ADAPTIVE LEAST SQUARES DIGITAL FILTERING ALGORITHMS FOR ONSHORE AND OFFSHORE OIL PROSPECTING

Abstract

Conventional seismic oil exploration is driven by the use of seismic detectors, viz. geophones (onshore) and hydrophones (offshore) for the detection and enhancement of seismic waves, etc. The problem, however, is that most of these equipment are not terrain sensitive as they were modeled with fixed coefficient digital filters, hence performing the same kind of enhancement on detected seismic waves irrespective of their origin(source) and geophysical characteristics. For instance, the same geophones/hydrophones that enhanced waves to locate oil in Niger Delta Nigeria were not effective in the location of oil in North Eastern Nigeria (Borno and the Chad Basins) even though oil had been commercially exploited in very geologically similar terrains in the nearby Chad Republic, hence the need for embedding variable (adaptive) coefficient digital filters in seismic detectors. Moreover, the quality of an inbuilt adaptive digital filter is dictated by the underling adaptive filtering algorithm. In this research, therefore, a composite adaptive deconvolution system that integrates a proposed hybrid Least Mean Square (LMS) and Recursive Least Squares (RLS) adaptive filtering algorithm with five existing algorithms is proposed. The composite model accepts input as reflections detected from rock interfaces (boundaries) in a prospective oil block. The system then removes echoes and reverberations using system’s identification principles before subjecting the emergent sequence (primary and secondary reflections) to adaptive deconvolution using a choice algorithm among multiple algorithms stacked for that purpose. The proposed system was implemented with MATLAB with a graphical user interface that shifts the choice of the algorithm for deconvolution to the user. Test data were acquired from the Borno State Department of Environment, Maiduguri, The Marine Geosciences Data System’s Warehouse and The Mathworks Inc. The output sequence (the estimated primary reflections), the error sequence, as well as the filter coefficient numbers/values were then graphically displayed for visual appraisal. Convergence was tested by comparing the output of each adaptive deconvolution algorithm with the standardized Albert Wiener’s signal deconvolution output. Results obtained by simulating the system with test data showed that the hybrid LMS/RLS algorithm converged faster to the Wiener’s coefficients at lower offsets and higher iteration values compared with any of the other algorithms. The LMS/RLS hybrid algorithm is therefore recommended for the implementation of seismic detectors for commercial seismic exploration of difficult terrains like the desert basins of Borno state in Nigeria.