Zero offset seismogram estimation using inversion to fit common conversion point data in three component reflection data

Abstract:

Stacking seismic reflection data typically involves grouping data from many different source and receivers locations by common mid points and then stacking the data after making moveout corrections. Moveout corrections are differences between the travel times of a phase that travels to a depth point at an offset and a phase reflecting from the same depth at zero offset. After these travel time corrections are made, the data from all offsets are averaged to produce a seismic trace that reflects zero offset travel times and amplitudes of the midpoint location. These corrections are dependent on accurate estimation of the velocity model. Near surface variations in velocity are impossible to estimate from the reflection data, so static corrections (time delays applied to the data to correct for delays at the stations and receivers) are estimated to further align these data before stacking. To stack PdS in three component reflection data, there is the added complication of estimating either Vs or the Vp/Vs ratio. This procedure generally works well, but traditional stacking with moveout and static corrections fail to adequately stack data from some areas within a data set. For these areas, we will use inversion methods to find the zero offset trace that best matches the data in the CMP rather than conventional stacking. Our first step is to use the area where data were stacked correctly to estimate a reference pulse and construct the zero offset trace for the problem regions. Starting with the Vp, Vs, and t0 for the nearest trace that stacked well, we find the Vp, Vs, and t0 that corresponds to the moveout and static correction that best fit the problematic CMP. This simplex will be repeated several times until the same t0 is found to a high certainty. The phase on each trace in the CMP corresponding to that t0 is subtracted from the CMP. The process is repeated until no further horizons are identified.