Research paperAn improved lossless group compression algorithm for seismic data in SEG-Y and MiniSEED file formats

- An algorithm was proposed to compress losslessly the SEG-Y and miniSEED Files.
- It describes the distribution of SEG-Y and miniSEED files using Gini coefficient.
- Different data sub-groups should use different compression algorithms.
- The improved algorithm can achieve a better compression ratio of 75-80%.
- It provides a better compression ratio than the Steim2 to miniSEED Files.

An improved lossless group compression algorithm is proposed for decreasing the size of SEG-Y files to relieve the enormous burden associated with the transmission and storage of large amounts of seismic exploration data. Because each data point is represented by 4 bytes in SEG-Y files, the file is broken down into 4 subgroups, and the Gini coefficient is employed to analyze the distribution of the overall data and each of the 4 data subgroups within the range [0,255]. The results show that each subgroup comprises characteristic frequency distributions suited to distinct compression algorithms. Therefore, the data of each subgroup was compressed using its best suited algorithm. After comparing the compression ratios obtained for each data subgroup using different algorithms, the Lempel-Ziv-Markov chain algorithm (LZMA) was selected for the compression of the first two subgroups and the Deflate algorithm for the latter two subgroups. The compression ratios and decompression times obtained with the improved algorithm were compared with those obtained with commonly employed compression algorithms for SEG-Y files with different sizes. The experimental results show that the improved algorithm provides a compression ratio of 75-80%, which is more effective than compression algorithms presently applied to SEG-Y files. In addition, the proposed algorithm is applied to the miniSEED format used in natural earthquake monitoring, and the results compared with those obtained using the Steim2 compression algorithm, the results again show that the proposed algorithm provides better data compression.