STRONG MOTION MODELLING OF RUPTURE PLANE ALONG AN IDENTIFIED PROBABLE CAUSATIVE FAULT

Abstract

Earthquakes are the most disastrous of all natural phenomenon. In this century alone, over five million lives have been lost due to earthquakes and they have been responsible for loss of property of over one hundred and thirty billion US dollars. At the time of an earthquake the elastic energy travels from the focus in all direction. In the epicentral region (i.e., near the source) earthquake recording instruments (seismographs) go off scale or get saturated because of high amplitude of ground motion. In such cases accelerograms or strong motion records provide useful insight into the earthquake processes. Analysis and interpretation of observed and simulated strong ground motion records holds promise for an enhanced understanding of the earthquake process and about the nature, type and extent of the causative fault of an earthquake. The main objectives of the present work are (i) to build a conceptual model of rupture plane along a causative fault identified on a map, (ii) to simulate accelerograms at selected observation points, and (iii) to compare the parameters extracted from simulated and field records for estimating the efficacy of the model. The most probable causative fault for an earthquake is identified on the basis of the following data (i) Isoacceleration contours prepared from field strong motion data, (ii) Isoseismal map, (iii) Tectonic map of the region, (iv) Location of aftershocks, (v) Geological cross sections and (vi) Fault plane solutions for the earthquake. It is not always possible to have all the above mentioned data for an earthquake, in which case, the available data is used to identify the causative fault. The fault thus identified is marked on the tectonic map as the most probable causative fault. The location and strike (0) of this fault is used for modelling. The rupture on the causative fault is assumed to occur on a rectangular plane embedded in a homogeneous isotropic half space. The model is based on following parameters : rupture length (L), downward extension of the rupture plane (D), dip (<J), rupture velocity (Vr), velocity of the medium (V), geometry of rupture propagation, total number of elements within rupture plane and nucleation element. The length, downward extension of rupture plane and rupture velocity are computed from empirical relations. Dip and strike are obtained from fault plane solutions. Velocity of the medium is assigned according to local geological conditions and rock types. The entire rupture plane is divided into equidimensional square elements of length (LJ. This is mapped in a three dimensional coordinate system. The element at which rupture initiates is called the nucleation element. Rupture propagates element by element in a radially outward direction till it covers the entire rupture plane. The nucleation element and the successive elements which get effected as the rupture propagates, emit a source wavelet. The time lag between the activation of two successive elements depends upon the rupture geometry within the rupture plane. From the nucleation element the rupture begins at time T=t0,and energy starts travelling with velocity Vand reaches the observation point at time T=t;which is the initial point (or starting point) on the synthetic record. The rupture spreads along the rupture plane. Total travel time between a particular element (other than nucleation element) and the observation point is the sum of : (a) The time taken by rupture to travel between nucleation element and that particular element with velocity Vr and (b) The travel time between that particular element and the observation point with the velocity of the medium (V). The source wavelet from different elements reach an observation point at different times. The simulated record at the observation point takes into account the appropriate time lags due to geometry of rupture propagation and travel times of wavelet through the medium. Various features have been extracted from both the field and simulated records, in time and frequency domains for quantitative comparison of the strong motion records. The parameters extracted are (i) Peak acceleration (PJ; (ii)Peak velocity (Pv); (iii)Peak displacement (Pd); (iv) Time of arrival of peak acceleration (TJ; (v) Duration of acceleration record (Td); (vi) Ratio of area covered by acceleration record above and below the abscissa (Rat) and (vii) Sum of acceleration values on both side of abscissa (T,rea). The parameters extracted from the autocorrelation function of the acceleration record are : (i) ACFt =Ti; (ii) ACF2 =T2; (iii) ACF3 =T3; (iv) ACF4 =Tm; (v) ACF5 = Aj/Ao; (vi) ACF6 =A2/A0; (vii)ACF7 =A3/A0; (viii)ACFg =Am/A0; (ix) ACF9 =Ratio of area under ACF from time T =0 to T = T, with area under ACF from time T =T, to T =T, and (x) ACF10 =area under ACF above the abscissa and area under ACF below the abscissa. Where A, = Autocorrelation function (ACF) at subscripted lag *i*;X, =Time of ith zero crossing in ACF (i =1,2,3) ; Tra =Time of global minima in ACF and Am = Value of global minima of ACF. Parameters extracted from the power spectrum of acceleration records are : (i) Fp frequency at which maximum power occurs; (ii) F, (i = 1,2,3) frequency at which 25th, 50th and 75th percentile of power occurs; (iii) F, (i = 4,5,6) frequency at which 25th, 50th and 75th percentile of frequency we.ghted power occurs. Three software packages were developed (i) to build a conceptual model of rupture plane along a causative fault identified on a map and to simulate strong motion records at selected observation points, (ii) to extract time domain parameters from simulated and field records and (iii) to extract frequency domain parameters from simulated and field records. The most probable causative fault was identified for three recent earthquakes in India, for which strong motion data was available from networks operating in the vicinity of the earthquake epicenter. Synthetic strong motion records forthese earthquakes was generated at selected observation points and twenty four features were extracted from field and simulated records. The data used and modelling parameters of earthquakes studied are listed in Tables 1 and 2, respectively. Acomparison of synthetic and field strong motion data was carried out for all twenty four extracted variables. Value of parameters Rat extracted from the synthetic record is atleast 80% of its value extracted from field record at all stations for the Dharamsala and the Uttarkashi earthquakes, while for the Meghalaya earthquake at eleven stations out of twelve. Value of parameter ACF10 extracted from the synthetic record is also atleast 80% of its value extracted from field record at all stations for the Dharamsala and the Uttarkashi earthquakes, while for the Meghalaya earthquake at eleven stations out of twelve. Value of parameter F3 extracted from the synthetic record is again atleast 80% of its value extracted from field record at two stations for the Dharamsala and the Uttarkashi earthquakes, while for the Meghalaya earthquake at eight stations out of twelve. The analysis brings out that three variables Ral, ACFU) and F3 are diagnostic parameters, i.e. the variable ofsynthetic record varies by a difference less than 20% that of the field record, at maximum numbers of stations for all three earthquakes. Strong motion records were simulated for a hypothetical earthquake of magnitude 6.5 nucleating within the North Almora Thrust. Developed software packages were used to model the rupture plane for simulating strong motion records at Tehri. Five different positions of nucleation points were selected for simulating strong motion records. Peak acceleration obtained from the five simulated records vary between 273 to 446 cm/sec2. This strongly suggests that from the hypothetical situation the peak acceleration at Tehri will be atleast 273 cm/sec 2 and can go upto as much as 446 cm/sec2, depending upon the position of nucleation point within the rupture plane. Ifthe strong motion records at or nearby Tehri are available, it will be of tremendous help in comparing parameters extracted from simulated records. This will add confidence in assigning design parameters to civil structures. Since limited strong motion data is available in the Himalayan (MBT and MCT) region, therefore data from these networks can be used to give better estimates of modelling and simulation techniques and eventually to give better estimate of design parameters.