AnonymouScientist/SiliciclasticReservoirs

SiliciclasticReservoirs数据集包含100万个合成的三维硅质碎屑岩储层地质立方体，这些立方体基于规则沉积模拟生成，模拟类型包括浊积叶状体、6种河流河道架构和三角洲扇状分流。每个立方体体素化为(64, 64, 32)单元格，每个样本包含岩相（二进制和6类）、孔隙度、渗透率以及一组结构化的地质条件参数。数据集专门设计用于训练可解释物理条件下的地下地质条件生成模型（如流匹配、扩散等）。数据集按8种储层架构分层，提供紧凑和完整的数据格式，支持文本条件实验，并具有完全可重现性。

SiliciclasticReservoirs is a dataset of 1,000,000 synthetic 3D siliciclastic-reservoir geology cubes generated from rule-based sedimentological simulations, including turbidite lobes, 6 fluvial-channel architectures, and delta-fan distributary. Cubes are voxelized at (64, 64, 32) cells. Each sample carries facies (binary and 6-class), porosity, permeability, and a structured set of geological conditioning parameters. It is designed to train conditional generative models (e.g., flow matching, diffusion) of subsurface geology under interpretable physical conditioning. The dataset is stratified across 8 reservoir architectures, offers compact and full reproducibility formats, includes captions for text-conditioning experiments, and is fully reproducible.