Introduction

In previous SEP reports I have proposed ways to remove effects of unequal radiation patterns in multi-component data. This subject is especially interesting in regard to diminishing biases introduced in inversion results and AVO, but even effects such as uneven stacks can be caused by unequal radiation patterns. I am working on equalizing radiation patterns within a given survey, in a component-consistent and source-location-consistent manner Karrenbach and Muir (1989); Karrenbach (1991). Such an equalization can be carried out in various ways. In this paper I am merely considering equalization to quasi-isotropic patterns by using no subsurface information whatsoever. This equalization technique has the advantage of being purely a preprocessing step. The method I am using dictates a reciprocal experimental geometry. The filtering technique is closely related to prediction error filtering, but should not be confused with surface consistent deconvolution Levin (1987). I am not trying to shape the wavelet to a desired form, although this step could be easily incorporated into the algorithm.