REFLECTION SEISMOLOGY: SYNTHETICS AND INVERSION

Several aspects of forward modelling and inversion techniques related to reflection seismology are investigated. An impedance well-log blocking algorithm that is based on a Markovian probability model is presented. The motivation for this algorithm is to provide a simpler input data set for forward modelling procedures in reflection seismology (i.e. creation of synthetic seismograms).; Seismic inversion currently has three major steps. They are stacking or velocity analysis, deconvolution, and migration. A new ray-tracing algorithm for the latter is presented. This algorithm will migrate data for two- or three-dimensional situations and for coincident or non-coincident sources and receivers. The initial data set required contains picked reflection events and the acoustic speed of each layer (the true value of the reflection coefficients could also be used).; A modification to maximum-likelihood deconvolution is suggested that will provide a faster computational time for this procedure. Currently this time is too long to practically use maximum-likelihood deconvolution regularly. The basic assumption of this deconvolution is that the reflectivity function is a Bernoulli-Gaussian process. It is shown that this assumption is similar to the Markovian assumption for the impedance function.