Well Log Segmentation in Spectral Domain

Abstract

Classic well log interpretation involves direct horizon mapping using log signature, attributes cross plot, etc to produce lithologic section for the delineation, exploration and production of hydrocarbon in oil and gas fields. The methods operate on recorded lithologic logs without adequate calibration. These result in interpretational ambiguities because of relatively poor resolution of well log owing to its recording in time, under sampling and coarse processing. In this paper, a new technique and algorithm for segmenting well log using discrete Fourier transform in the interpretation of well data obtained from the Niger Delta is presented. The aims and objectives of the study are to segment well logs into their constituent lithology in time domain, transform the well data from time to frequency domain and segment, and deduce viable diagnostic attributes such as magnitude, phase and frequency from the transform coefficients which could be used to identify the most probable zonation/contact in the well. The algorithm adopts Short time Fourier transform technique in the time to frequency transformation and is implementable on both standard and general seismic and well log interpretational platforms. It directly computes the spectral equivalent of the adopted lithologic log (Gamma-ray) and recovers hitherto lost frequency information. The results of the spectral decomposition of the well data yielded frequency (pseudo) logs that reveal subtle sub-well horizons and differences in lithology. By revealing masked horizons and better delineating and delimiting reservoirs, more hydrocarbons will be recovered and field development will be enhanced.

Keywords: Fourier transform, Spectral decomposition