Time-lapse Processing of 2D Seismic over the Japan Trench
Abstract
The seismic surveys SR97-101 (baseline) and D19 (monitor), acquired over the JapanTrench, were processed using an open-source software (Madagascar) with aims ondetecting changes from the Tohoku-Oki earthquake. The surveys were interpolated toa datum plane using a modified Shepard algorithm together with differential NMOfunction to fix the source and receiver coordinates. The data were then processedthrough frequency filter, despiking, debubble filter, and top mute. After sorting thedata into CMP gathers, velocity scanning and picking were used with inner mutedsemblances in order to avoid picking of multiples when using an automatic pickingalgorithm. The data were NMO corrected with smoothed velocity models where themonitor survey had refractional events being overcorrected at shallow depths, whichwere solved by reducing the fold prior to stacking. The data were then stacked toobserve the first seismic images of the subduction zone.
The CMP gathers were then processed through a Kirchhoff pre-stack time migrationand an experimental depth migration was performed using the full acoustic wavepre-stack migration on the common shot gathers. Similar overcorrections wereobserved on the shallow depth common offset image gathers obtained from the timemigration, and where the monitor survey went through similar procedures done priorto stacking. The migrated stacks were converted to SEGY format for interpretation inPetrel where the baseline survey was shifted to the east due to problems with thenavigation and to make the stacks comparable. Horst and graben structures, thrustcomplex, and deformations zone were characterized on the baseline survey andcompared with the monitor survey. 11.84 m uplift was observed on the backstopinterface on the Okhotsk Plate while 5.92 m subsidence was observed on the OceanicPlate near the trench with high uncertainties due to the navigational shift. Largechanges on the horst structure on the Okhotsk Plate were also observed, butdiscussed to be unrealistic. The depth migration was unstable due to unstable depthvelocity models and resulted in depth errors, but managed to image reflectors atshallow depth with better resolution compared to the time migration.