This paper utilizes a time-frequency analysis tapproach based on the wavelet transform and the observation that the presence of oil and/or gas results in an increase in the high frequency attenuation gradient and an increase in low frequency energy. Application of these methods in a prospective reservoir suggests that the combined use of both the high and low frequency information may be a more robust method for locating hydrocarbons than the use of either alone.
The study of scattering theory indicates that rocks containing oil or gas will attenuate the seismic signal, with higher frequencies more attenuated than lower frequencies low-frequency energy information can be utilized to make oil and gas prediction. When a reservoir contains fluid or gas, the seismic wave propagation will appear with effect of low frequency resonance scattering, leading to the seismic energy of low-frequency abnormally amplified, that is low-frequency energy grow relatively stronger, therefore the seismic of low-frequency information can indicate hydrocarbon information. In many studies of the time-frequency analysis technique, attention has focused on the use of high-frequency information, at the expense of the low-frequency data for detecting oil and gas. With improvements in seismic data acquisition, we can get more reliable low-frequency information. In this study we describe how using both high frequency attenuation and summed low frequency energy may lead to more reliable hydrocarbon detection. Base on oil gas testing data of well Ma2, we choose the gas layer and non-gas layer of well Ma2 to analyze the spectrum response. As shown in its result (Fig.1), the gas layer shows relative high-frequency energy attenuatioin and low-frequency energy enhancement. So we can use this spectrum characteristic to predict the hydrocatrbon layer.
Time-frequency analysis is a commonly used signal processing method that combines time and frequency domain information Cara (1982). This well known method uses the Fast Fourier Transform to convert time-domain seismic data into the frequency domain, and can provide Time-frequency analysis methods mainly experience a short-time Fourier transform, wavelet transform, etc. Gabor (1946) proposed the concept of short time Fourier transform. The actual signal analysis requires lowfrequency signal to obtain a higher frequency resolution, and high-frequency signal to obtain a higher time resolution, which needs window function to adjust its width and bandwidth according to the signal frequency changes. Apparently, Fourier transform can not meet the need of high-precision time-frequency analysis in a short time. Morlet (1984) discovered an obvious feature of signals in his analysis of synthetic seismic, the low-frequency end of the signal requires a very high frequency resolution, while the frequency resolution of high-frequency end can be lower. Based on Heisenberg’s uncertainty principle, It is that kind of the high-frequency signals has high time resolution, and the time resolution in low-frequency can be lower. According to this characteristic of seismic signals, Morlet proposed wavelet transform. Instantaneous spectral analysis (ISA) based on the wavelet transform has solved the problem that length of time window will affect the results, and has greatly improved the stability and resolution.
