Estimation of Vertical Continuous Stochastic Parameters from Seismic Reflection Data

Abstract

This paper presents a scheme to invert reflected seismic wave fields for vertical spatial statistics of the Earth’s continuous stochastic velocity distribution. The forward model is based on a single scattering theory, where a seismogram is modeled as a convolution of a source wavelet and stochastic reflectivity, which obeys a von Karman autocorrelation function. Tests with synthetic data indicate that the method presented here recovers vertical characteristic length from seismic reflection data that is similar to that of the original velocity model; however, the inversion does not reliably recover the Hurst exponent, ν. The input wavelet’s power spectrum is a free parameter in the inversion and is estimated as well, but is band-limited relative to the original source wavelet. Noise degrades the quality of the estimates of the stochastic parameters such that the recovered stochastic parameters are biased according to the power spectrum of the noise.