Smoothing spectral gamma logs: A simple but effective technique

Abstract

Spectral gamma‐ray logs suffer from unusually poor counting statistics because of the extremely low counting rates used to compute the concentrations of potassium (K), uranium (U), and thorium (T). Filtering is therefore a prerequisite to interpretation. Kalman filtering has been suggested, but this approach is complex and involves uncertain assumptions. Simple weighted averaging, on the other hand, fails to take into account abrupt changes that can occur in geologic response. Effective filtering of real logging data is possible, however, by a simple adaptive filter which uses the total gamma responses of a gamma‐ray tool to compute filter weights based on the error function. This filter is mathematically and computationally simple to implement for real time or postprocessing of spectral gamma logs.