Marmousi, model and data

Abstract

In 1988 a complex 2D model was created and synthetic seismie data were generated from this model. Model and data were designed specifically by the IFP for the blind test of the Copenhagen workshop. This artiele describes the philosophy bebind the model and the creation of model and data. The Model Philosophy behind the modelBefore the creation of the model several demands were formulated to which the model (and the data) should adhere. First, the model should be geologically plausible, meaning that it preferably should be based on a real situation. Second, the model should be complex, meaning that it must contain many reflectors, steep dips and strong velocity gradients in both vertical and lateral directions. Thus, the data should be so complex that the assumptions on which conventional processing relies should not hold: obtaining a good image from the data should require the use of depth imaging methods like prestack depth migration or inversion.Obviously the creation of such a model requires the cooperation of geologists, geophysicists and computer scientists. Model generationThe Marmousi model is based on a profile through the North Quenguela trough in the Cuanza basin (Angola) (fig. 1) as described in Verrier and Branco (1972).Based on this section a geometrie model containing 160 layers was created using the MIMIC™ module of the SIERRA package (colour figure 1). Then the velocity and density distributions were defined by introducing realistic horizontal and vertical velocity gradients. The velo city distribution in the shallower part of the series (the detrital series) has a high vertical gradient to model the strong compaction. This model was transformed into a 2-D velocity/density grid (colour figures 2 and 3) with dimensions 9200 meters (horizontal distanee) by 3000 meters (depth) with a grid size of 4 meters (both horizontal and vertical). Geological model descriptionThe geological model of the basin consists of:• A deltaic sediment interval, thickening from west to east, deposited upon a saliferous evaporitic series. In the eastern part, which seems to he more affected by the continuous lateral salt creep, normal growth faults are developed.• presaliferous folded carbonate platform deposits, in which a structural hydrocarbon trap is expected. :1;1 1 Q)s:s:.~s:~a .-Cl ) Cf)s:Geological HistoryThe geological history consists of two quite distinct phases:The fust phase corresponds to a continuous platform sedimentation (of marls and carbonates). At the end of this sedimentation these deposits were slightly folded and then eroded (with the erosion surface being flat).The second phase began with the deposition of an isopachous saliferous evaporitic series. On this series, a clayey-marly series rich in organic matter was deposited.These sediments were followed by a thick deposit of shaly-sandy detrital sediments whose facies thickness was govemed by continuous lateral creep of salt, resulting from the overburden pressure. Linked to this salt creep, which may locally cause complete disappearance of t...

Toward an automatic full-wave inversion: Synthetic study cases

“…Our first case study is based on the Marmousi model (Bourgeois et al, 1991). Although this is a well-known model, which can be inverted with FWI easily given a good initial model, we propose a more challenging experiment, where we remove low-frequencies from the data and and then high-pass filter both source wavelet and data at 4 Hz, in order to remove as much as possible information from the lowest frequency.…”

A local adaptive method for the numerical approximation in seismic wave modelling

“…For example, in a marine seismic acquisition generally the minimum velocity is the water velocity situated on the top of the studied region, and the geological structures situated below the sea floor have a higher acoustic velocity. In particular, Figure 4, b) shows the 2D acoustic Marmousi velocity model (see [23]), that is discussed more in details in Section 5. By these considerations the choice of a classical finite difference approach to model the seismic wave propagation is inefficient.…”

Scattering angle-based filtering via extension in velocity

“…We show the application of the frequency scaling filter on the example of inversion for the Marmousi model (Bourgeois et al, 1991). For modeling we use a pseudo-spectral method (Kosloff and Baysal, 1982;Tessmer, 2011;Virieux et al, 2011) with a 4-th order Taylor expansion in time for time stepping.…”