 21 0 6 2 21 1 — 0.1 3 28-111-54 20 h 50 20 9 20.9 21 2 6 4 21 4 — 0.2 4 l-IV-54 1 8h 60 21 0 20.5 21 1 6 0 20 8 + 0.3 5 18-VTI-54 09h 40 20 8 20.7 21 0 6 2 21 1 — 0.1 6 9-VIII-54 197; 60 20 5 20.9 21 1 6 2 21 1 0.0 7 30-VIII-54 07 h 60 20 7 20.7 21 0 5 9 20 7 + 0.3 8 6-IX-54 18 h 60 20 7 20.9 21 1 6 2 21 1 0.0 9 23-IX-54 21h 60 20 6 20.8 21 0 6 3 21 3 — 0.3 10 18-VIII-57 21 h 40 20 3 20.3 20 6 6 4 21 4 — 0.8 11 3-1-57 127i 560 20.45 20.65 20 9 6 4 21 4 — 0.5 moment, the forces forming a plane extending at 303° and inclined at 10°. The movement in the rupture plane is nearly horizontal. The determination of the direction of forces in the source of this earthquake was helpful in determining / (e, a) which varied between 0 and 1 for the system of the stations used. I t turned out that the scatter of E values for individual stations exceeded / (e, a) variations. For the other 10 earthquakes we assumed / (e, a) = 1. I n calculating the average E value for each earthquake, we deter- mmed the average geometrical of all values calculated for individual seismic stations. E — | Ex • E2 En , where n = number of stations whose data served to determine the energy. D E T E R M I N I N G E N E R G Y OK E L A S T I C W A V E S C A U S E D B Y E A R T H Q U A K E 6 1 In this case E = \ El • E2 • E3... jQn • ek(di+ez+d3 + where E = Ee k° , k = 0.00012a.,,,-1 (Gutenberg's value). at, 3, • -T v n • Oo* • • a <1 • • M. • OO . 1 Fig. 3. - Experimental d a t a on A lg Ep and A lg Es A lg Ep = lg Ep* (individual) - lg Ep (average) A ]g Es = lg Es* (individual) - lg Es (average) as a function of the epic.entral distance (0). . . . - experimental d a t a on earthquakes No. 1-10 ooo - experimental d a t a on e a r t h q u a k e No. 11 Taking into account all the values indicated, we obtained the values of the energy of P and S waves reduced to the source. They are pre- sented in Table 2. The data of the Table enable the following conclusions to be drawn: 1. The average values of lg Ep and lg Es are very similar. 6 2 V . L. B E L O T E L O V , N . V . K O N D O R S K A Y A , E. TIL. S A V A R E N S K Y 2. All the earthquakes under consideration have nearly iden- tical values of M. The energy values obtained are also rather close. 3. If the relationship of the M value and the energy as obtained by Gutenberg and Richter ( u ) is assumed to be lg E = 11.8 + 1.5 M, then the values of Ig E derived agree with the energy values obtained by our method. 4. The values of A lg Fjp and A lg Es were considered, which re- present the deviations from the average values of lg Ep and lg Es for each station within individual regions. One can observe a regularity of A lg Ep and A lg Es values for in- dividual stations. These deviations may be regarded as corrections which should be introduced when making determinations according to the data from individual stations. I t is noteworthy that these cor- rections agree with the features of the hodograph which were previous- ly noted by N. Y. Kondorskaya (12). I t was shown that the travel times of seismic waves to individual stations within the region exhibit a re- gularity with respect to the averaged Jeffreys-Buben hodograph: the travel times to some stations are always less than to others. The deviation signs of A lg Ep and A lg Es for individual stations agree with deviation signs of travel times to these stations from the Jeffreys-Buben hodograph. The most probable cause of these deviations obtained evidently bes in the peculiarities of the geological structure in the region of each seismic station. To analyse ab the obtained data statistically, a single graph was used for plotting the deviations of individual values of lg E * for indivi- dual stations from the average value of lg E for each earthquake. Such E* plotting enables to obtain the dependence lg j - = A \gE = f (0) for earthquakes of different intensities. The respective dependences A IgE = f (6) for P and S waves are given in Fig. 3. In spite of the considerable scatter of individual values, the general trend is toward an increase in A Ig E with the epicentral distance which may be due to the decrease in the absorption coefficient of energy with depth. SUMMARY The energy of 11 earthquakes in the Far East was examined by using observations of seismic stations of the USSR with epicentral distances from 20° up. DETERMINING ENERGY OK ELASTIC WAVES CAUSED BY EARTHQUAKE 63 The absolute value of energy of elastic ivaves (P and S) ivas determined on the basis of development of Galitzin method. The results agree with the value of energy calculated on the basis of Gutenberg and Richter formula. IgE = 11.8 + 1.5 M. The dependence of the value of lg E with epicentral distance was no- ticed. It is suggested that this tvay be due to the decrease with absorption coefficient of energy with depth. RIASSUNTO E stata esaminata I'energia di 11 terremoti nclV Estremo Oriente uti- lizzando le osservazioni delle stazioni sismiche dell'U.R.S.S. con distanze dalVepicentro da 200 in poi. II valore assoluto delVenergia delle onde elastiche (P ed E) e stato determinato sulla base dello sviluppo del metodo di Galitzin. I risultati concordano con il valore delVenergia calcolato sulla base della formula di Gutenberg e Richter lg E = 11,8 + 1,5 M R E F E R E N C E S (1) E. B. rojimiHHj O 3eMjiempjiceHuu 18/11/1911 r., « H3BCTHH POCC. A H , » c e p . 6, 1915 r . (2) JEFFREYS H . , The Pamir earthquake of 1911 February 18 in relation to depth of earthquake foci. « M o n . N o t . R o y . A s t r o n o m . S o c . G e o p l i y s . n , Suppl. 1, 2, (1923). (3) E. . CABAPEHCKHH, H . B. KOHAOPCKAH, B. JI. BEJIOTEJIOBB, 06 onpede.ieuuu sHepeuu ynpyzux eo.au, nopOMcdaeMUX 3eM>iempnceHueM, H3BecTHH AH CCCP, Cepn reo(J)H3HMeciH3., N° 7, 1957.