Active seismic monitoring of underground formation under production, for use in hydrocarbon industry, involves allowing separation of induced microseismic signals from seismic signals emitted during surveillance

摘要

The induced microseismic signals are separated from the recorded surveillance signals by comparison with a reference spectral model and taking into account the spectral contributions from each source (Si) at the fundamental frequencies emitted and at their respective harmonics, and reconstituting by inversion in the time domain the microseismic signals. Seismic recording cycles are produced with emitting seismic waves into the formation, coupling with it one or more seismic sources (5) which, in this case, emit simultaneously orthogonal signals relative to each other to form a composite vibratory signal. The signals returned by the formation are received and recorded by at least one seismic detector (2), and the recorded signals treated to separate the respective contributions of the seismic sources to the signals received and reconstruct the seismograms equivalent to those that would have been obtained by operating the seismic sources separately. The spectral contribution of the microseismic signals to the received signals is obtained by retrenching the amplitudes and phase values associated with the reference spectral model against the recorded amplitudes and phase values. The spectral reference model is a running model formed by using a previous spectral model and taking into account the spectral contribution of previous recording cycles. The running spectrum model is formed by determining a mean or median value of the frequency spectra from previous and/or later recordings obtained from the same source at the same frequency, or by extrapolation or interpolation of the frequency spectrum from adjacent spectral values. Using N seismic sources (Si), for each recording n of a recording cycle p, the respective contributions (Cp,i,n) of the different sources are calculated. The ratio (Ep,n) of the contribution to a running spectral model (Mp,n) is then calculated, with the running model formed by updating a previous spectral model (Mp,n-1) from the frequencies emitted during the previous recording (n-1) and their harmonics. The part (Ap,n) of the recording n for cycle p which can be associated with the active seismic surveillance is then deduced and thus the part (Pp,n) of the recording n of cycle p which is relevant only to passive microseismic activity is deduced. The seismograms for the active seismic surveillance are formed by inverting in the time domain the respective spectral contributions (Cp,i,n) for each seismic source (Si) to the fundamental frequencies and their harmonics, after completing a measurement cycle. The underlying microseismic signals are found by inversion in the time domain of the relevant part (Pp,n) of the signals. The spectral contribution (Cp,i,n) is obtained by multiplying a transfer function (Tp,i,n,r) between a wavelet characteristic of the source and a seismogram associated with a receiver (r) by a wavelet characteristic of the source. The transfer function is continually updated during a current cycle from an estimation (Tp,i,n-1,r) made during the previous cycle and an initial estimate (T0p,i,n,r) made during the current cycle by the relation Tp,i,n,r = (1- h)Tp,i,n-1,r + hT0p,i,n,r