Detailed elastic modelling to characterise noise contributions to seismic data from the Gippsland Basin

Abstract

Strong noise contributions currently prevent adequate definition of deep exploration targets in marine seismic data from the Gippsland Basin. We computed an elastic synthetic seismogram (Kennett, 1979, 1980) from a highly detailed depth model to match the raw field records with considerable accuracy. As part of a sensitivity analysis, we computed additional synthetics to gauge the effects of poorly known parts of the depth model upon the elastic synthetic. These synthetics also helped us to identify noise events in the field records. Subsequent processing of the synthetics led to the identification of persistent noise contributions to stacked data from the Gippsland Basin. Noise contributions in the target zone consist of "shear reflections" resulting from strong mode conversion occurring at the Miocene channels and the Latrobe Group coal/shale sequence. These formations then reconvert the S-waves to P-waves so that they are recorded by the hydrophones. The timing and curvature of the shear reflections is such that they strongly interfere with primary reflections in the target zone stack response. At greater depths, the stack response is dominated by long-period multiples associated with the coal sequence, the Miocene channels, the hard sea floor and the sea surface. Lateral variation of these noise interferences is due to structural variation in the Miocene channels and the lower part of the coal sequence. Strong interbed multiples (including mode converted peglegs) are generated in the coal sequence and constitute a second type of noise interference by preventing direct interpretation of the target zone primary reflections. Our elastic synthetics also characterised shear refractions in the field data using a compactional trend to derive the shallow depth model. The modelled shear refractions were highly sensitive to the shallow shear velocity profile and this makes them an important elastic modelling constraint. Further variation of the shear velocity model also revealed the insensitivity of stacked elastic synthetics to our 'artificial Vs log' based on approximate formation s values. The amplitude and timing of prominent reflections barely changed as we adjusted their formation s values.