多年冻土退化的一个后果是热融湖塘的发育，而这些湖塘可能是二氧化碳（CO₂）和甲烷（CH₄）排放的热点区域。然而，由于数据分辨率的限制，小型热融湖塘（<500 m²）在区域碳排放估算中长期被排除在外。本研究结合深度学习方法与高分辨率（3 m）PlanetScope遥感影像，并辅以人工修正，构建了一个精度更高的青藏高原多年冻土区热融湖塘数据集。共识别出329,848个热融湖塘，覆盖面积约为2,893 km²，较已有湖塘识别结果多出51.4%，揭示了此前对多年冻土区热融湖塘估算的重大不足。", "ds_source": "
本数据集结合深度学习方法与高分辨率（3米）PlanetScope遥感影像，并辅以人工修正，构建了一个精度更高的青藏高原多年冻土区热融湖数据集。", "ds_process_way": "
本数据集的制作主要包括两个阶段：深度学习预测和人工修正。在深度学习阶段，采用 U-Net 模型进行热融湖塘识别。该模型具有经典的编码器—解码器结构，并通过跳跃连接融合低层次与高层次语义信息，从而提升分割精度。输入影像包含红、绿、蓝和近红外四个波段，输出为表示湖泊存在的二值分割结果。模型训练使用已有热融湖数据构建样本，覆盖青藏高原多年冻土区内不同形态和尺度的热融湖塘。在训练过程中，充分考虑山影、冰雪和云层等干扰因素，通过不断补充具有代表性的训练样本逐步优化模型，直至模型性能趋于稳定。最终共生成4452个样本，并按照7:3的比例划分训练集和测试集。\n在获得二值分割结果后，首先进行影像拼接和矢量化处理，并结合中国第二次冰川编目数据排除冰川湖。随后，根据面积阈值（<3 km²）筛选出热融湖塘。为进一步提高数据的准确性和可靠性，对提取结果的矢量多边形进行了系统的人工校正，最终构建高精度热融湖数据集。", "ds_quality": "
本数据集主要采用F1-score和IOU作为评价模型性能的核心指标。结果显示，F1-score和IOU分别达到0.9541和0.9495，表明模型在湖塘识别完整性和空间一致性方面具有较好的自动化提取能力。尽管如此，定量评价结果和目视解译均表明，在局部区域仍然存在一定的分类不确定性。为进一步提高提取结果的可靠性，在自动分类结果基础上仍需进行人工校正与优化，从而确保最终数据集的精度和质量。", "ds_acq_start_time": "2020-07-01 00:00:00", "ds_acq_end_time": "2020-08-31 00:00:00", "ds_acq_place": "青藏高原多年冻土区", "ds_acq_lon_east": 103.12888888888888, "ds_acq_lat_south": 27.96388888888889, "ds_acq_lon_west": 74.03999999999999, "ds_acq_lat_north": 39.56666666666667, "ds_acq_alt_low": 4219.0, "ds_acq_alt_high": 5047.0, "ds_share_type": "open-access", "ds_total_size": 118500436, "ds_files_count": 8, "ds_format": "*.shp", "ds_space_res": "3m", "ds_time_res": "年", "ds_coordinate": "WGS84", "ds_projection": "Albers", "ds_thumbnail": "c4fd7755-9252-48c0-bd2e-f70606a13c08.png", "ds_thumb_from": 2, "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": [ "170.45" ], "quality_level": 3, "publish_time": "2026-03-18 11:29:32", "last_updated": "2026-03-18 18:01:07", "protected": false, "protected_to": null, "lang": "zh", "cstr": "11738.11.NCDC.LZU.DB7178.2026", "i18n": { "en": { "title": "High resolution thermokarst lake dataset in the Permafrost region of the Qinghai–Tibet Plateau (2020)", "ds_format": "*.shp", "ds_source": "
This study employs a deep learning method combined with high-resolution (3 m) PlanetScope imagery complemented by manual corrections, to produce a new high-accuracy dataset of thermokarst lakes in the permafrost regions of the Qinghai-Tibet Plateau (QTP).", "ds_quality": "
The F1-score and IOU were adopted as the primary metrics for evaluating model performance in this dataset. The results show that the F1-score and IOU reached 0.9541 and 0.9495, respectively, indicating that the model achieves strong automated extraction performance in terms of lake identification completeness and spatial consistency. Nevertheless, both quantitative evaluation and visual interpretation reveal that certain classification uncertainties still exist in localized areas. To further improve the reliability of the extracted results, manual correction and refinement are required based on the automated classification outputs, thereby ensuring the overall accuracy and quality of the final dataset.", "ds_ref_way": "", "ds_abstract": "
One of the consequences of permafrost degradation has been the development of thermokarst lakes, which are potential hotspots for CO2 and CH4 emissions. However, due to data resolution issues, small thermokarst lakes (<500 m2) have traditionally been excluded from regional carbon emission estimates. This study employs a deep learning method combined with high-resolution (3 m) PlanetScope imagery complemented by manual corrections, to produce a new high-accuracy dataset of thermokarst lakes in the permafrost regions of the Qinghai-Tibet Plateau (QTP). A total of 329,848 thermokarst lakes were detected, covering an area of approximately 2,893 km2. This exceeds existing estimates of identified lakes by 51.4%, thereby uncovering a critical shortcoming in prior thermokarst lake estimates for the permafrost region of the QTP.", "ds_time_res": "年", "ds_acq_place": "Permafrost region of the Qinghai–Tibet Plateau", "ds_space_res": "3m", "ds_projection": "", "ds_process_way": "
The construction of this dataset mainly consists of two stages: deep learning–based mapping and manual refinement. In the deep learning–based mapping stage, thermokarst lakes were extracted using a U-Net model, which adopts a classic encoder-decoder architecture with skip connections to integrate low-level and high-level semantic information, thereby enhancing the network's segmentation performance. We used images from four channels: red, green, blue, and near infrared as inputs, generating binary images that indicate the presence of lakes. The existing thermokarst lake data is used to train our model, leveraging a comprehensive set of samples from the permafrost regions of the QTP that includes a variety of shapes and sizes of thermokarst lakes. Additionally, we took into account factors such as mountain shadows, ice and snow, and cloud cover that could impact the accuracy of lake extraction. During the training process, we iteratively increased the training samples with diverse features based on the feedback from the prediction results, continuing this cycle until no further improvements in training performance. A total of 4452 training samples were generated, with a training to testing ratio of 7:3.\nAfter model inference, binary output images were obtained, followed by stitching and vectorization. Glacial lakes were excluded based on China’s second glacier survey. Subsequently, we filtered thermokarst lakes based on an area threshold of less than 3 km2. To further ensure the accuracy and reliability of the dataset, the extracted vector polygons were systematically checked and manually refined, resulting in a high-precision thermokarst lake dataset.", "ds_ref_instruction": "" } }, "submit_center_id": "ncdc", "data_level": 0, "license_type": "CC 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": [ "热融湖塘", "深度学习", "高分辨率遥感影像" ], "ds_subject_tags": [ "地理学" ], "ds_class_tags": [], "ds_locus_tags": [ "青藏高原多年冻土区" ], "ds_time_tags": [ 2020 ], "ds_contributors": [ { "true_name": "彭小清", "email": "pengxq@lzu.edu.cn", "work_for": "兰州大学资源环境学院 ", "country": "中国" }, { "true_name": "罗恒星", "email": "luohx2024@lzu.edu.cn", "work_for": "兰州大学资源环境学院", "country": "中国" }, { "true_name": "田伟伟", "email": "tianwei_9510@163.com", "work_for": "兰州大学资源环境学院", "country": "中国" } ], "ds_meta_authors": [ { "true_name": "彭小清", "email": "pengxq@lzu.edu.cn", "work_for": "兰州大学资源环境学院 ", "country": "中国" }, { "true_name": "罗恒星", "email": "luohx2024@lzu.edu.cn", "work_for": "兰州大学资源环境学院", "country": "中国" }, { "true_name": "田伟伟", "email": "tianwei_9510@163.com", "work_for": "兰州大学资源环境学院", "country": "中国" } ], "ds_managers": [ { "true_name": "彭小清", "email": "pengxq@lzu.edu.cn", "work_for": "兰州大学资源环境学院 ", "country": "中国" }, { "true_name": "罗恒星", "email": "luohx2024@lzu.edu.cn", "work_for": "兰州大学资源环境学院", "country": "中国" }, { "true_name": "田伟伟", "email": "tianwei_9510@163.com", "work_for": "兰州大学资源环境学院", "country": "中国" } ], "category": "冻土" }