Effective Seismic Modeling in 3D Earth Models

Wolfgang Velten. ( 1998 )
Institut National Polytechnique de Lorraine

Abstract

Although mankind prepares to get grip in outer space, the knowledge in the other direction, the inner of the earth, still remains imperfect. Geophysical methods, like gravimetrics, magnetics and seismic procedures allow nevertheless to have a look at the inner of the earth. For doing so, recording travel times and signal strength of waves of earthquakes or of artificially created waves, is an important procedure. The recorded data however don't result in readily images of the inner of the earth, but have to be processed and interpreted before. The task to get images of the subsurface from raw recorded data is called migration, or more general imaging. But one also undertakes the inverse way, which is to simulate seismic events for already given models of the subsurface. This procedure is called forward modeling. On one side it is possible to compare the simulated data with real data, and gain so information about the validity of the used model. On the other side forward modeling is also needed for certain types of migration. There are different approaches to simulate wave propagation in the earth. One of them is the tracing of rays which propagate in a medium. Like in geometrical optics these rays can be reflected and refracted. Unlike in geometrical optics however, the wave velocity can vary within the medium, not only across medium boundaries. The rays therefore are in general not straight, but curved. In seismic recordings signal strength plays, besides travel times, an important role. The theory of ray tracing is nowadays good developed, both what concerns travel time calculation and signal strength. Often however, rays are calculated only in 2D models or 3D models, but which don't allow velocity discontinuities. But just in complex structures, as in salt domes or faulted areas, one can expect to find oil. The given model builder program GOCAD allows for modeling complex geological features, including the modeling of layers, faults and even overhangs. While GOCAD allows easily to model geological objects, a ray tracing, basing on GOCAD and using GOCAD objects, was not developed until now. First it was planned to develop basic routines, which would allow other people to continue work on this topic. Later, this work was supported by an industrial sponsor (Mobil Oil), and also the development of applications was envisaged. The ray tracing development should cover: 1. the possibility to calculate travel times in models of complexity comparable to those of models in exploration and production; 2. the possibility to calculate signal strengths; 3. the possibility to calculate wave fronts; 4. the possibility to calculate travel times on regular grids in the subsurface; 5. the possibility to represent velocity on tetrahedra based models, where velocity should be smooth inside model regions, but should allow discontinuities across region boundaries; 6. these listed features should be demonstrated by producing synthetic seismograms, in common shot configurations, using sufficiently complex models; 7. efficient computation is regarded as essential.

Download / Links

BibTeX Reference

@PHDTHESIS{Velten1998a,
    author = { Velten, Wolfgang },
     title = { Effective Seismic Modeling in 3D Earth Models },
      year = { 1998 },
    school = { Institut National Polytechnique de Lorraine },
  abstract = { Although mankind prepares to get grip in outer space, the knowledge in the other direction, the inner of the earth, still remains imperfect. Geophysical methods, like gravimetrics, magnetics and seismic procedures allow nevertheless to have a look at the inner of the earth. For doing so, recording travel times and signal strength of waves of earthquakes or of artificially created waves, is an important procedure. The recorded data however don't result in readily images of the inner of the earth, but have to be processed and interpreted before. The task to get images of the subsurface from raw recorded data is called migration, or more general imaging. But one also undertakes the inverse way, which is to simulate seismic events for already given models of the subsurface. This procedure is called forward modeling. On one side it is possible to compare the simulated data with real data, and gain so information about the validity of the used model. On the other side forward modeling is also needed for certain types of migration. There are different approaches to simulate wave propagation in the earth. One of them is the tracing of rays which propagate in a medium. Like in geometrical optics these rays can be reflected and refracted. Unlike in geometrical optics however, the wave velocity can vary within the medium, not only across medium boundaries. The rays therefore are in general not straight, but curved. In seismic recordings signal strength plays, besides travel times, an important role. The theory of ray tracing is nowadays good developed, both what concerns travel time calculation and signal strength. Often however, rays are calculated only in 2D models or 3D models, but which don't allow velocity discontinuities. But just in complex structures, as in salt domes or faulted areas, one can expect to find oil. The given model builder program GOCAD allows for modeling complex geological features, including the modeling of layers, faults and even overhangs. While GOCAD allows easily to model geological objects, a ray tracing, basing on GOCAD and using GOCAD objects, was not developed until now. First it was planned to develop basic routines, which would allow other people to continue work on this topic. Later, this work was supported by an industrial sponsor (Mobil Oil), and also the development of applications was envisaged. The ray tracing development should cover:
1.	the possibility to calculate travel times in models of complexity comparable to those of models in exploration and production;
2.	the possibility to calculate signal strengths;
3.	the possibility to calculate wave fronts;
4.	the possibility to calculate travel times on regular grids in the subsurface;
5.	the possibility to represent velocity on tetrahedra based models, where velocity should be smooth inside model regions, but should allow discontinuities across region boundaries;
6.	these listed features should be demonstrated by producing synthetic seismograms, in common shot configurations, using sufficiently complex models;
7.	efficient computation is regarded as essential. }
}