YOLOv8-Detected Gravity Wave Event Catalog from VIIRS Day-Night Band Imagery (NOAA-21, 2023–2024)

This dataset contains gravity wave event lists detected by a YOLOv8 machine-learning model applied to VIIRS Day/Night Band (DNB) imagery acquired by the NOAA-21 satellite from 2023 to 2024. The detailed methodology is described in a manuscript being prepared for submission to Journal of Geophysical Research: Machine Learning and Computation: Yuta Hozumi, Jia Yue, Seraj Al Mahmud Mostafa, Chenxi Wang, Jianwu Wang, Sanjay Purushotham, and Steven D. Miller, “Localization and Classification of Gravity Wave Events from VIIRS Day/Night Band Satellite Imagery Using Machine Learning Techniques.” Dataset 1: wave_event_VJ202DNB.csv This file provides the catalog of detected wave events in CSV format. The event list is the result of a survey of moonless NOAA-21 VIIRS/DNB images from 2023 to 2024 using the trained YOLOv8 model. The columns are: 1. Satellite name. 2–3. Intensity and geolocation data file names, respectively. The original data files can be downloaded from https://ladsweb.modaps.eosdis.nasa.gov/archive/allData/5201/VJ202DNB/ and https://ladsweb.modaps.eosdis.nasa.gov/archive/allData/5201/VJ203DNB/. 4. Observation time of the granule in UTC. The format is “YYYY-MM-DD hh:mm:ss,” where YYYY, MM, DD, hh, mm, and ss denote year, month, day, hour, minute, and second, respectively. 5. Total number of detected objects in the granule. 6. Object index within the granule. 7–8. Class ID and class name of the detected wave event, respectively. The class IDs are: 0 for concentric gravity waves (cgw), 1 for frontal waves, 2 for ripples, and 3 for other gravity waves (gw). 9. Detection confidence score, ranging from 0.3 to 1.0. 10–13. Normalized coordinates of the center of the bounding box (x_center, y_center), followed by the normalized width and height of the bounding box. 14–15. Latitude and longitude of the center of the bounding box. 16–23. Latitudes and longitudes of the four corners of the bounding box. Dataset 2: YYYYMM.zip Monthly zip archives of DNB images with detected wave events annotated by bounding boxes. Dataset 3: map_figs.zip Monthly global maps of detected wave events for each wave class.本数据集包含由YOLOv8机器学习模型（YOLOv8）应用于NOAA-21卫星2023至2024年获取的VIIRS昼夜波段（Day/Night Band, DNB）影像所检测到的重力波事件列表。 详细研究方法已撰写为手稿，拟投稿至《地球物理研究杂志：机器学习与计算》（Journal of Geophysical Research: Machine Learning and Computation），作者为Yuta Hozumi、Jia Yue、Seraj Al Mahmud Mostafa、Chenxi Wang、Jianwu Wang、Sanjay Purushotham及Steven D. Miller，论文标题为《基于机器学习技术的VIIRS昼夜波段卫星影像重力波事件定位与分类》（"Localization and Classification of Gravity Wave Events from VIIRS Day/Night Band Satellite Imagery Using Machine Learning Techniques"）。 数据集1：wave_event_VJ202DNB.csv 该文件以CSV格式提供检测到的重力波事件目录，此事件列表为借助训练完成的YOLOv8模型，对2023至2024年无月NOAA-21 VIIRS/DNB影像开展检索所得的结果。 各字段说明如下： 1. 卫星名称。 2–3. 分别为强度与地理定位数据文件名。原始数据文件可从以下地址下载：https://ladsweb.modaps.eosdis.nasa.gov/archive/allData/5201/VJ202DNB/ 与 https://ladsweb.modaps.eosdis.nasa.gov/archive/allData/5201/VJ203DNB/。 4. 该影像幅的观测时间（UTC），格式为"YYYY-MM-DD hh:mm:ss"，其中YYYY、MM、DD、hh、mm、ss分别代表年、月、日、时、分、秒。 5. 该影像幅中检测到的目标总数量。 6. 该影像幅内的目标索引。 7–8. 分别为检测到的重力波事件的类别ID与类别名称。类别ID对应关系如下：0代表同心重力波（concentric gravity waves, cgw），1代表锋面波，2代表涟漪，3代表其他重力波（other gravity waves, gw）。 9. 检测置信度评分，取值范围为0.3至1.0。 10–13. 边界框中心的归一化坐标（x_center, y_center），随后为边界框的归一化宽度与高度。 14–15. 边界框中心的纬度与经度。 16–23. 边界框四个角点的纬度与经度。 数据集2：YYYYMM.zip 该文件为按月归档的DNB影像压缩包，其中包含已用边界框标注的检测到的重力波事件。 数据集3：map_figs.zip 该压缩包包含各波浪类别的检测重力波事件月度全球分布图。