Non-uniform optimal sampling for simultaneous source survey design

Abstract

Summary Optimal selection of locations for sensors in a seismic survey has been a long-standing issue for geophysicists. If we could sample the earth at two points per wavelength or better in all dimensions according to Nyquist sampling theory, design would not be an issue. The reality of limited access and funding requires us to make do with orders of magnitude fewer sampling points than Nyquist theory would dictate. The field of Compressive Sensing provides a new theory for non-uniform sampling that allows for using significantly fewer sensors than current practice in seismic exploration (Herrmann, 2010). We use these principles to define a pragmatic framework for seismic survey design, acquisition, processing and imaging, that we refer to as Compressive Seismic Imaging (CSI). In previous work we have described the CSI frameworks used for creating optimally sampled locations for sources and receivers that maximize our ability to recover the available bandwidth in seismic data. These same principles can be used to design surveys that use multiple simultaneous sources. In this paper we describe work flows for designing Non-Uniform Optimal Sampling (NUOS) locations for sources that maximize our ability to de-blend the data at high signal-to-noise ratios back to individual shot records. These work flows were used to design a blended dual-source survey that was shot immediately after the completion of a traditional single-source survey. Shooting time for the blended survey was reduced by more than half, with comparable or better data quality obtained for the blended source survey compared to the single source survey. After simple fast track processing, 4D differences between the blended and single source data were comparable to those obtained for 4D projects with similar geometries in nearby areas.