OpenSeis-RAW: A Public Benchmark of Raw Seismic Surveys and Standardized Processing Chains

The geophysical community lacks an openly shareable, standardized corpus of \emph{raw} (pre-stack) seismic surveys with aligned intermediate processing targets. This gap hinders reproducible evaluation of modern learning-based and hybrid physics--AI pipelines for denoising, deghosting, multiple suppression, interpolation, velocity analysis, and migration. We propose \textbf{OpenSeis-RAW}, a two-tier dataset coupling (i) openly licensed field surveys where pre-stack traces are available, with (ii) high-fidelity synthetic surveys providing ground-truth subsurface models and per-stage supervision. OpenSeis-RAW defines task protocols, metrics, and a hidden test split for leaderboard benchmarking, enabling rapid, cross-disciplinary progress in subsurface imaging for energy transition (CCUS, geothermal) and offshore infrastructure.