NONE

1. OverviewThe Global_Extent_of_Human_Regulated_Water_Bodies.shp (GHRW) dataset provides a comprehensive global inventory of human-regulated water bodies (HRWs), identified using remote sensing and deep learning techniques. The dataset includes 1.1 million HRWs, ranging in size from 0.001 km², with a complete record for HRWs larger than 0.1 km². 2. Production MethodologyThe dataset was produced through a multi-step process involving the identification, classification, and validation of water bodies using the Joint Research Centre Global Surface Water (GSW) dataset, Sentinel-2 imagery, and a deep learning classification model (EfficientNet_V2_S). The GSW dataset was utilized to extract water bodies larger than 0.001 km², and Sentinel-2 cloud-free composite images were used to classify these water bodies into human-regulated or natural categories. The deep learning model was trained with manually labeled HRWs and natural water bodies, followed by iterative refinement using a human-in-the-loop approach. 3. Attribute DescriptionsThe dataset consists of the following attributes: area_km2: This attribute represents the area of each water body in square kilometers (km²). It was calculated from the GSW maximum extent layer and other data sources. source: In cases where the developed product overlapped with datasets from CRD, GOODD, or GeoDAR, we prioritized the sources in the following order: CRD, followed by GeoDAR, and finally GOODD. Specifically, if an overlap occurred between the developed product and any of these datasets, we retained the source data from CRD as the highest priority. If CRD data was not available, we used data from GeoDAR, with GOODD being used as the last option. 'Hao et al. 2024': Data directly derived and classified using the methodologies described in Hao et al. 'GeoDAR v1.1': HRWs identified from the GeoDAR v1.1 dataset. 'GOODD': HRWs identified from the GOODD dataset. 'CRD': HRWs identified from the CRD dataset. lon: Longitude coordinate of the centroid of the water body. lat: Latitude coordinate of the centroid of the water body. Reference: Hao, Z. (2024). Global_Extent_of_Human_Regulated_Water_Bodies (GHRW) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.13777813 Song, C., Fan, C., Zhu, J., Wang, J., Sheng, Y., Liu, K., ... & Ke, L. (2022). A comprehensive geospatial database of nearly 100 000 reservoirs in China. Earth System Science Data, 14(9), 4017-4034. Wang, J., Walter, B. A., Yao, F., Song, C., Ding, M., Maroof, A. S., ... & Wada, Y. (2021). GeoDAR: Georeferenced global dam and reservoir dataset for bridging attributes and geolocations. Earth System Science Data Discussions, 2021, 1-52. Mulligan, M., van Soesbergen, A., & Sáenz, L. (2020). GOODD, a global dataset of more than 38,000 georeferenced dams. Scientific Data, 7(1), 31.1. 概述 全球人工调控水体范围（Global_Extent_of_Human_Regulated_Water_Bodies.shp，简称GHRW）数据集提供了一套全面的全球人工调控水体（human-regulated water bodies，简称HRW）普查数据，通过遥感与深度学习技术识别获取。该数据集共计包含110万个人工调控水体，水体面积覆盖0.001平方千米以上区间，其中面积大于0.1平方千米的人工调控水体拥有完整记录。 2. 制作方法 本数据集通过多阶段流程制作完成，包括利用欧盟联合研究中心（Joint Research Centre，简称JRC）全球地表水体（Global Surface Water，简称GSW）数据集、Sentinel-2影像以及深度学习分类模型（EfficientNet_V2_S）对水体进行识别、分类与验证。首先借助GSW数据集提取面积大于0.001平方千米的水体，随后利用无云合成的Sentinel-2影像将这些水体划分为人工调控与自然水体两类。该深度学习模型以人工标注的人工调控水体与自然水体样本进行训练，并通过人机协同（human-in-the-loop）的方式进行迭代优化。 3. 属性说明 本数据集包含以下属性： area_km2：该属性代表每个水体的面积，单位为平方千米（km²），其计算基于GSW最大范围图层与其他数据源。 source：当本数据集成果与CRD、GOODD或GeoDAR数据集存在重叠时，我们按照以下优先级选取数据源：CRD优先，其次为GeoDAR，最后为GOODD。具体而言，若本成果与上述任一数据集存在重叠，则优先保留CRD的源数据；若无CRD数据，则采用GeoDAR数据；仅当前两者均不可用时，才使用GOODD数据。 以下为具体来源说明： 'Hao等（2024）'：通过Hao等提出的方法直接衍生并分类得到的数据。 'GeoDAR v1.1'：从GeoDAR v1.1数据集中识别得到的人工调控水体。 'GOODD'：从GOODD数据集中识别得到的人工调控水体。 'CRD'：从CRD数据集中识别得到的人工调控水体。 lon：水体质心的经度坐标。 lat：水体质心的纬度坐标。 参考文献： Hao, Z. (2024). 全球人工调控水体范围（GHRW）[数据集]. Zenodo. https://doi.org/10.5281/zenodo.13777813 Song, C., Fan, C., Zhu, J., Wang, J., Sheng, Y., Liu, K. 等 (2022). 中国近10万座水库的综合地理空间数据库. 地球系统科学数据, 14(9), 4017-4034. Wang, J., Walter, B. A., Yao, F., Song, C., Ding, M., Maroof, A. S. 等 (2021). GeoDAR：用于衔接属性与地理位置的全球参考大坝与水库数据集. 地球系统科学数据讨论, 2021, 1-52. Mulligan, M., van Soesbergen, A., & Sáenz, L. (2020). GOODD：全球超3.8万个地理参考大坝数据集. 科学数据, 7(1), 31.