TJNU multimodal ground-based cloud database (MGCD)
收藏资源简介:
TJNU多模态地面云数据库(MGCD)是在中国天津收集的,时间从2017年到2018年。它包含8000个多模态地面样本。根据世界气象组织(WMO)发布的国际云分类系统标准以及实践中云的视觉相似性,天空条件被分为七种类型:1) 积云,2) 高积云和卷积云,3) 卷云和卷层云,4) 晴空,5) 层积云、层云和高层云,6) 积雨云和雨层云,7) 混合云。此外,云量不超过10%的云图像被归类为晴空。MGCD由来自七个类别的4000个训练样本和4000个测试样本组成,每个样本包含一张地面云图像和一组多模态云信息,这些信息是一一对应的。云图像由带有鱼眼镜头的天空相机拍摄,并以JPEG格式存储,像素分辨率为1024×1024。多模态云信息由气象站收集,包括温度、湿度、压力和风速,并以四个元素的向量形式存储。所有样本均由气象学家和地面云研究人员共同标注。MGCD将免费提供给云相关研究人员,以促进研究。
TJNU Multimodal Ground-based Cloud Database (MGCD) was collected in Tianjin, China, from 2017 to 2018. It contains 8000 multimodal ground-based samples. Based on the International Cloud Classification System standards issued by the World Meteorological Organization (WMO) and the visual similarity of clouds in practical scenarios, sky conditions are categorized into seven types: 1) Cumulus, 2) Altocumulus and Cirrocumulus, 3) Cirrus and Cirrostratus, 4) Clear sky, 5) Stratocumulus, Stratus and Altostratus, 6) Cumulonimbus and Nimbostratus, 7) Mixed clouds. Additionally, cloud images with cloud coverage not exceeding 10% are classified as clear sky. MGCD comprises 4000 training samples and 4000 test samples across the seven categories, with each sample containing one ground-based cloud image and a set of paired multimodal cloud information. The cloud images are captured by a sky camera equipped with a fisheye lens, stored in JPEG format with a resolution of 1024×1024 pixels. The multimodal cloud information is collected by meteorological stations, including temperature, humidity, atmospheric pressure and wind speed, and stored as a four-element vector. All samples were jointly annotated by meteorologists and ground cloud researchers. MGCD will be freely available to researchers working on cloud-related studies to facilitate relevant academic research.
Multimodal-Ground-based-Cloud-Database
数据集概述
- 名称: TJNU multimodal ground-based cloud database (MGCD)
- 收集地点: 中国天津
- 收集时间: 2017年至2018年
- 样本数量: 8000个多模态地面样本
- 样本分类: 根据国际云分类系统标准和实际视觉相似性,将天空条件分为七种类型:
- 积云
- 高积云和卷积云
- 卷云和卷层云
- 晴空
- 层积云、层云和高积云
- 积雨云和雨层云
- 混合云
- 样本组成: 每个样本包含一张地面云图像和一组多模态云信息,两者一一对应。
数据详情
- 图像:
- 采集设备: 鱼眼镜头的天空相机
- 格式: JPEG
- 分辨率: 1024×1024像素
- 多模态云信息:
- 采集设备: 气象站
- 信息类型: 温度、湿度、压力和风速
- 存储格式: 四元素向量
数据集结构
- 训练样本: 4000个
- 测试样本: 4000个
- 标注: 由气象学家和地面云研究人员共同标注
使用许可
- 免费提供: 供云相关研究人员使用,以促进研究
- 下载前要求: 需阅读并理解MGCD Agreement,下载和使用数据集即表示同意所有限制和要求。
引用
- 参考文献:
- Liu, S., Li, M., Zhang, Z., Cao, X., & Durrani, T. S. (2020). Ground-Based Cloud Classification Using Task-Based Graph Convolutional Network. Geophysical Research Letters, 47(5), e2020GL087338.
- Liu, S., Li, M., Zhang, Z., Xiao, B., & Durrani, T. S. (2020). Multi-evidence and Multi-modal Fusion Network for Ground-based Cloud Recognition. Remote Sensing, 12(3), 464.




