It is well known that CSEM data are sensitive to the presence of hydrocarbons, but they lack the sensitivity to fine scale structural detail that can be attained from seismic data. In contrast, seismic data can provide detailed structural information and resolve some rock properties, notably porosity, but cannot unambiguously map out fluid properties. In this regard, joint seismic and CSEM data interpretation is becoming popular. In this paper, we present a global, Genetic Algorithms (GA), joint inversion algorithm for estimating reservoir parameters from both seismic AVA and CSEM data. Through an example, based on the Luva gas field, we demonstrate the power of the two datasets in the retrieval of reservoir parameters when jointly interpreted.
There were recently reported a number of successful applications of the joint inversion of seismic and CSEM data in reservoir characterizations (e.g. Gallardo and Meju, 2004; Hoversten et al., 2006). However, due to the fact that the relationship between reservoir parameters and the corresponding elastic and electrical properties is found to be non-unique and subject to uncertainties, and that the sensitivities of seismic and electromagnetic data are quite different, to date two major approaches have been applied: (1) methods that involve the use of structural properties, i.e. boundaries of geological targets, as a common factor between seismic and resistivity models (e.g. Gallardo and Meju, 2004); (2) methods that involve the use of petrophysical characteristics to relate the two datasets, for example, water saturation and porosity can be used to provide a link between resistivity and seismic velocity in porous media (e.g. Tillmann and Stocker, 2000). Strategies for joint interpretation of EM and seismic data therefore range from what we term cooperative, in which both datasets are interpreted to determine a mutually consistent geological model, to fully coupled joint inversion. The first class of joint interpretation is generally aimed at finding resistive targets with structural help from seismic. A fully coupled, joint simultaneous inversion can also be applied and used to directly constrain reservoir rock properties. However, the opportunity to carry out such joint inversion depends on the development of a robust rockphysics model, which links the elastic and electrical parameters to reservoir rock and fluid properties of interest. In this paper we present a method for jointly inverting seismic AVA partial stacks and CSEM data, with an example based on the Luva gas field, North Sea.
The Genetic Algorithm (GA) approach is a global optimization that is analogous to the natural processes of biological evolution. The technique uses stochastic processes to produce an initial population of models with associated data misfits. In general, the fittest individuals of any population (those with the lowest data misfit) tend to reproduce and survive to the next generation, thus improving successive generations. However, inferior individuals can, by chance (through mutation), survive and also reproduce. The GA does not require the objective function to possess ‘nice’ properties such as continuity, differentiability and satisfaction of the Lipschitz Condition as local optimizations does.