Fast imaging and inversion of on and off shore electromagnetic data

Abstract

We consider thin layer algorithms for inversion of electromagnetic data acquired on large experimental grids. These algorithms do not involve targeted search of the model parameters. Instead, they determine an integrated characteristic of the model, i.e. heterogeneous conductance or transverse resistance, using a filtration technique developed for thin layer models, which adequately represent the geo-electric situation by a sequence of laterally homogeneous and heterogeneous layers. The corresponding filter reflects impedance/admittance relations between different components of the electromagnetic field in the stratified part of the model. Convolution of such filters with the data obtained from experimental measurements represents the most time consuming part of the computation. The convolution is carried out using a high performance algorithm, which makes the process of inversion extremely fast. The choice of a particular inversion algorithm is dictated by the specific geo-electric situation in the area of interest and the electromagnetic data subject to interpretation. Typically, most computations can be carried out in up to several minutes using a regular desktop or laptop PC. We restrict our consideration to two simplest algorithms for determination of the conductance of heterogeneous conductive layers ( S-algorithm) and the transverse resistance of heterogeneous resistive layer ( T-algorithm). The algorithms can be useful for interpretation of data acquired in on- and off-shore conditions with natural and controlled sources. Inversion can be jointly carried out for data sets collected for different source locations, at different frequencies, etc. Likewise, the algorithms can be used for interpreting the data collected using a moving source. Performance of the algorithms is demonstrated on a set of test inversions of numerically simulated data sets. The first group of the examples is typical for environmental applications. It is studied using the S-algorithm applied to natural and controlled source measurements for shallow and deep targets. The T-algorithm is applied for delineation of a gas hydrate type of reservoir using the data numerically simulated for such a problem. In geo-electrical situations that can be addressed using thin layer models, results of inversion hardly require any further elaboration. For instance, this happens when parameters of the heterogeneous layer satisfy thin layer conditions and its thickness can be determined from seismic measurements. When this is not the case, the model found by a thin layer inversion usually reveals correct location of the anomalous body and provides reasonable estimate of its integrated electric properties. In such situations the algorithm works as an imaging technique, which together with seismic information on the layer boundaries provides a justified initial model for a full scale 3D inversion and hopefully allows for avoiding pitfalls associated with a straightforward application of more traditional 3D inversion algorithms.