2D Seismic reflection data filtering using Time Slice Singular Spectrum Analysis for noise suppression: A case study from Singareni coalfield, India

Complex noises that arise due to the nonlinear interaction of unwanted seismic signals (coherent and random noises), alter the primary reflections and create severe problems than the simple random noise in recognizing geological structures from seismic stack sections. We present here Time Slice Singular Spectrum Analysis (TSSSA) for the suppression of such noises from seismic records in time domain. The TSSSA involves organizing the spatial data (corrected for NMO) which corresponds to constant time into trajectory matrix for the reduction of noises that do not show large spatial coherency. The singular value decomposition based rank reduction of the trajectory matrix formulated from constant time slice helps to identify the noise in TSSSA with low Eigen values. We test the method on synthetic data contaminated with complex noises to demonstrate its 'robustness' for the identification of faults and then apply to high-resolution seismic reflection observations from Singareni coalfield, India. We find a good correlation between de-noised and pure synthetic data, which indicates the suppression of complex noise without any loss of seismic features. The application of TSSSA to pre and post stack seismic field data suggests significant improvement in signal to noise ratio. The reflections resembling the coal beds in the pre and post stack TSSSA processed depth sections clearly match with the reflectors in the synthetic trace generated from well log data. Finally, improvement in SNR and clear matching of fault structures and coal beds identified in the TSSSA processed data with regional fault structures and available geological information suggest the TSSSA as a robust method for seismic data conditioning.