Parkfield Earthquake of June 28, 1966: Magnitude and Source Mechanism By

The Parkfield earthquake of June 28, 1966 (04:26:12.4 GMT)is studied using short-period and long-period teleseismic records. It is found that (1) Mb = 5.8 and M~ = 6.4 as compared to Mb = 5.4 and Ms = 5.4 for the foreshock (04:08:54) , (2) both the Rayleigh and Love wave radiation patterns conform to those of a double couple at a depth of about 8.6 km, (3) the main shock can be represented by a series of shocks separated in space and time. The near-field strong-motion data support the last conclusion. Based on strong-motion seismograms, and the surficial evidences of the dimensions of the fault, the energy is found to be 102I ergs. INTRODUCTION The Parkfield earthquake of 28 June 1966 has aroused extensive interest because of its location on the San Andreas fault where every breath has been monitored. The recordings of this event fall into three main categories: strong-motion seismometer and seismoscope records in the epicentral region; short-period seismometer records from Berkeley, Caltech and USCGS network (within an epicentral distance of 200 kin); and long period WWNSS and Canadian net records. In the first category we can procure information about the amplitude period and orbital motion of waves in the source region. This is the first time in seismological history such information was gathered systematically with an array within the source region. However, the mission of this instrument cluster being non-seismological in its design, no absolute or relative time was available on the records; it is impossible to decipher the nature of the waves recorded. In the second category, the seismograms are used to determine accurately the origin time and the epicenter, and also the first motions. The last category is the teleseismie records on long-period instruments (to = 15 sec, Ig = 30 see) ; the body waves that were so predominant on the second category records now appear only as a faint trace in most cases. These last category records are the chief source of data for this study. It must be noted that while the near-source waves must bear some relation to the far-field or teleseismic waves, the manner in which the orbital motions of the nearsource wave behave and the magnitude of the strain involved in these waves throw some doubt on a direct relationship; namely, the near-source waves may be nonlinear in the sense that they diminish much faster than linearly diminished body waves, and/or the material behaved non-linearly under the high stress. In either ease the waves would have suffered a very severe attenuation in the immediate region close to the som'ce; to recover the information contained in these waves from teleseismic waves well described by first-order linear elasticity theory, would be difficult if not impossible. In using the teleseismic signals for source mechanism studies we are looking for 689 690 BULLETIN OF THE SEISMOLOGICAL SOCIETY OF AMERICA average features of the earthquake processes such as the average velocity propagation of the source, the average length of the fault and so on. M A G N I T U D E S OF THE FORESHOCK AND THE M A I N SHOCK Richter magnitude scale as originally designed by Richter (1935) employed the maximum trace of short-period waves recorded on Wood-Anderson torsional seismometers to study the statistical characteristics of local earthquakes. Later, Gutenberg and Richter (1936) extended the same idea to categorize teleseisms based on the body waves P, S, PP and the more prominent phase on a teleseismic record, the "20 second" surface waves. A relation between these two magnitude scales was subsequently sought to equalize them to yield the same result (Gutenberg and Richter, 1956). A basic assumption involved here is that the source spectrum for an earthquake of certain size is fixed in shape; thereby, knowing the amplitude at one frequency would suffice to calibrate the earthquake. In the process of developing the relation, a large number of stations and earthquakes were used; the relation yields rather consistent results for "average" earthquakes. The main shock of the Parkfield event is, however, a rather outstanding exception. The fundamentM formula used in calculation of magnitude M based on surfacewave amplitude is M~ = loglo A~ lOglo B -1~ qMn + ~7(7.1 M,). This formula, a modification (Bath, 1952) of Gutenberg and Richter's original formula, is for exclusive application with vertical component Rayleigh waves, where A = the amplitude (in microns) of 20 second Rayleigh waves recorded on the vertical component of longperiod seismometers, ~ogl0 B -distance correction factor, + MR = correction for depth, path, radiation pattern, etc., = 0.2 for normal shock, Me = the sum of the first four terms in the formula. In the present investigation ~ -tMR is set to 0, since we used a number of stations, and the radiation pattern is aetuMly worked out. For body waves we use the formula Mb = l o g w / T + Q