next up [*] print clean
Next: Introduction Up: Table of Contents

Predictive signal/noise separation of ground-roll-contaminated data

Morgan Brown, Robert G. Clapp, and Kurt Marfurt

morgan@sep.stanford.edu, bob@sep.stanford.edu, kmarfurt@tank.agl.uh.edu

ABSTRACT

Coherent noise contamination is a first order problem plaguing the imaging of seismic data acquired in both land and marine environments. We present a new method for the predictive separation of coherent noise from prestack data which operates in the t-x domain. We apply the new method to real 2-D receiver lines coming from a 3-D shot gather, and nondestructively separate hyberbolic ground roll from primary reflections. This method performs favorably compared to other common techniques, even with an imperfect model of the ground roll, a fact which makes the method attractive in cases where the noise is difficult or expensive to model explicitly.



 
next up [*] print clean
Next: Introduction Up: Table of Contents
Stanford Exploration Project
10/25/1999