北极陆表植被作为气候变化的灵敏指示器，其空间分布格局在北极生态、地气能-水-碳循环及全球气候变化均不可或缺，但植被类型的准确提取又充满挑战，主要源于其显著的空间异质性、复杂地表特征和极端环境条件。为应对这些挑战，本研究开发了ARC-2024环北极地表覆盖数据产品，该产品具有10米空间分辨率，涵盖多样化苔原植被类型。ARC-2024采用专门的北极分类系统，包括10个类别：低灌木(LS)、直立矮灌木(DS)、草地(GR)、小型低矮的草本植物(SF)、苔藓地衣(ML)、水域(WA)、湿地(WL)和冰雪(IS)，特别设计用于表征北极环境的生态多样性。ARC-2024基于时间序列匹配分类方法构建，通过部分物候信号匹配机制，有效突破了传统方法对完整等距时间序列的依赖，特别适用于无云观测时空分布不均的大尺度分类任务。\n多层次验证结果显示，ARC-2024总体精度达87.78%，Kappa系数为0.858，较现有产品平均提升超过16%。特别地，该产品在低矮灌木、矮灌木、草地和小型杂草等细粒度苔原植被类型识别方面表现突出，这些植被类型往往在现有地表覆盖产品中被忽略，但在ARC-2024中得到重点关注并实现了显著的精度提升。\nARC-2024在空间异质性识别方面具有显著优势，能够精细刻画灌木-草本生态过渡带等复杂区域特征。该数据产品代表了北极地表覆盖制图的重要进展，可为极地生态研究、植被格局分析和气候变化建模等提供重要数据支撑。", "ds_source": "
ARC-2024环北极土地覆盖产品基于欧洲空间局哥白尼计划提供的Sentinel-2 Level-2A光学影像数据构建，影像获取时间覆盖2024年北极生长季。该地表反射率产品具有10米空间分辨率，经过大气效应校正并包含完整的辐射测量参数。数据预处理在Google Earth Engine平台完成，包括基于QA60质量保证波段的云掩膜处理以及NDVI、NDWI、NDBI、NDSI等关键光谱指数计算。产品开发过程中集成了5个代表性土地覆盖数据集进行样本生成和精度验证，分别为ESA WorldCover（全球覆盖，总体精度74.4%）、FROM-GLC（10米全球制图，精度80.6%）、GLC_FCS30（基于Landsat时间序列，精度82.5%）、CAVM（标准化环极植被分类）和CALC-2020（10米北极专用制图，精度79.3%）。训练样本采用创新的时空一致性分析方法自动生成，通过GLC_FCS30长时间序列（1985-2022年）进行37年稳定性筛选，最终产生26,877个高置信度训练点，经500个随机样本的人工验证显示94.2%的标签一致性。独立验证体系包括POI-Sample II和PAT-Sample两个数据集：前者包含1,069个参考点（900个人工解译点和169个实地验证样本），数据源自北极植被档案库、AmeriFlux/FLUXNET监测站点和PhenoCam观测网络；后者由226个5km×5km景观斑块组成，系统性分布于不同异质性梯度上，确保了产品质量评估的全面性和客观性。", "ds_process_way": "
（1）影像预处理：基于Sentinel-2 Level-2A产品，利用QA60波段进行云掩膜处理，计算NDVI、NDWI、NDBI、NDSI等光谱指数，在Google Earth Engine平台完成标准化预处理。\n
（2）自动化样本生成：采用时空一致性分析框架，通过GLC_FCS30时间序列(1985-2022年)37年稳定性筛选，结合多产品共识验证和分层空间滤波，基于生物气候亚区和海拔梯度分层采样，生成26,877个高置信度训练样本，人工验证准确率达94.2%。\n
（3）时间序列匹配分类：开发变长时域匹配策略，通过部分物候信号匹配机制突破传统等距时间序列限制，最大化利用生长季内所有可用无云观测数据。重点提取物候转换期特征信号，增强光谱相似苔原植被类型的可分性，实现自动化分类处理。\n
（4）多尺度验证：建立点集验证(POI-Sample II，1,069个参考点)、斑块验证(PAT-Sample，226个5km×5km景观斑块)、区域一致性分析和视觉评估的四维验证体系，采用双标签系统确保评估客观性。", "ds_quality": "
采用总体精度（OA）、Kappa系数、用户精度（UA）和生产者精度（PA）等标准指标，通过点基验证、斑块基验证、区域一致性分析和视觉一致性评估四种互补验证方法对ARC-2024土地覆盖分类产品进行全面精度评估。验证数据集包括POI-Sample II（1,069个参考点：900个人工解译点+169个实地验证点）和PAT-Sample（226个5km×5km景观斑块），数据源涵盖北极植被档案库、AmeriFlux/FLUXNET站点和PhenoCam网络，采用双标签系统确保与不同分类产品比较的公平性。\n多尺度验证结果表明，ARC-2024在环极地北极地区达到87.78%的总体精度和0.858的Kappa系数。斑块尺度验证显示良好的空间表现，在226个景观斑块中展现出较强的空间一致性。网格化精度分析表明，86.72%的100×100km网格单元总体精度超过80%，且在不同生物气候亚区（A-E）均保持相对稳定的性能。区域一致性分析显示，ARC-2024在稀疏森林（SF）、矮灌木（DS）等北极特有植被类型的识别上表现出合理的面积估算结果。视觉评估通过9个代表性地块验证了产品在复杂地貌和植被过渡带的空间结构表达能力。\n与现有产品的比较分析显示，ARC-2024在多个验证维度展现出良好表现，特别是在针对北极环境设计的分类类别方面。不同产品由于设计目标和方法学差异呈现各自特点，ARC-2024的北极专门化设计使其在该区域的植被制图中具有一定优势。因此，ARC-2024可作为北极土地覆盖制图、生态系统监测和气候变化研究的可靠数据资源。", "ds_acq_start_time": "2024-05-01 00:00:00", "ds_acq_end_time": "2024-10-31 00:00:00", "ds_acq_place": "泛北极区域", "ds_acq_lon_east": 180.0, "ds_acq_lat_south": 55.79, "ds_acq_lon_west": -180.0, "ds_acq_lat_north": 84.08999999999999, "ds_acq_alt_low": null, "ds_acq_alt_high": null, "ds_share_type": "login-access", "ds_total_size": 15132975833, "ds_files_count": 9, "ds_format": "*.tif", "ds_space_res": "10m", "ds_time_res": "年", "ds_coordinate": "WGS84", "ds_projection": "Albers Equal Area Conic Projection System", "ds_thumbnail": "bd8018b1-ed52-48c3-9e7d-59a1799cb6d0.png", "ds_thumb_from": 2, "ds_ref_way": "", "paper_ref_way": "", "ds_ref_instruction": "", "ds_from_station": null, "organization_id": "53943799-d453-4bf2-a141-56c205c1355b", "ds_serv_man": "李红星", "ds_serv_phone": "0931-4967592", "ds_serv_mail": "ncdc@lzb.ac.cn", "doi_value": "", "subject_codes": [ "170.45" ], "quality_level": 3, "publish_time": "2026-02-12 19:50:14", "last_updated": "2026-02-13 10:28:26", "protected": false, "protected_to": null, "lang": "zh", "cstr": "11738.11.NCDC.ARCTIC-CHANGE.DB7128.2026", "i18n": { "en": { "title": "ARC-2024: 10m Resolution Land Cover Map of Circumpolar Arctic in 2024", "ds_format": "*.tif", "ds_source": "
The ARC-2024 Circumpolar Arctic land cover product is constructed based on Sentinel-2 Level-2A optical imagery data provided by the European Space Agency's Copernicus Programme, with image acquisition covering the 2024 Arctic growing season. This surface reflectance product features 10-meter spatial resolution, atmospheric correction, and complete radiometric parameters. Data preprocessing was completed on the Google Earth Engine platform, including cloud masking based on the QA60 quality assurance band and calculation of key spectral indices such as NDVI, NDWI, NDBI, and NDSI. The product development process integrated 5 representative land cover datasets for sample generation and accuracy validation: ESA WorldCover (global coverage, overall accuracy 74.4%), FROM-GLC (10m global mapping, accuracy 80.6%), GLC_FCS30 (Landsat time series-based, accuracy 82.5%), CAVM (Circumpolar Arctic Vegetation Map), and CALC-2020 (10m Arctic-specific mapping, accuracy 79.3%). Training samples were automatically generated using an innovative spatiotemporal consistency analysis method, with 37-year stability screening through GLC_FCS30 long-term time series (1985-2022), ultimately producing 26,877 high-confidence training points. Manual validation of 500 random samples showed 94.2% label consistency. The independent validation system includes two datasets: POI-Sample II and PAT-Sample. The former contains 1,069 reference points (900 manually interpreted points and 169 field validation samples), sourced from the Arctic Vegetation Archive, AmeriFlux/FLUXNET monitoring sites, and PhenoCam observation network. The latter consists of 226 landscape patches of 5km×5km, systematically distributed across different heterogeneity gradients, ensuring comprehensive and objective product quality assessment.", "ds_quality": "
Comprehensive accuracy assessment of the ARC-2024 land cover classification product was conducted using standard metrics including Overall Accuracy (OA), Kappa coefficient, User's Accuracy (UA), and Producer's Accuracy (PA), through four complementary validation methods: point-based validation, patch-based validation, regional consistency analysis, and visual consistency assessment. The validation datasets include POI-Sample II (1,069 reference points: 900 manually interpreted points + 169 field validation points) and PAT-Sample (226 landscape patches of 5km×5km), with data sources encompassing the Arctic Vegetation Archive, AmeriFlux/FLUXNET sites, and PhenoCam network. A dual-label system was adopted to ensure fairness in comparisons with different classification products.\nMulti-scale validation results demonstrate that ARC-2024 achieves an overall accuracy of 87.78% and a Kappa coefficient of 0.858 across the circumpolar Arctic region. Patch-scale validation shows good spatial performance, exhibiting strong spatial consistency across 226 landscape patches. Gridded accuracy analysis indicates that 86.72% of 100×100km grid cells achieve overall accuracy exceeding 80%, maintaining relatively stable performance across different bioclimatic subzones (A-E). Regional consistency analysis shows that ARC-2024 demonstrates reasonable area estimation results for Arctic-specific vegetation types such as Sparse Forest (SF) and Dwarf Shrubs (DS). Visual assessment through 9 representative plots validated the product's capability to express spatial structures in complex terrain and vegetation transition zones.\nComparative analysis with existing products shows that ARC-2024 exhibits good performance across multiple validation dimensions, particularly in classification categories designed specifically for Arctic environments. Different products present distinct characteristics due to variations in design objectives and methodologies, with ARC-2024's Arctic-specialized design providing certain advantages in vegetation mapping for this region. Therefore, ARC-2024 can serve as a reliable data resource for Arctic land cover mapping, ecosystem monitoring, and climate change research.", "ds_ref_way": "", "ds_abstract": "
Arctic land cover mapping faces significant challenges, primarily stemming from pronounced spatial heterogeneity, complex surface characteristics, and extreme environmental conditions. To address these challenges, this study developed the ARC-2024 Circumpolar Arctic land cover data product, which features 10-meter spatial resolution and encompasses diverse tundra vegetation types. ARC-2024 employs a specialized Arctic classification system comprising 10 categories: Low-shrub (LS), Dwarf-shrub (DS), Grass (GR), Small, low-growing forb (SF), Moss and Lichen (ML), Water (WA), Wetlands (WL), and Ice and Snow (IS), specifically designed to characterize the ecological diversity of Arctic environments. ARC-2024 is constructed based on a time-series matching classification approach that, through a partial phenological signal matching mechanism, effectively overcomes the traditional dependence on complete and equidistant time series, making it particularly suitable for large-scale classification tasks where cloud-free observations exhibit spatiotemporally uneven distribution.\nValidation results demonstrate that ARC-2024 exhibits exceptional classification performance, achieving an overall accuracy of 87.78% and a Kappa coefficient of 0.858, representing an average improvement of over 16% compared to existing products. Particularly noteworthy is the product's outstanding performance in identifying fine-grained tundra vegetation types, including low shrubs, dwarf shrubs, grasslands, and small forbs. These vegetation types are often overlooked in existing land cover products but receive focused attention in ARC-2024, achieving significant accuracy improvements.\nARC-2024 demonstrates significant advantages in spatial heterogeneity identification, enabling fine-scale characterization of complex regional features such as shrub-herbaceous ecotones. This data product represents an important advancement in Arctic land cover mapping and can provide crucial data support for polar ecological research, vegetation pattern analysis, and climate change modeling.", "ds_time_res": "年", "ds_acq_place": "Circumpolar Arctic", "ds_space_res": "10m", "ds_projection": "", "ds_process_way": "
(1) Image Preprocessing: Based on Sentinel-2 Level-2A products, cloud masking was performed using the QA60 band, and spectral indices including NDVI, NDWI, NDBI, and NDSI were calculated. Standardized preprocessing was completed on the Google Earth Engine platform.\n
(2) Automated Sample Generation: Employing a spatiotemporal consistency analysis framework, 37-year stability screening was conducted through GLC_FCS30 time series (1985-2022), combined with multi-product consensus validation and stratified spatial filtering. Stratified sampling based on bioclimatic subzones and elevation gradients generated 26,877 high-confidence training samples, with manual validation accuracy reaching 94.2%.\n
(3) Time-Series Matching Classification: A variable-length temporal matching strategy was developed, breaking through traditional equidistant time-series limitations through a partial phenological signal matching mechanism, maximizing the utilization of all available cloud-free observations during the growing season. Focus was placed on extracting characteristic signals from phenological transition periods to enhance the separability of spectrally similar tundra vegetation types, achieving automated classification processing.\n
(4) Multi-scale Validation: A four-dimensional validation system was established, comprising point-based validation (POI-Sample II, 1,069 reference points), patch validation (PAT-Sample, 226 landscape patches of 5km×5km), regional consistency analysis, and visual assessment. A dual-label system was adopted to ensure assessment objectivity.", "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": [ "泛北极", "土地覆盖制图", "时间序列分类", "苔原植被", "Sentinel-2" ], "ds_subject_tags": [ "地理学" ], "ds_class_tags": [], "ds_locus_tags": [ "泛北极" ], "ds_time_tags": [ 2024 ], "ds_contributors": [ { "true_name": "帅艳民", "email": "shuaiym@zjnu.edu.cn", "work_for": "浙江师范大学", "country": "中国" }, { "true_name": "曲歌", "email": "471920609@stu.lntu.edu.cn", "work_for": "辽宁工程技术大学", "country": "中国" }, { "true_name": "霍思慧", "email": "Huosihui@zjnu.edu.cn", "work_for": "浙江师范大学", "country": "中国" }, { "true_name": "马现伟", "email": "maxianwei_lntu@126.com", "work_for": "辽宁工程技术大学", "country": "中国" } ], "ds_meta_authors": [ { "true_name": "曲歌", "email": "471920609@stu.lntu.edu.cn", "work_for": "辽宁工程技术大学", "country": "中国" }, { "true_name": "帅艳民", "email": "shuaiym@zjnu.edu.cn", "work_for": "浙江师范大学", "country": "中国" } ], "ds_managers": [ { "true_name": "帅艳民", "email": "shuaiym@zjnu.edu.cn", "work_for": "浙江师范大学", "country": "中国" } ], "category": "基础地理" }