Seismic Data Compression Using Linear Predictive Coding

摘要

This report is on the application of linear predictive coding, which has been applied successfully to many types of sample sequences taken from natural signals, to the compression of seismic data. Designs for encoding and decoding data from a single seismic channel and also from multiple seismic channels in a seismic array are studied. Single-channel prediction, as opposed to multiple-channel prediction using a reference signal, turns out to offer the largest reduction in overall data rate for a given accuracy of reconstruction. When applied to data taken during several NRSA events at 40 samples/s, single-channel predictive encoding consistently produced savings (relative to ordinary quantization) of approximately 200 bits/s over a wide range of data rates corresponding with varying degrees of accuracy of the encoded signals. With background data preceding these events the savings were less, but still approached 100 bits/s. (ERA citation 12:042305)