methaneset
收藏资源简介:
MethaneSET 是一个用于卫星甲烷羽流检测的 AI 就绪数据集集合,涵盖三种传感器(Sentinel-2、Landsat-8/9 和 EMIT)并具有统一的元数据。该数据集包含带有专家验证的羽流分割掩码的标记场景、用于自监督学习的无羽流场景、用于数据增强的 WRF-LES 模拟合成羽流库,以及基于 MODTRAN 的透射率查找表,用于物理信息羽流注入。数据集总规模约为 2.0 TB,包括约 5,500 个标记场景和约 55,000 个无羽流场景。MethaneSET 支持多光谱和高光谱传感器的甲烷检测任务,适用于气候变化研究和环境监测。数据集遵循 CC-BY-4.0 许可,并提供了详细的元数据字段和快速入门指南。
MethaneSET is an AI-ready dataset collection for satellite methane plume detection, covering three types of sensors (Sentinel-2, Landsat-8/9, and EMIT) with unified metadata. This dataset includes labeled scenes with expert-validated plume segmentation masks, non-plume scenes for self-supervised learning, a WRF-LES simulated synthetic plume library for data augmentation, and MODTRAN-based transmittance lookup tables for physics-informed plume injection. The total size of the dataset is approximately 2.0 TB, containing around 5,500 labeled scenes and 55,000 non-plume scenes. MethaneSET supports methane detection tasks for multispectral and hyperspectral sensors, and is applicable to climate change research and environmental monitoring. The dataset is licensed under CC-BY-4.0, and provides detailed metadata fields and a quick start guide.



