Travel time analysis of P waves arising from six underground nuclear explosion at Novaya Zemlya

SUMMARY. - The analysis of the travel times of the P waves generated by six underground nuclear explosions in Novaya Zemlya shows notable early arrivals (oo4sec) at the Fennoscandia stations. These observations, together with a marked linear dependence of the travel times on the epicentral distance throughout the interval 8.5 < / J °< 20.4. were evaluated with the aim of producing a satisfactory model for the upper mantle of the Baltic zone. In particular, there appears a considerable mantle zone with almost uniform velocity, or at least, with not of increasing velocity. This mantle zone is delimited above by a discontinuity at the 86 km depth (v = 8.45 km/sec), and delimited below by the «20° discontinuity» (h = 507 km; v = 10.56 km/sec). The proposed model is amply verified on the basis of the foreseen travel times compared with the corresponding ones, observed from a fair number of impulses associated with teleseismic Pn, noticeable as far as 40°, and from phases reflected and refracted on the discontinuity at 507 km.

The proposed model is amply verified on the basis of the foreseen travel times compared with the corresponding ones, observed from a fair number of impulses associated with teleseismic Pn, noticeable as far as 40°, and from phases reflected and refracted on the discontinuity at 507 km.
(*) Istituto Geofisico e Geodetico, Università di Messina -(Italy). INTRODUCTION. It is generally agreed t'hat the fine structure of the upper mantle cannot be assimilated to a single type for the whole of the earth's sphere. This consideration, already expressed by various authors, has been recently emphasized by Bottari and Federico (1975) ('), who, availing themselves of the support of a vast amount of seismological observations, have studied the upper mantle in the Mediterranean area, with particular regard to the « 20° discontinuity ». Similarly, following much research the upper mantle in the Fennoscandia zone presents peculiar o structures of no less seismological interest (Tryggvason, 1961;BATH, 1966BATH, , 1971Sellevoll and Pomeroy, 1968;et al. ( H , 2 , 4 , 13 ). In fact, in addition to what has already discovered, we consider remarkable both the sharp discontinuity of the seismic parameter corresponding to the epicentral distance 20.8°, and the notable early arrivals of the first arrivals at the Fennoscandia stations in concomitance with some underground explosions in Novaya Zemlya.
OBSERVATION DATA AND ANALYSIS. This work was commenced as a result of the mentioned observations, or more precisely, as a result of the following two: the explosions in Novaya Zemlya give rise in Messina to seismograms with clear phases between P and PP which are not identifiable with known phases; the first arrrivals at the Fennoscandia stations show notable early arrivals with respect to standard travel-times.
With the aim of analysing and interpreting these observations, the statistical epicenters of 6 nuclear explosions were determined (Table 1, col. 3), assuming as provisional coordinates -for calculation purpose-those published by the I. S. C. of Edinburgh. Also utilized were the initial times of the P of around hundred stations more favorably located than Novaya Zemlya. The times elaborated are, however, relative to stations with epicentral distance greater than 20°, with the purpose of obtaining focal locations as for as possible independent of Ihe propagation anomalies which exist in the upper mantle  '' and in particular in the crustal and sub-crustal structures of the Fennoscandia zone. The statistical procedure was carried out using the standard travel-time of both Jeffreys-Bullen (1967) ( 8 ) and Herrin et al. (1968) ( 7 ). The final explosion locations are not appreciably different in either case, but it was thought expedient to take for the epicentral coordinates those arrived at using the Herrin's travel-times, given that, with respect to these, the standard deviations of the observational residuals are smaller (     The travel-times compatible with [2] result for A°> 7 longer than the corresponding ones deduced using [1|. Furthermore, it not being possible to have data on the epicentral range 2-8°, the 'hypothesis that the first impulses in the range 8.5 to 20.8° are referible to seismic rays which have followed a deeper refractory horizon than the base of the crust, was put forward. In order to individuate this horizon, the condition of zeroing the known term was imposed for [11, adopting t'he above mentioned values of thickness and velocity for the crust and below it. The above conditions is satisfied at the level r s = 6285 km, to which belongs the true velocity of 8.45 km/sec. The observation that the first impulses recorded are distributed along a single straight line in an apreciably wide interval of 8.5 -20.8°, leads to the belief that the medium underlying the level r s = 6285 km is very homogeneous. So much so that the velocity v (r) is uniform, or at least doesn' t increase with an increase in depth. Such a model agrees  The hypothesis that, below the level r s = 6285 km, the P-wave velocity can be taken to be uniform was examined closely by approximating the travel-times observed reduced to the level r s , by means of an equation of the type T s = 2 r s /v s'mA s /2, which as is known, represents the travel-time curve in a homogeneous spheric medium. In Table 3   In the interval 20.8<4°<95, the travel-times observed concord excellently with the standard ones, and they are represented by the polynomial T = (68.73±0.96) + (l 1.043±0.061) z\ -(0.03363±0.00118) zi 2 + -(0.0000122±0.0000070) A 3 [6] This results confirms the acceptable quality of the statistical determination of the explosion location, and indicates, relatively to the velocity law of the P, a gradual return to normality below the 20.8° discontinuity.
The sharp incline different of the travel-time curves [1] and [(>] which intersect in correspondence to the epicentral distance 20.8° poses the depth of the « 20° discontinuity » as our first objective. To this end, only the data from Baltic Shield stations (15</l°<25) were utilized. If we consider that, with respect to the source of the P waves the mean azimuth of the Fennoscandia stations is consistent with that for Messina, it was deemed expedient to consider the observational data of only the azimuthal sector having its vertix in Novaya Zeml/a and its opening within the azimuthal range 220<a°<260. From the examination of the observations sent us, we find successive impulses to the first arrivals, as in Uppsala (18.17°, 260. On the other hand, the very close alignment of the first impulses in the distance interval 8 -20.4", and the sharp discontinuity of the seismic ray parameter in the proximity of 20.8°, suggest that the mantle zone with uniforme velocity underlying the discontinuity at 86 km has as its bottom limit the « 20° discontinuity ». Furthermore, the « 20° discontinuity » depth was determined, relative to the Baltic Shield, by limiting the reduction procedure to only the Fennoscandia stations (15<zl°<25), and elaborating the observation data according to a procedure already introduced for the study of the « 20° discontinuity :» in Europe and the Mediterranean Basin (Girlanda and Federico, 1966; Bottari and Federico. 1975) ( 6 , 3 ). In this case the procedure was suitably modified to adapt it to the peculiar features of the area under study. This calculation method provides for the utilization of the travel times associated with the seismic rays having points of maximum depth below the discontinuity, and a knowledge of the velocity laws in ihe crust and in the mantle overlying the discontinuity itself.
Taking in to account [7] and following the above indicated procedure, we obtain the cancellation of the known polinomial term at the depth /? = 506.57 km (/'= 5864.43 km) which level was arrived at by extrapolation of the travel-time curves of equations 7 = 3.55 + 9. The elements which allow the comparison between the reflected phase calculated and that identified in the seismograms are reported in       From a fair number of recordings kindly furnished us by various observatories, we can find, up to distances of around 40°, clear impulses whose travel times have t'he same trend of the Pn, initially caracterized by the relation [1], (Table 4, col. 4 and 5; table 5, col. 9 and 10).
In addition, the notable impulses observed at Messina (Figg. 2, 3, 4, 5 and 6) 13 sec after the first arrivals, already mentioned in the preceding paragraph of this paper, are adequately confirmed in the numerous seismograms of other stations, and are interpretable as P refracted on the «20° discontinuity» (P-¿ 0 )-The observed data relative to this phase, and those calculated in line with the crust-mantle model previously defined, are reported in Table 5 (Col. 3 and 5).
In this connection we emphasize that for this refracted phase the relating travel-time equation has been obtained excluding the observational data of Messina. In fact, at this station, the travel times exhibit systematically considerable early arrivals as well as the observed first arrivals (P waves). This concomitance is revealing the disagreement of the model adopted with the travel times observed at Messina.
The impulses of the P¿ u » recorded concord with the equation of the travel times 7=77.69 + 9.8804 A [12] which was obtained by assigning to the « 20° discontinuity » the calculated depth (506.57 km) and to the seismic ray parameter the value corresponding limit incidence on the same discontinuity (p = 9.8803962 sec/'zl 0      gri-.    CONCLUSIONS.
The analysis of the travel times of the P waves originating concomitantly with six underground nuclear explosions in Novaya Zemlya puts into evidence various observational elements and puts forward solutions which can be summarized as follows: (i) the first arrivals in the distance range 9<zl°<20.4, are noticeably early arrivals with respect to the standard travel-times of Herrin and Jeffreys-Bullen.
(ii) the travel times of the initial impulses result, for the entire distance interval corresponding to the Fennoscandia area, as being strictly linearly dependent on the distance and show an appreciably greater value for the apparent velocity than those reported in literature for the sub-Moho in the same region. We can therefore point out the existence of a refractory 'horizon located at a depth of 86 km, below which the velocity of the longitudinal waves can be considered uniform down to the level of the « 20° discontinuity ».
(iii) the proposed model is confirmed by the observation of Pn waves noticeable as far as 40° and of P vaves reflected (P d P) and refracted (P>o") on the « 20° discontinuity ».

ACKNOWLEDGEMENTS. o
We thank Professor Markus Bath Director of the Seismological Institute, Uppsala, Sweden, for his keen interest and also for providing seismograms of t'he Swedish Network. The writers are indebted to all observatories which have provide to send records and observational data.