Introduction:

Reverse Time Migration (RTM) is a powerful seismic imaging approach, widely used for migrating areas of complex structures of steep-dips and subsalt regions, despite its high computational cost. As in our work we are solving the isotropic acoustic 2nd order wave equation using explicit time domain finite differences (FD), we can identify the most computational part as follows : The FD kernel by itself , the boundary condition, the way we handle snapshots and the communication between domains.

Seismic data management is facing continuous challenges with growing amounts of data while traditional storage technologies are not equipped to handle that volume with evolving workloads. Seismic data stored in a standard file consisting in a set of headers and wiggle traces, a trace being a time-series of waves’ amplitudes reflecting in the Earth. A seismic survey translates into hundreds of terabytes where existing parallel file systems on rotating media and SSDs today suffer from a number of issues causing performance degradations.

source code available here

Advanced computing technologies are extending the performance limits of computing power by increasing the number of microprocessor processing cores, clock rate, and word size; thereby, making advanced seismic imaging and migration techniques, such as Reverse-Time-Migration (RTM) economically feasible. The RTM seismic imaging approach is widely used for migrating areas of complex structures. The standard implementation of keeping a snapshot for every time step is the easiest though most IO traffic intensive implementation. Handling snapshots in memory has clear limitations due to large footprint requirements, extra data movement and extra communication between domains that frequently generated load unbalance.

Advanced computing technologies are extending the performance limits of computing power by increasing the number of microprocessor processing cores, clock rate, and word size; thereby, making advanced seismic imaging and migration techniques, such as Reverse-Time-Migration (RTM) economically feasible. The RTM seismic imaging approach is widely used for migrating areas of complex structures. The standard implementation of keeping a snapshot for every time step is the easiest though most IO traffic intensive implementation. Handling snapshots in memory has clear limitations due to large footprint requirements and the penalty of moving extra data; this would work only for a very small workload.

Pre-stack Time Migration (PsTM) has been widely adopted as a flexible seismic migration method. PsTM is computationally intensive even with the high performance clusters; some 3D seismic data may take days or weeks to migrate the final image. The most intensive PsTM’ modules are travel-time calculations and diffraction summations; both of which exhibit trade off between travel-time computations and other resources.