A Note on the Characteristics of P Coda Waves From Experimental Results

Seismologists customarily use coda wav�� to study the inhomogeneous characteristics of the earth. Howe'\'·er, P coda waves ha'\'·e seldom been stud­ ied due to their weak signals and short duration time when the epicenter distance i s less than 100 km. In order to manifest the characteristics of the P coda, a measurement of 2-D physical modeling was made by systemati­ cally changing the scattering attenuation and the intrinsic attenuation of the medium. Based on the experimental results, the decay rate of the P coda with elapse time is independent of the scattering attenuation; in contrast, the excitation level of the P coda is proportio11al to the scattering attenuation. Moreo\7er, the amplitude decay with the elapse time of P coda waves can he satisfactorily interpreted by the n1odel (.>f energy transfer (Shang and Gao, 1988). (Ke)1 w·ords: P coda, Physical model, Scattering attenuation)


INTRODUCTION
The decay rate ot' a coda �iave is independent of the distance. and the path between the epicenter and a station. Similarly, it is independent of earthqt1ake magnitude and receiver station representi11g �ln average pr()pert)l ()f the i·egion Ct)ntaining the source an(i receiv·er (Aki and Chouet., 1975). Since it is easy to completely and ltniforml)' 1ne'=1sure the C()da Q over a large area, it is C()n\.renie. nt t() use C()da wav·es to stt1d)1 the st)urce characteristics and p1· < .

EXPERll\ilENTS AND RESULTS
In an ultrasonic 2-D physical modeling experiment here, the char�tcteristics ()f. the P coda wav·es \\lere observed. The detailed experimental se. tup and procedures are presented in Chang ( 1995). A plexiglass plate and a duralumin plate \:vere individually taken as a 2-D homoge ne()US medium. The former was specit'ied \\1ith a strong intrinsic attenuation C()efficient, ""'·as taken as a soft and low velocity area, st1ch as a sedimentar)1 basin. The latter \i\1ith a weak one, w<:1s taken as a harder and high ''el(lcity· ()ne, e.g., an igneous area. The physical properties and dimensions of the plates are sho\�1n in Table. I. A disk transducer PZT-4, inlaid in the plate, was like a point source and ge11e. rated P-wav·es into the model. Chang ( 1995) describes the characte. ristics of' the tra11sducer clearly. The holes \Vere distributed randomly' in the plate and were used t:ls scatterers in the present st11dy.  The percentages ot� the volume. ratio (YR) ()f the holes to the plate for the plexiglass models were ().5, 1. 2 and 3.3 f'()f i11creasing scatterer numbers, but the YR for the duralumin ffi(>dels were 1.24 and 2. The dominant t'requenC)' ot� the S()Urce wav1elet \\i·as 100 KHz and the rc1dit1s of the void scatterers was 0.1 cm for the plexiglass plate. Howev1er, as for the duralumin plates, the domin,1nt f'reqLtencies were 500 K .
Hz, 200 KHz and I 00 KHz with the radii of the scatters being 0.1 ctn (t11d 1 cm. The radiant C()mponent and trans\'erse component are mea sured for analysis. The values of' sc�tttering Q (Qs) of' the plexiglass plate and duralumin plate with \/ariant scatterers C(lU1d be estimated b)l the a\'eraging \\lavef()fffi method (Hsieh and Chang, 1996).     equivalent (see Figure 4). This indicates that the . decay rate is also independent of scattering

DISCUSSION AND CONCLUSIONS
The decay of S coda wav·es with elapsed time depends only on the intrinsic attenuation (Frankel and Wennerberg, 1987;Shang and Gao, 1 . 988;Hoshiba, 1991� Matsunami, 1991, and on the basis of' this study, it may be concluded that the P coda wav·es are the same. The theoretical curve C)f the P coda wave decay with elapse time can be compu.ted by the energy tlux method (Shang and Gao, 1988) when the source and detector are placed at the same point in the 2-D model by· the following equation : (1) where A C ()(1a( t) is the env·elop of coda \Vaves, ,r.J.�' is the scattering attenuation coefficient, gt is the intrinsic atte . nuatio11 coet. t' icient� Vis the ve . locit)' of the medium, and tis the time. By using the P-\\7ave velocity and its intrinsic attenuation coefficient of the plexiglass and duralumin plate, the theoretical decay curves (thick dashed lines) ()f' the ,P coda waves depicted by Equation (1) are shown in Figures 2 and 4, respectively. These figures show that the energy flux model can describe the decay rate ot' the P coda very \\1ell, e\'en though the energy flux model was origi nally use -d for the S coda waves. By comparing the decay rates ()f the P coda waves in Figures  2 and 4, it is noted that the decay rates in the duralumin plate are slower than those in the. plexiglass plate. This rnight be due to the fact that the intrinsic attenuation of the duralumin plate is weaker than that of the plexiglass plate.
Figt1re 3 shows that the excitation lev'el of the P coda \\,aves is proportional to the scatter ing attenuation. This result is the same as that found by Matsunami ( 199 1 ), Aki and Chouet ( 197 5), Shang and Gae) ( 1988) and Hoshiba ( l 991). They sl10\\1ed that the excitation level of the S coda is proporti<)nal to the square of the scattering attenuation when the source and detector are placed at the same point.
The present results can be applied tc) P coda waves generated from an artificial explosion. The P coda excitation level is related to the P-\:vave scattering attenuation, while the decay, rate is dependent on intrinsic attenuation.