4. Data Processing/Signal Processing
Investigation of using Fractal Geometry for Vertical Resolution Improvement of Seismic Data
Hoda Tahani(1) and Siavash Tarkhan(2)
(1) EOR Research Institute-Research and Development Directorate of NIOC(National Iranian Oil Company), Iran. (2) Research and Development of KEPCO(Khazar Exploration and Development Company) of NIOC., Iran.
Abstract
In petroleum engineering, there is no sufficient information about subsurface geology and layering. Seismic profiles can not often accommodate information more then one billionth of reservoir volume. With this limited data, all of statistical determination is just in doubt. So, we try to increase the seismic data if possible. One of the ways is to do this with fractal interpolation. Fractal geometry is a new concept to describe the natural shapes and is based on that natural shapes are self-similar and follow the power-law; it means that in different scales have similar properties. In other words, it is a collection of things that have fractal dimensions (non-round). Studies show that seismic traces are fGn. fGn s are sequences with fractal properties that are used for description of dispersion of reservoir and subsurface layers characteristics. Therefore, using of fractal model for subsurface modeling can be suggested. Using these parameters and fractal interpolation, the numbers of points in a seismic trace are increased, or produced signal have a lot of points more than the original trace. In this paper, we investigate usage of fractal geometry for seismic data improvement. For this purpose, 5 neighbor seismic traces of a stacked section from Bushgan oil field of Iran have been selected. These seismic traces are recorded with sampling rate of 2 ms in 7 seconds. For investigation of fractal geometry, at first, five fractal signals with sampling numbers of 4 times more than the initial seismic trace has been produced. So, the software was written for scaling and interpolating between fractal signal and seismic trace, the obtained signal by software is called "Noise". By adding Noise to original seismic trace, a trace affected fractal geometry is produced that only in sampling points is similar to the previously one, but, the curve is changed between two adjacent samples (2ms) that this can increase vertical resolution of seismic data, if subsurface layers have more fractal properties and self-similarities. Of course it should be mentioned that with each parameter and in every computer run, different signal "Noise" is obtained. This is because of statistical increment vertical resolution of subsurface layers and if we do R/S or Fourier analysis produced signals, we reach to the same number. In this paper, we just investigate fractal geometry effects on seismic data. So, ideal parameters and conditions of this method were not considered that it should be continued in the future.
Last modified: Fri Apr 21 20:40:33 2006