One-dimensional seismic inversion using adaptive deconvolution

Abstract This paper presents an adaptive deconvolution approach to one-dimensional normal incidence inversion of noisy seismic data. A layered system is considered, which is characterized by reflection coefficients and travel-times. The reflector model is assumed to be random, while the seismic source signature wavelet, which travels into the earth, is unknown and time-varying. A recursive instrumental variables algorithm is appropriately combined with a fixed-lag deconvolution and detection technique for adaptive estimation of the wavelet, reflection coefficients and travel-times. The proposed procedure includes a multiples suppression method, which is fed with the above estimates, constructs and removes from the given signal the part containing all significant multiple, transmission and absorption effects. Examples are given, which illustrate the derived results.