Improving velocity analysis beneath a submarine canyon using velocity re-datuming

Abstract

A number of large submarine canyons are present in the Gippsland Basin. These canyons cause seismic rays to be refracted in a complicated manner, such that in time-offset (t,y) space, events beneath the canyon do not have hyperbolic moveout. Using model seismic data constructed by ray-tracing over typical canyon profiles, CMP gathers are analyzed by considering three parameters measured in t2-y2 space: Vstack, To and the RMS error value. Vstack is calculated from the slope of the best-fit line, To is calculated from the intercept of the best-fit line, and the RMS error measures how badly the event picks are positioned with respect to this line. Reference values for Vstack, and To are Vrms and the two-way normal incidence time respectively and are calculated from the zero-offset model data. Analyzing the errors in the three parameters reveals that the Vstack parameter is the most sensitive to the presence of these canyons with relative errors exceeding 60%. Conventional replacement statics do not remove all of these errors. In particular, errors in To are increased, while errors in Vstack calculated with the water layer replaced, are only partially reduced when referenced to Vrms derived from the corresponding zero-offset, replaced model. A modified replacement static technique called ``velocity re-datuming'' when applied to the model seismic data, reduces the errors in Vstack beneath the canyon to about 5%. The method only requires a water bottom profile and is not computer intensive. It is used only to estimate velocities for post-stack migration. Field data are stacked without replacement statics to avoid the increased errors in To. Velocities derived by velocity re-datuming are applied to post-stack data for better migration and depth conversion of stacked seismic data in the Gippsland Basin.