Iterative inversion of seismic data using singular value decomposition

Abstract

Most geophysical problems are a kind of inversion problem in that we try to learn more. about the physical parameters of the earth from measured data. In this paper we constrain ourselves to the inversion of seismic data. The earth model we try to find is composed of reflector positions and reflection coefficients. We show that no stationarity and minimum phase assumptions are required for the wavelet in the inversion. We also show how reflector positions and amplitudes can be determined with high resolution, even if we start the inversion with the wrong number of reflectors at the wrong positions.