Seismic Attenuation Assessment in Passive Seismic Investigations Using Pulse Width Analysis and Considering Source Effects

Transcription

1 Seisic Attenuation Assessent in Passive Seisic Investigations Using Pulse Width Analysis and Consideing Souce Effects G.A. selentis* (Univesity of Patas), P. Paaskevopoulos (Univesity of Patas), N. Matakis (Landech Entepise SA), A. Sepetsidaki (Univesity of Patas) & E. Sokos (Univesity of Patas)

2 Suay Knowledge of attenuation can be vey useful in any stages of conventional seisic data pocessing as its eoval inceases esolution. Recent advances in passive seisic investigations involving icoeathquakes indicate that attenuation assessent has a geat potential as a diect hydocabon indicato and esevoi stuctual delineation. he pupose of this investigation is to assess the use of pulse width analysis to invet fo attenuation stuctue. Intoduction Geological stuctues of hydocabon inteest ae coonly stongly heteogeneous. his heteogeneity affects the seisic wave popagation and ay poduce phenoena like focusing and defocusing of seisic ays, wave scatteing, aplitude absoption etc. Futheoe, a esevoi full of hydocabons tends to be acoustically softe than if it is full of an incopessible fluid such as wate. All these phenoena ae paaeteized by the - paaete (o quality facto) which accounts oveall fo the enegy decay as a function of distance. heoetical odels of seisic wave attenuation pedict that a wave passing a hydocabon esevoi should suffe oe attenuation than in suounding geological ateials. Attepts to extact attenuation fo conventional seisic data have et with little success since aplitude specta of the seisic ecod contain the ipint of the aplitude spectu of the eath s eflectivity as well as the aplitude spectu of the seisic wavelet. he ai of the pesent investigation is to extend the application of passive seisic ethodologies towads evaluating accuate values of the attenuation stuctue coecting fo the seisic souce. heoy he isetie τ and pulse width of a seisic signal geneated by a finite souce duation can be epesented by the following nonlinea elations (DeLoenzo et al. 2004) τ = τ 0+ γ ( 0, p, θ ) + λ( 0, p, θ ) Τ= Τ0 + C Whee τ Τ 0 = = sin θ ; 0 p p sinθ θ is the take off angle and o the fault adious (assuing cicula cack), the uptue velocity, p the p-wave velocity, C has been expeientally deteined to be 0.5 (Gladwin and Stace, 1974) and the paaetes γ, λ ae those given by DeLoenzo et al. 2004).

3 Figue 1: Rise tie Τ and pulse width τ of a seisic phase on a velocity seisoga a velocity seisoga, τ epesents the tie duation of the fist half cycle of the p-wave phase, while is the tie duation of the coplete cycle of the wave (Figue 1). Given N icoeathquakes ecoded at a netwok of M eceives, the data space consists of the easueents of ise ties τij and pulse widths ij, with i=1,n and j=1,mi, whee Mi the nube of eceives that ecoded the ith event. he take off angles θij can be expessed in tes of the dip δi and stike φi diections of the plane containing the cicula fault cack of the i diections of the plane containing the cicula fault cack of the ith event (Aki and Richads 2002). Let us assue that by applying conventional passive seisic toogaphic invesion on ou dataset (selentis et al. 2007), we have a detailed 3D p-wave velocity odel of the egion unde investigation. By dividing the egion into blocks the whole path attenuation te / can be expessed as a su of the contibutions i/i of each block cossed by the p-wave ays P = i i i he nonlinea syste of equations to be solved is theefoe given by P 0, i 0, i i, j, τ ij = sinω( δ i, φi ) + γ i, j + λi, j p = 1 Τ ij = 0, i + 0, i p P i, j, sinω( δ i, φi ) λ = 1 i, j With i=1,n and j=1,m and i,j, epesenting the tavel tie of the fist p-wave geneated by the ith event ecoded at the jth station passing though the th block and Ω given as a function of stike and dip of the fault plane (Aki and Richad 2002) Since the above equations show a nonlinea dependence in souce and attenuation paaetes we define the following cost function cf cf M 2 { ( ) ( 2 W i τ τ + W Τ Τ )} τ, j i, jobs i, jest Τi, j i, jobs i, jest = N Whee i= 1 j= 1 W i, j Τ and W i, j τ ae the weight factos. he next stage is to solve this equation by using a global optiization technique fo seaching fo its iniu, eploying the downhill Siplex optiization ethod (Ichinose et al. 1997). We assue soe a pioi constaints of the fo each discetised subsuface block deived

4 fo the iniu and axiu slope of the elation of ise tie and pulse width vesus tavel tie data. he data In this investigation we used the souce and eceive configuation of the Rio-Antiio 2004 passive icoeathquake expeient (selentis et al. 2007). he data set consists of 150 welllocated icoeathquakes with ML anging fo 0.5 to 3.0 ecoded on a netwok of 70 stations (Figue 2). (a) Figue 2: (a) Seisic station layout (tiangles) and soe velocigas of the the fist beaks ecoded fo a local icoeathquake (shown in 6 stations) he tie segent shown is the sae fo all velocigas (b) 3D distibution of icoeathquake hypocentes ecoded by the local netwok (b) Each station consists of a 24bit EathData PR24 digitize suppleented with Landech s L- S01 thee coponent velocity sensos. he icoeathquake hypocentes (figue 2b) and the 3-D P-wave velocity odel wee obtained fo a joint toogaphic invesion and ae descibed in selentis et al (Figue 3).

5 Figue 3: he 3D p (P wave) velocity odel obtained fo tavel tie toogaphy. he basic paaetes used in the pesent study ae the ise tie τ and the pulse width Τ of fist p-wave aivals. Figue 4 pesents the deived ise ties and pulse width fo the ecoded icoeathquakes vesus the P wave fist beak tavel tie. A geneal positive tend can be seen, even though a consideable scatte of the data is obseved. he incease of τ and Τ with tavel tie is usually attibuted to ay path attenuation effects. he dispesion of the data aound the line of best fit ay be due eithe to diectivity souce effects and/o spatial vaiation in the attenuation popeties of the sapled by the p-wave ays volue. (a) Figue 4: Obseved (a) ise ties and (b) pulse widths vesus tavel tie. (b) Chaacteistic esults of the invesion pocess descibed peviously, as it was applied in the study aea ae depicted in Figue 5. In this figue only layes with the highe values ae pesented. In this way the non coloed egions with low values can also be delineated. he low egions appea to coincide quite well with the eathquake epicentes that wee used in the seisic toogaphy. Also they ae in ageeent with well known sedientay and tectonic zones of the egion fo which we expect highe seisic attenuation.

6 Figue 5: 3D stuctue of the investigated aea. Lowe egions have been eoved. Conclusions Since the elationship between velocity and attenuation will depend on the coposition and cack poosity with depth, in situ easueents of both velocity and attenuation povide ipotant constaints on the lithology and physical popeties of hydocabon esevois and the subsuface in geneal. Futheoe, the assessent of velocity and attenuation odels has ipotant applications to the pocessing and iaging of seisic eflection data. his includes invese -filteing, -phase copensation, aplitude statics and the incopoation of attenuation in igation algoiths. Acknowledgeents he authos want to thank Landech Entepises SA fo peitting to publish the esults of the pesent investigation. Refeences Aki, K. and Richads, P. G. [2002] uantitative seisology: Univesity Science Books, Sausalito. De Loenzo, S., Di Gazia, G., Giapiccolo, E., Gesta, S., Lange, H., usa, G. and Usino, A. [2004] Souce and p paaetes fo pulse width invesion of icoeathquake data in Southeasten Sicily, Italy. Jounal of Geophysical Reseach, 109, B Gladwin, M.., and Stacey, F.D. [1974] Anelastic Degadation of Acoustic Pulses in Rock. Physics of the Eath and Planetay Inteios, 8, Ichinose, G. A., Sith, K. D., and Andeson, J. G. [1997] Souce paaetes of the 15 Novebe 1995 Bode own, Nevada, eathquake sequence. Bulletin of the Seisological Society of Aeica, 87, selentis, G-A., Sepetsidaki, A., Matakis, N., Sokos, E., Paaskevopoulos, P. and Kapotas, S. [2007] Local high esolution passive seisic toogaphy and Kohonen neual netwoks, application at the Rio-Antiio Stait, Cental Geece. Geophysics, 72, B 93- B 106.