Time-frequency Analysis Of Seismic Signals And Its Application In Gas-bearing Reservoir Detection

Marine carbonate reservoirs and tight sandstone reservoirs are the focus anddifficulty of oil and gas exploration and development in the future. Currently, mostreservoirs are deeply buried with the characteristics of weak seismic response. Porefluid response is feeble. And a small difference exists between the reservoir andnon-reservoir. These all lead to difficulties of the gas detection with this type ofreservoir. For more effectively conducting hydrocarbon detection of marine carbonatereservoirs and tight sandstone reservoirs and further improving the use of seismic datato fully exploit the effective geological information and further improving the accuracyof instantaneous attributes such as instantaneous amplitude and instantaneous frequency,etc., the empirical mode decomposition (EMD) based gas detection methods andtechniques are mainly studied. The main contents and results of the paper are asfollows:(1) Based on the evaluation of a variety of time-frequency analysis methods, theEMD-based time-frequency methods using for hydrocarbon detection are mainlystudied. The mechanism and characteristics of the conventional seismic signaltime-frequency methods are analyzed in detail. On this basis, the basic principles andthe current existing problems of the Hilbert-Huang transform (HHT) method are mainlydiscussed and the corresponding solutions are given to some extent. The mechanism ofhydrocarbon detection based on spectral decomposition technology is illustrated. Here,the EMD-based time-frequency methods include: HHT method, NHT method, HUmethod, EMD/TK methods and EMDWave methods. The EMD-based time-frequencymethods use IMF/IMFs which reflects the most oil and gas information to detecthydrocarbons, but not all of the original seismic signals. Such type methods expand thespectral decomposition analysis and attenuation gradient analysis techniques.(2) For direct hydrocarbon detection, the EMD-based time-frequency methodsmainly use the characteristics of the "low-frequency energy enhancement andhigh-frequency energy absorption ". Selecting the correct IMF/IMFs obtained from the EMD method is the premise of the effective use of the EMD-based time-frequencymethods to direct hydrocarbon detection. Reasonable choice of the lowfrequency/frequency band and the high frequency/frequency band by spectral analysisof the selected IMF/IMFs for spectral decomposition is the key. Model test and realseismic data analysis show that for gas detection, the EMD-based time-frequencymethods can give a good gas-containing statistical interpretation to the reservoir whichhas weak reflection amplitudes and weak oil and gas response characteristics.(3) The concept and the calculation method of formation absorption section areproposed. Model analysis and real seismic data processing show that the results offormation absorption section and common frequency section are consistent. Butformation absorption section can give a more intuitive gas-containing result. Otherwise,formation absorption section is conducive to fusion manipulation with the attributesextracted by other methods so as to give a final favorable gas-containing distribution.Formation absorption section technology can avoid the inconvenience occurred by aseries of common frequency sections analysis using the traditional spectraldecomposition method.(4) Gas-bearing detection technologies based on the NHT method and the HUmethod are developed. The principle and characteristics of HHT as a spectraldecomposition technique are analyzed. On this basis, for the HHT method restricted byBedrosian theory and Nuttall theory, the NHT method and the HU method areintroduced for hydrocarbon detection. The principles of the NHT method and the HUmethod are given and the spectral decomposition implementations are further illustrated.A comparison analysis between HHT、NHT and HU is conducted by model test and realseismic data processing. The results show that these three methods are all suitable forprocessing a large amount of seismic signals. The execution time of HHT is the shortestand the execution time of HU method is the longest. For Jingbian data, theinstantaneous amplitudes and instantaneous frequencies obtained by the NHT methodand the HU method are better than them from the HHT method. The endpointsdeviations of the NHT methods and the HU method are smaller than that of the HHTmethod. Considering all aspects, we believe that for the seismic datain Jingbian gasfield, the NHT method has better results than the other two methods.(5) One spectral decomposition technique named EMD/TK is proposed based onthe EMD method and TK energy operator. The basis of this method is that seismic waveenergy directly associates with the Kaiser established physical model. The majorcomponent which can reflect the details and the oil and gas information most is first extracted by the EMD method. The instantaneous attributes such as the instantaneousamplitude and the instantaneous frequency are further extracted by TK energyseparation algorithm. Instantaneous attributes extracted by TK energy separationalgorithm have higher accuracy than that of the Hilbert transform. And TK energyseparation algorithm has the ability to track instantaneously-varying special patternsmore quickly than that of the Hilbert transform. But TK energy separation algorithmapplies only to monocomponent signal, and the application of EMD ensures thefeasibility of TK energy separation algorithm. Model validation and real seismic dataprocessing and comparison of different methods show the effectiveness and superiorityof the EMD/TK method. The EMD/TK method can enhance the responsecharacteristics of oil and gas and give a better gas-containling statistical interpretation.(6) A spectral decomposition analysis and attenuation gradient analysis methodnamed EMDWave method is proposed based on the EMD method and the wavelettransform. The major component which can reflect the oil and gas information most isextracted by the EMD method. Spectral decomposition analysis based on the wavelettransform is conducted to the selected major component. The high-precision attenuationgradient analysis is then implemented combined with the least squares method. Thismethod combines the advantages of EMD and the wavelet transform. EMDWavemethod can highlight the response characteristics of oil and gas, and it can also givemore details. The real seismic data processing and comparison of different methodsshow the effectiveness and superiority of this method.(7) The application of the EMD-based time-frequency methods to marinecarbonate reservoir in the western Sichuan is conducted. In the study area, the depth ofthe target layer is about5000meters and seismic response characteristics are weak.Currently only three wells are in the region and logging data are less. Reservoirprediction and gas detection are relatively difficult. The seismic sections intersectingwell Chuanke1, Xinshen1and Xiaoshen1are analyzed respectively and the analysisresults are consistent with the logging information. These confirm that the resultsobtained by the EMD-based time-frequency methods are reliable. Model test and theanalysis on seismic section intersecting gas wells show that the prediction using theEMD-based time-frequency methods forecast in this region is feasible. The thicknessof the main gas production layer of well Chuanke1analyzed by the EMD-based time-frequency methods is consistent with that of the well logging data. This fact furtherindicates the validity and reliability of the EMD-based time-frequency methods. Finally,under the control of the horizons, slices along the horizons in the target area are analyzed and selected. The corresponding optimal formation absorption sections andattenuation gradient formation are extracted by using HHT, NHT, EMD/TK andEMDWave methods respectively. The favorable gas-containing distribution of the FMLeikoupo is given by the use of cluster analysis to the EMD-based time-frequencymethods.(8) The research on the application of the EMD-based time-frequency methods tothe tight sandstone reservoir in the Sulige gas field is carried out. The effectivereservoirs in the Sulige gas field distributed sporadically and the application of AVOanalysis technologies is subject to certain restrictions. Here, starting directly from theseismic records, the EMD-based time-frequency methods are studied and developed.Reservoir characteristics of the study area are given. The effectiveness of theEMD-based time-frequency methods is validated by using seismic section intersectingwell Zhao68etc. and model test further confirms the feasibility of using the EMD-basedtime-frequency methods to the tight sandstone reservoir. The results of thethree-dimensional data processing in the study area show that all of the class I and theclass II wells are detected and the majority of class III wells are detected. TheEMD-based time-frequency methods using for gas-containing detection are high inaccuracy and reliability. The EMD-based time-frequency methods can improve therecognition of the effective reservoir or the gas-containing reservoir.