High Performance Seismic Trace Compression

P. L. Donoho, R. A. Ergas, and J. D. Villasenor

We have developed an algorithm for the compression of seismic data capable of compression ratios in excess of 100:1 for a wide range of datasets representing both prestack and poststack data. Although the compression algorithm is lossy, in the sense that its use introduces noise into the final reconstituted dataset, the compression noise leads to no observable loss of geophysical information at compression ratios substantially greater than 1OO:1. The compression algorithm employs the wavelet transform to represent the data in a basis characterized by a number of subbands having different temporal and spatial frequency content. The coherency and redundancy of multidimensional seismic datasets, ranging from prestack field data to migrated 3D stack cubes, permits efficient qantization of the data by each wavelet-transform subband, such that a very small number of bits per data sample will accurately represent the data. The ability to compress data by such large factors leads to the possibility of transmitting entire field datasets from a multistreamer marine survey directly to a processing center in nearly real time. Complete 3D prestack datasets can be easily maintained online with realizable disk storage at 100:1 compression ratio. Network transmission and disk storage costs for other applications can also be reduced by large factors. The algorithm described here has been optimized for speed in compressing large 3D datasets, requiring typically the same amount of time as other processing steps, such as NMO.

AAPG Search and Discover Article #91019©1996 AAPG Convention and Exhibition 19-22 May 1996, San Diego, California