This thesis investigates the application of the Empirical Mode Decomposition (EMD) and Ensemble Empirical Mode Decomposition (EEMD) techniques in de-noising of marine seismic data. EMD decomposes the seismic dataset into sub-datasets called Intrinsic Mode Functions (IMFs); the sum of these IMFs produces the original seismic data. EMD is a proven technique to capture the non-stationarity and non linearity of a signal; therefore, the key idea behind EMD is its use as a non-stationary filter. Marine seismic noise, such as swell, cablestrum, reflected interference, ground roll, and refracted waves, causes non-stationary seismic data and represents a major challenge. Therefore, I propose to investigate in this thesis if seismic noise can be separated from target reflections by using the EMD/EEMD techniques.
The motivation of this work is to establish a reliable methodology to de-noise seismic data using the EMD/EEMD techniques. The main idea is that non-stationarity caused by noise is collected in just a few IMFs that do not carry the target (primary) reflections in the seismicdataset. The filtered section can be obtained by subtracting these IMFs from the original data leading to signal-to-noise enhancement of the data. In order to benchmark the effectiveness of the method I will compare the filtered EMD/EEMD sections to filtered reference sections generated using the GeoCluster software propriety to CGGVeritas.