Benchmarking Error-Bounded Compressors for Big Seismic Data

Abstract The rapid advancement in seismic data acquisition technology, with its high-channel counts and point-source, point-receiver systems, has resulted in unprecedented volumes of seismic data. Single sensor survey can easily generate hundreds of terabytes, challenging conventional storage and data management infrastructures. Efficient compression of seismic data emerges as a critical solution to mitigate storage demands and accelerate data processing workflows. This study evaluates several state-of-the-art lossy compressors -Bitcomp, SZ3, and ZFP - focusing on their compression effectiveness and the impact of compression errors on seismic data quality. We analyze performance metrics such as compression ratio and throughput for both compression and decompression, while also examining data fidelity through PSNR (Peak Signal-to-Noise Ratio) measurements in pre-stack and post-stack domains. Our work further includes a comparison with commonly used lossless compressors, highlighting their limited compression capability for seismic datasets.