Here we present a case study of improved subsalt imaging for a wide azimuth (WAZ) survey in the Mississippi Canyon/Atwater Valley area in the Gulf of Mexico. The key technologies to impact the subsalt images are: 1) WAZ acquisition, 2) True Azimuth Multiple Elimination (TAME), 3) Tilted Transverse Isotropic (TTI) model building and imaging, and 4) Reverse Time Migration (RTM) based Delayed Imaging Time (DIT) scans to update the subsalt velocities. The area was previously imaged with Narrow Azimuth (NAZ) Vertically Transverse Isotropic (VTI) Kirchhoff migration and WAZ VTI Kirchhoff and RTM algorithms. Application of TTI RTM has resulted in significant improvements of subsalt images.

No abstract

We present a case study of enhanced imaging of wideazimuth data from the Gulf of Mexico utilizing recent technologies; and we discuss the resulting improvements in image quality, especially in subsalt areas, relative to prior methodologies. The input seismic data set is taken from the large scale Freedom WAZ survey located in the Mississippi Canyon and Atwater Valley areas. In the course of developing the enhanced wide-azimuth processing flow, the following three key steps are found to have the most impact for improved subsalt imaging. 1) Data regularization to prepare the data for multiple attenuation as well as for the final run of anisotropic reverse time migration; 2) 3D true azimuth SRME to remove multiple energy, in particular, complex multiples beneath salt; 3) reverse time migration based delayed imaging time (DIT) scan to update the complex subsalt velocity model. The DIT scan further improves the accuracy of the subsalt velocity model after the conventional ray-based subsalt tomography. In this paper, we focus on the depth imaging aspects of the project, with particular emphasis on the application of the DIT scanning technique. We also demonstrate the uplift of acquiring a wide-azimuth data set relative to a standard narrowazimuth (NAZ) data set.

No abstract

Exascale architecture computers will be limited not only by hardware but also by power consumption. In these bounded power situations, a system can deliver better results by overprovisioninghaving more hardware than can be fully powered. Overprovisioned systems require power to be an integral part of any scheduling algorithm. This paper introduces a system called PANN that uses neural networks to dynamically allocate power in overprovisioned systems. Traces of applications are used to train a neural network power controller, which is then used as an online power allocation system. Simulation results were obtained on traces of ParaDiS and work is continuing on more applications. We found in simulations PANN completes jobs up to 24% faster than static allocation. For tightly constrained systems PANN performs 6% to 11% better than Conductor. A runtime system has been constructed, but it is not yet performing as expected, reasons for this are explored.