Publications

The spectral decomposition attributes, which is analyzed in the form of time-frequency domain, is believed to be effectively differentiated both lateral and vertical lithologic and pore-fluid changes. There are some common methods for generating time-frequency spectrum such as Short Time Fourier Transform (STFT), Continues Wavelet Transform (CWT) and Matching Pursuit Decomposition (MPD). However, all of the methods need an intensive computation and predefined function or filter such as mother wavelet in CWT or wavelet dictionary in MPD. The last development, which has superiority in improving signal-to-noise ratio and fast computation, is Empirical Mode Decomposition (EMD) and S-Transform.In this paper, we perform the comparison of the spectral decomposition algorithms, those are EMD and S-Transform for characterizing the hydrocarbon reservoir in the meaning of time frequency analysis. The advantage and disadvantage of these two methods are compared particularly in providing the quality of timefrequency spectrum. In addition, we perform the analysis of low frequency shadow zone, which is produced by these two methods, to delineate the existence of the hydrocarbon. In its implementation, the EMD offers simplicity, which does not need a predefined function and provides good quality of time-frequency spectrum. In other hand the S-Transform provides less quality than EMD but it performs fast computation. Thus, EMD is more suitable for seismic data processing for better quality than S-Transform.Our experiences on the seismic data set of East Java basin shows that these two methods is useful for finding direct hydrocarbon indicator (e.g. low frequency shadow zone) and delineating reservoir distribution based on time-frequency energy spectrum and iso-frequency slices.Keywords: Spectral Decomposition, EMD, S-Transform and Hydrocarbon Reservoir.