6. Multi-scale Imaging/Interpretation Methodologies
Automatic Detection of Wavelets Interfering in Seismic Sections using Fractal Methods
Shaban Ali Gholamy(1), Abdolrahim Javaherian(1) and Abdolreza Ghods(2)
(1) University of Tehran, I. R. Iran. (2) Institute for Advanced Studies in Basic Sciences, Zanjan, I. R. Iran.
Abstract
Wavelets interfering from overlapping of seismic wavelets reflect from base and top of layers cause an ambiguity of reflectors detection. When depth increases, attenuation increases too and frequency content of the source wavelet tends to become lower; therefore, wavelength increases and consequently effect of interfering increases. In deeper horizons, increase in wave velocity causes interfering effects on thick layers too. In this paper, wavelets interfering are studied using fractal dimension of instantaneous attributes. Measurement of fractal dimension using moving window technique yields changes in fractal dimension that relates with roughness. If a seismic attribute has a large roughness in reflectors and low roughness in other places, its fractal dimension has a sharp peak in reflectors. Among instantaneous attributes (amplitude, phase and frequency), phase shows this property as well. We examined four fractal dimension measurement methods including divider, Hurst, spectral and variance fractal dimension. Since these methods are applied in moving window technique, optimum window length for each method has to be determined. We found that the Hurst method is the best method with window length of 50 samples that indicates reflectors position. In order to study wavelets interfering using fractal dimension of instantaneous attributes we generated two synthetic seismic sections with wedge shaped models with two types of wavelets, zero phase and minimum phase. One of the seismic sections includes a wedge shaped model that reflections of the upper and lower surface of the wedge have the same polarity, whereas the other model has the opposite polarity. We observed that the fractal analysis of instantaneous phase can separate interference of wavelets that their onsets have distance greater than in the same polarity and greater than in the opposite polarity. We examined the effect of signal to noise ratio on synthetic models. We found that signal to noise ratio of 10 has no effect on the fractal analysis results. Increasing signal to noise ratio, generates more than one peak for each reflector. In this case, onset of interfered reflectors cannot be determined.
Last modified: Thu May 25 10:01:34 2006