一种地震数据有损压缩方法及误差分析

Lossy compression method and its error analysis for seismic data

对某地震处理系统将32位浮点数压缩为N位有符号短整型数的方法进行了推算和分析。得出了其压缩和优化的方法,认为其压缩前、后的误差基本为10^(-5)数量级,压缩率约为60%,远高于常用无损压缩软件的压缩率(约为95%)。进一步讨论了舍入造成的差异及提高压缩率的可能性。通过某工区资料展示了压缩方法对实际数据的应用效果,表明了压缩方法的可用性。

The method of compressing 32-bit floating point number into N-bit signed short integer number in a seismic processing system is deduced and analyzed. The method of compression and optimization is proposed. The error before and after the compression is about 10-5 order of magnitude, the compression ratio is about 60%, much higher than that of lossless compression software(about 95%). The difference caused by rounding and the possibility of increasing compression ratio are discussed further. The application effect of the compression method to the actual data is demonstrated by using the data of a certain working area, and the availability of the compression method is demonstrated also.