Seismic inversion of the Fortuna National 3D survey (Tabaso, Mexico)

A stratigraphic deconvolution has been carried out on the Fortuna National 3D seismic cube (Fig. 1) in collaboration with the Pemex Macuspana exploration business unit. The seismic inversion processing enabled a relationship to be established between the reservoir parameters, the acoustic impedance and the seismic response. In stratigraphic deconvolution processing, the 'classic' wiggle trace is replaced by a spiky response corresponding to a blocked acoustic impedance (AI) output. The spikes are coincident with those AI contrasts that are well aligned with meaningful geological boundaries. This is because the tuning effects of thin beds have effectively been removed in the inversion scheme. The procedure enhances the seismic resolution and discrimination potential in the zone under investigation. Seismic amplitudes should be preserved during the processing (see Hilterman 2001). The basic input for the model-driven inversion processing consists of: • a pre-stack time-migrated seismic cube (PSTM); • well logs of the MAC-201 borehole converted to time; • a 3D macro-layer model with continuous TWT horizons and initial acoustic impedance values. The current study was carried out in parallel with an AVO study. The AVO analysis required rigorous pre-processing of the seismic data. The seismic data cube, obtained in this way, was used in the final inversion procedure. The wavelet is extracted from the seismic cube by cross-correlation procedures. This is basically performed in a 1D manner and the result is therefore strongly influenced by local noise. The bulkshift for the logs is determined, and a zero-phasing operator for the seismic cube is computed. Based on the well-to-seismic match, the phase rotation is determined in order to obtain a zero-phase seismic cube. The 3D macro-layer model consists of TWT horizon surfaces with specification of the initial acoustic impedance values between them. The inversion algorithm creates microlayers in the model and perturbs the AI values. The perturbed AI trace is convolved with the seismic wavelet to compute a new synthetic trace. The difference between this synthetic trace and the original seismic trace is calculated at the same x-y position (Fig. 2). The program terminates the AI perturbation at a certain threshold value for this difference. It stores the AI model and goes on to the next trace. A true 3D approach is applied with several traces as input around the target trace. This approach stabilizes the outcome of the inversion, trace by trace. Testing is carried out to optimize the parameters of the inversion algorithm around the calibration well. The seismic inversion method never gives a unique result and several acoustic impedance models may equally explain the surface seismic response. By putting in constraints (e.g. geological, petrophysical), the number of solutions is reduced and the result is described by a most plausible scenario. The validity of this scenario is calibrated by the additional well control available and, in favourable circumstances, it will increase the interpreter's confidence in the correctness of the inversion procedure. The results of the inversion are summarized in AI section plots and layer maps (microlayers). Meaningful reservoir characteristics are to be extracted with care from the inverted acoustic impedance cube. In this respect, a multidisciplinary approach with close cooperation between the seismic processor, the interpreter, geophysicist, geologist, reservoir engineer and petrophysicist can be considered of prime importance.