多年冻土的功能与脆弱性在很大程度上取决于其近地表的地下冰含量。然而，目前北半球仍缺乏高质量、网格化的地下冰分布图。本研究首次构建了整个北半球空间分辨率为1 km的多年冻土上限以下5米深度的体积含冰量数据集。该数据集整合了大量地下冰观测数据（共1,178个）和多源地理空间数据，包括古气候、遥感数据以及地质地貌单元，并结合多种机器学习模型采用Copula嵌入式贝叶斯模型平均（COP-BMA）方法实现。验证结果表明模型误差较低（R² = 0.86，RMSE = 7.08% VIC，bias = 0.02% VIC），而基于95%预测区间（PI）计算的不确定性为16.08 ± 3.55% VIC。结果显示，北半球近地表多年冻土的总冰储量约为 54,600 km³ （47,800–62,300 km³），约为国际多年冻土协会（IPA）地下冰图的两倍。这种差异主要归因于制图技术的进步、更丰富的观测数据整合以及空间分辨率的显著提升。高含冰量区（>80%）主要分布在低洼平原、湿地和沼泽地带；而青藏高原、蒙古高原等山地地区的VIC较低，通常在 20%–40% 之间。新生成的地下冰图在空间格局上与以往成果基本一致，但空间细节得到了显著提升。该高分辨率地图可作为监测多年冻土变化及评估其在气候、水文、生态系统和基础设施等方面提供数据支持。
", "ds_source": "该数据集整编了涵盖北美、欧亚大陆及青藏高原等主要多年冻土区共1178个钻孔的地下冰调查数据，这些数据主要来自于已发表的文献和地质勘察报告，代表了多年冻土上限以下5米的体积含冰量。预测因子系统性地考虑了影响地下冰形成与保存的关键环境要素，包括现代气候与古气候、遥感变量、地形因子以及地质与地貌单元信息等。该数据集以Geotiff格式存储，文件命名为：T5m_VIC.tif，代表了的是多年冻土上限以下5米深度的体积含冰量。
", "ds_process_way": "该数据集整合了多源地理空间数据，包括古气候与现代气候变量、遥感数据、地形因子以及地质地貌单元信息，并采用 Copula 嵌入式贝叶斯模型平均（COP-BMA） 方法，对四种机器学习模型（支持向量回归、广义加性回归、随机森林和 轻量级梯度提升机）的预测结果进行集成，以提升模拟结果的稳定性与可靠性。
", "ds_quality": "验证结果表明模型误差较低（R² = 0.86，RMSE = 7.08% VIC，bias = 0.02% VIC），而基于95%预测区间（PI）计算的不确定性为16.08 ± 3.55% VIC。
", "ds_acq_start_time": null, "ds_acq_end_time": null, "ds_acq_place": "北半球", "ds_acq_lon_east": 180.0, "ds_acq_lat_south": 25.0, "ds_acq_lon_west": 180.0, "ds_acq_lat_north": 84.0, "ds_acq_alt_low": null, "ds_acq_alt_high": null, "ds_share_type": "open-access", "ds_total_size": 1656308457, "ds_files_count": 5, "ds_format": "*.tif", "ds_space_res": "1000", "ds_time_res": "", "ds_coordinate": "WGS84", "ds_projection": "", "ds_thumbnail": "5c51614f-2d69-4285-bf73-67566c367f68.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": "李红星", "ds_serv_phone": "0931-4967592", "ds_serv_mail": "ncdc@lzb.ac.cn", "doi_value": "", "subject_codes": [ "170.4510" ], "quality_level": 3, "publish_time": "2026-02-05 16:50:19", "last_updated": "2026-05-11 16:26:41", "protected": false, "protected_to": null, "lang": "zh", "cstr": "11738.11.NCDC.NIEER.DB7052.2026", "i18n": { "en": { "title": "Near-surface ground ice dataset for the Northern Hemisphere", "ds_format": "*.tif", "ds_source": "This dataset compiles ground-ice survey data from 1,178 boreholes across the major permafrost regions of North America, Eurasia, and the Qinghai–Xizang Plateau. These records are primarily derived from published literature and geological investigation reports, and represent volumetric ice content within the 5 m interval below the permafrost table. The predictors systematically incorporates key environmental controls on ground-ice formation and preservation, including modern and paleoclimate variables, remote-sensing indicators, topographic factors, and geological and geomorphological information. The dataset is stored in GeoTIFF format, and the file is named T5m_VIC.tif, which represents the volumetric ice content within 5 m below the permafrost table.", "ds_quality": "
The validation results indicate relatively low model errors (R² = 0.86, RMSE = 7.08% VIC, bias = 0.02% VIC), while the uncertainty derived from the 95% prediction interval (PI) is 16.08 ± 3.55% VIC.", "ds_ref_way": "", "ds_abstract": "
The functioning and vulnerability of permafrost are largely determined by near-surface ground ice content. However, high-quality, grid-based ground ice maps for the Northern Hemisphere are currently unavailable. This study presents the first 1-km resolution grid-based ground ice map within 5 m below the permafrost table across the Northern Hemisphere. The map integrates an unprecedentedly amount (1,178 boreholes) of field measurement for volumetric ice content (VIC) and multisource geospatial data, especially paleoclimate, remote sensing data, surficial geology units, using Copula-Embedded Bayesian Model Averaging (COP-BMA) techniques with multiple machine learning models and 200 ensemble simulations. The validation indicates relatively low errors (R2=0.86, RMSE=7.08%VIC, bias=0.02%VIC), while the uncertainty, represented by the 95% prediction interval (PI), is 16.08% ± 3.55%VIC. The map indicates that the total ice storage of near‐surface permafrost across the Northern Hemisphere is approximately 54,600 km3 (47,800–62,300 km³), about twice the value from the International Permafrost Association map. This difference may be due to, but is not limited to, advancements in mapping techniques, the integration of additional measurement data, and improved spatial resolution. High VIC (>80%) is predominantly concentrated in low-lying plains, wetlands, and marshes. In contrast, mountainous regions, including the Qinghai-Xizang Plateau and Mongolian Plateau, exhibit lower VIC, typically ranging from 20% to 40%. The new ground ice map exhibits a spatial pattern that is largely consistent with previous maps while providing enhanced spatial detail. This high-resolution map serves as a benchmark for tracing permafrost changes and assessing impacts on climate, hydrology, ecosystems, and infrastructure in permafrost regions.", "ds_time_res": "", "ds_acq_place": "Northern Hemisphere", "ds_space_res": "", "ds_projection": "", "ds_process_way": "
This dataset integrates multi-source geospatial data, including paleoclimate and modern climate variables, remote-sensing data, topographic factors, and geological–geomorphological information. It further applies a Copula-embedded Bayesian model averaging (COP-BMA) approach to ensemble the predictions from four machine-learning models—support vector regression, generalized additive regression, random forest, and Light Gradient Boosting Machine (LightGBM)—thereby improving the stability and reliability of the simulation results.", "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, "ds_topic_tags": [ "冰冻圈", "地下冰", "青藏高原", "北极", "机器学习" ], "ds_subject_tags": [ "自然地理学" ], "ds_class_tags": [], "ds_locus_tags": [ "北半球" ], "ds_time_tags": [], "ds_contributors": [ { "true_name": "王冰泉", "email": "wangbingquan@nieer.ac.cn", "work_for": "中国科学院西北生态环境资源研究院", "country": "中国" }, { "true_name": "冉有华", "email": "ranyh@lzb.ac.cn", "work_for": "中国科学院西北生态环境资源研究院", "country": "中国" }, { "true_name": "李新", "email": "xinli@itpcas.ac.cn", "work_for": "中国科学院青藏高原研究所", "country": "中国" } ], "ds_meta_authors": [ { "true_name": "王冰泉", "email": "wangbingquan@nieer.ac.cn", "work_for": "中国科学院西北生态环境资源研究院", "country": "中国" } ], "ds_managers": [ { "true_name": "王冰泉", "email": "wangbingquan@nieer.ac.cn", "work_for": "中国科学院西北生态环境资源研究院", "country": "中国" } ], "category": "冻土" }