本数据集包括全球水库清单（GREI_v1）和全球水库泥沙淤积数据库（GREI_Sed_v1）。GREI_v1记录全球555,960座面积大于0.001 km²水库的空间位置、边界及属性信息，包括水库名称、面积、位置、库容、平均水深等，总面积为469,639.9 km²，其中85,843座水库具有核验库容信息，总库容为6,851.7 km³；GREI_Sed_v1在该清单基础上，利用6,133个实测泥沙淤积样本和机器学习模型估算全球水库年淤积率与库容损失。该数据集弥补了现有全球水库数据库对小型水库和新建水库覆盖不足的缺陷，可服务于水资源管理、水文学、泥沙输移、气候适应与可持续发展研究。\n
数据详细信息参见：Liu, K., Fan, C., Song, C. et al. Global patterns of reservoir sedimentation and overlooked risks in small reservoirs. Nat Sustain (2026). https://doi.org/10.1038/s41893-026-01859-y", "ds_source": "
数据包主要包括：GREI_v1.gdb（全球水库空间清单及属性）、GREI_Sed_v1.gdb（全球水库泥沙淤积估算及相关属性）、Sedimentation_sample_sources.xlsx（6,133个实测淤积样本来源）、GREI_data_sources.xlsx（数据源汇总）和GREI_data_dictionary.xlsx（变量、数据类型、单位与属性定义）。水库位置来源包括GRanD、FHReD、GOODD、GeoDAR、GREI-p2k、Geo-referenced Database on Dams等全球数据库，16个国家/区域尺度数据集（包括中国、美国、印度等），以及OpenStreetMap。水库边界主要基于Global Surface Water（GSW）和GLAD长期水体发生数据（30 m Landsat，1984–2020），小型水库（<0.01 km²）进一步使用10 m Sentinel-2水频率数据（2019–2021）补充。泥沙淤积样本来源于中国全国水库淤积调查、美国Reservoir Sedimentation Survey Information System-II、印度Central Water Commission资料，以及Web of Science文献检索获取的100余篇相关研究。", "ds_process_way": "
数据生产主要包括以下步骤：（1）整合并标准化多源水库位置数据，对全球公共数据库、国家/区域数据集和OSM数据进行清洗、地理配准、人工检查和高分辨率卫星影像交叉验证，并剔除110个受调控天然湖泊；（2）基于GSW和GLAD水体发生数据提取1984–2020年历史最大淹没范围，对小型水库使用2019–2021年Sentinel-2 10 m水频率数据精细化边界，并对狭长河谷型水库结合更高分辨率影像和人工判读修正；（3）开展拓扑校正和空间一致性检查，形成全球水库边界数据库；（4）汇集库容和名称等属性信息，对缺失库容通过平均水深机器学习模型估算（库容=平均水深×面积）；（5）对实测淤积样本进行质量控制，剔除1930年前调查记录和调查间隔短于5年的记录，进行空间匹配并对重复记录保留调查间隔最长者；（6）构建水库平均水深和年淤积率预测模型，使用气候、地貌、土壤、水文、土地覆盖、人类活动和水库形态等变量，并分别在水库区、1 km邻域区和上游汇水区提取统计量；（7）比较CNN、SVM、XGBoost及多种树模型，采用70%训练/30%测试、五折交叉验证和网格搜索优化，最终选择XGBoost模型。", "ds_quality": "
数据质量控制贯穿位置整合、边界提取、属性汇编和模型估算全过程。水库位置记录经过系统清洗、人工检查和高分辨率影像交叉验证，以去除错误或错位水库；边界提取后开展拓扑校正和空间一致性检查。最终数据集包含555,960座全球水库，总面积469,639.9 km²。库容数据方面，85,843座水库具有核验库容信息，合计6,851.7 km³，代表全球净库容的90.4%。泥沙样本方面，原始调查资料经过调查年代、调查间隔、空间匹配和重复记录筛选后，最终保留6,133个与水库多边形匹配的实测淤积记录；样本覆盖近60个国家，水库面积范围为0.001–5060 km²，年淤积率5–95分位范围为0.29%–1.65%。模型质量采用MAE、RMSE和一倍标准差误差评估，并通过五折交叉验证、测试集评估、变量重要性和训练样本策略分析验证模型稳健性与空间可迁移性。", "ds_acq_start_time": "1930-01-01 00:00:00", "ds_acq_end_time": "2021-12-31 00:00:00", "ds_acq_place": "全球", "ds_acq_lon_east": 180.0, "ds_acq_lat_south": -90.0, "ds_acq_lon_west": -180.0, "ds_acq_lat_north": 90.0, "ds_acq_alt_low": null, "ds_acq_alt_high": null, "ds_share_type": "login-access", "ds_total_size": 310980397, "ds_files_count": 0, "ds_format": "*.tif", "ds_space_res": "30m", "ds_time_res": "年", "ds_coordinate": "WGS84", "ds_projection": "", "ds_thumbnail": "3478e129-dea1-45b6-b98d-e7001f7bec22.png", "ds_thumb_from": 0, "ds_ref_way": "", "paper_ref_way": "", "ds_ref_instruction": "", "ds_from_station": null, "organization_id": "52b7b79b-860c-49a5-9083-9a70cf8bed5a", "ds_serv_man": null, "ds_serv_phone": null, "ds_serv_mail": null, "doi_value": "", "subject_codes": [ "170.55" ], "quality_level": 0, "publish_time": "2026-06-12 18:08:09", "last_updated": "2026-06-12 18:09:32", "protected": false, "protected_to": null, "lang": "zh", "cstr": "11738.11.ncdc.hydrology.db7452.2026", "i18n": { "en": { "title": "Global REservoir Inventory and Reservoir Sedimentation Dataset (GREI_v1 and GREI_Sed_v1)", "ds_format": "*.tif", "ds_source": "
The package contains GREI_v1.gdb (global reservoir inventory and attributes), GREI_Sed_v1.gdb (global reservoir sedimentation estimates and associated attributes), Sedimentation_sample_sources.xlsx (sources of the 6,133 field-surveyed sedimentation samples), GREI_data_sources.xlsx (summary of input datasets), and GREI_data_dictionary.xlsx (variable names, data types, units and definitions). Reservoir locations were compiled from global databases including GRanD, FHReD, GOODD, GeoDAR, GREI-p2k and the Geo-referenced Database on Dams, regional or national inventories from 16 countries including China, the USA and India, and OpenStreetMap. Reservoir boundaries were delineated mainly from Global Surface Water (GSW) and GLAD long-term water occurrence data (30 m Landsat, 1984–2020), with additional refinement for small reservoirs (<0.01 km²) using 10 m Sentinel-2 water-frequency data for 2019–2021. Sedimentation samples came from national field surveys in China, the US Reservoir Sedimentation Survey Information System-II, India Central Water Commission records, and more than 100 Web of Science publications.", "ds_quality": "
Quality control was implemented throughout reservoir-location integration, boundary delineation, attribute compilation and model prediction. Reservoir locations were cleaned, manually inspected and cross-validated using high-resolution satellite imagery to remove false or mislocated records. Topology correction and spatial-consistency checks were applied after boundary extraction. The final dataset includes 555,960 reservoirs with a total surface area of 469,639.9 km². For storage capacity, 85,843 reservoirs have verified records totaling 6,851.7 km³, representing 90.4% of global net capacity. For sedimentation, field-surveyed records were screened by survey year, survey interval, spatial matching and duplicate handling, resulting in 6,133 records matched to reservoir polygons. The samples span nearly 60 countries, with reservoir areas from 0.001 to 5060 km² and annual sedimentation rates from 0.29% to 1.65% within the 5th–95th percentiles. Model performance was evaluated using MAE, RMSE and one-sigma error, and model robustness and transferability were assessed through fivefold cross-validation, test-set evaluation, variable-importance analysis and training-sample selection experiments.", "ds_ref_way": "", "ds_abstract": "
This dataset includes the Global REservoir Inventory (GREI_v1) and the global reservoir sedimentation database (GREI_Sed_v1). GREI_v1 records the spatial locations, boundaries and attributes of 555,960 reservoirs worldwide larger than 0.001 km², including reservoir name, area, location, storage capacity and mean depth. The final inventory covers 469,639.9 km² of reservoir surface area; 85,843 reservoirs have verified storage-capacity records totaling 6,851.7 km³. GREI_Sed_v1 estimates annual sedimentation rates and storage loss using 6,133 field-surveyed sedimentation samples and machine-learning models. The dataset improves the representation of small and newly dammed reservoirs and supports research and management in water resources, hydrology, sediment transport, climate adaptation and sustainability planning.\r\n
For detailed data information, refer to: Liu, K., Fan, C., Song, C. et al. Global patterns of reservoir sedimentation and overlooked risks in small reservoirs. Nature Sustainability (2026). https://doi.org/10.1038/s41893-026-01859-y", "ds_time_res": "", "ds_acq_place": "global", "ds_space_res": "", "ds_projection": "", "ds_process_way": "
The dataset was produced through the following steps: (1) integration and standardization of multi-source reservoir locations, including cleaning, geolocation, manual inspection and high-resolution satellite-image cross-validation, while excluding 110 regulated natural lakes; (2) delineation of the historically maximum inundation extent for 1984–2020 using GSW and GLAD water-occurrence data, refinement of small-reservoir boundaries with 10 m Sentinel-2 water-frequency data for 2019–2021, and manual correction for narrow steep-valley or river-like reservoirs; (3) topology correction and spatial-consistency checks; (4) compilation of storage capacity, and name attributes, with missing storage capacity estimated using a mean-depth machine-learning model (storage capacity = mean depth × area); (5) quality control of field-surveyed sedimentation samples by excluding pre-1930 records and survey intervals shorter than 5 years, geo-matching samples to GREI polygons, and retaining the longest survey interval for duplicates; and (6) development of predictive models for mean depth and annual sedimentation rate using climatic, geomorphological, soil, hydrological, land-cover, anthropogenic and reservoir-morphological variables summarized in reservoir, 1 km vicinity and upstream catchment zones. CNN, SVM, XGBoost and tree-based models were compared using a 70/30 train-test split, fivefold cross-validation and grid-search hyperparameter optimization, with XGBoost selected as the final model.", "ds_ref_instruction": "" } }, "submit_center_id": "ncdc", "data_level": 0, "recommendation_value": 0, "license_type": "https://creativecommons.org/licenses/by/4.0/", "doi_reg_from": "reg_local", "cstr_reg_from": "reg_local", "doi_not_reg_reason": null, "cstr_not_reg_reason": null, "is_paper_in_submitting": false, "belong_to_nieer": false, "ds_topic_tags": [ "水库", "泥沙淤积", "库容损失", "机器学习", "遥感" ], "ds_subject_tags": [ "水文学" ], "ds_class_tags": [], "ds_locus_tags": [ "全球水库" ], "ds_time_tags": [ 1930, 1931, 1932, 1933, 1934, 1935, 1936, 1937, 1938, 1939, 1940, 1941, 1942, 1943, 1944, 1945, 1946, 1947, 1948, 1949, 1950, 1951, 1952, 1953, 1954, 1955, 1956, 1957, 1958, 1959, 1960, 1961, 1962, 1963, 1964, 1965, 1966, 1967, 1968, 1969, 1970, 1971, 1972, 1973, 1974, 1975, 1976, 1977, 1978, 1979, 1980, 1981, 1982, 1983, 1984, 1985, 1986, 1987, 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021 ], "ds_contributors": [ { "true_name": "宋春桥", "email": "cqsong@niglas.ac.cn", "work_for": "中国科学院南京地理与湖泊研究所水安全湖泊与流域科学国家重点实验室", "country": "中国" }, { "true_name": "刘凯", "email": "kliu@niglas.ac.cn", "work_for": "中国科学院南京地理与湖泊研究所", "country": "中国" }, { "true_name": "范晨雨", "email": "fanchenyu@niglas.ac.cn", "work_for": "中国科学院南京地理与湖泊研究所", "country": "中国" } ], "ds_meta_authors": [ { "true_name": "宋春桥", "email": "cqsong@niglas.ac.cn", "work_for": "中国科学院南京地理与湖泊研究所水安全湖泊与流域科学国家重点实验室", "country": "中国" }, { "true_name": "刘凯", "email": "kliu@niglas.ac.cn", "work_for": "中国科学院南京地理与湖泊研究所", "country": "中国" }, { "true_name": "范晨雨", "email": "fanchenyu@niglas.ac.cn", "work_for": "中国科学院南京地理与湖泊研究所", "country": "中国" } ], "ds_managers": [ { "true_name": "宋春桥", "email": "cqsong@niglas.ac.cn", "work_for": "中国科学院南京地理与湖泊研究所水安全湖泊与流域科学国家重点实验室", "country": "中国" } ], "category": "水文" }