基于青藏高原30 m / 8 d NDVI时间序列数据集（2000–2020），结合“青藏工程走廊热融滑塌遥感识别数据”和高分辨率 PlanetScope 影像，应用 LandTrendr 与分段线性函数拟合（PWLF）方法对 NDVI 突变过程进行识别与解析，揭示青藏高原楚玛尔河附近区域2010–2020年间热融滑塌灾变发生的关键时间节点、扰动幅度与持续效应，并利用多源遥感图像变化验证其地貌扰动机制。本数据集适用于多年冻土区地表过程监测、滑塌灾害时序分析及区域生态变化研究。", "ds_source": "
NDVI 数据来自《青藏高原时空分辨率（30 m、8 d）NDVI 时间序列数据集（2000–2020）》，DOI：https://doi.org/10.11888/Terre.tpdc.272681\n
热融滑塌滑坡清单数据来自青藏高原国家数据中心《青藏高原多年冻土区热融滑塌分布数据（2018–2020）》，DOI：https://doi.org/10.11888/Cryos.tpdc.300333\n
独立验证数据来自《青藏工程走廊沿线热融滑塌综合调查（2019）》，DOI：https://doi.org/10.11888/Cryos.tpdc.272672", "ds_process_way": "
（1）LandTrendr（Landsat-based Detection of Trends in Disturbance and Recovery）\n算法原理：将 NDVI 时间序列划分为多段线性段，通过迭代优化识别扰动与恢复阶段拐点。\n应用目的：识别 NDVI 突变年份、下降幅度与后期恢复趋势。\n
（2）PWLF 分段线性函数拟合（Piecewise Linear Function Fitting）\n用途：对 NDVI 拐点前后的变化趋势进行精细线性分段拟合，量化扰动强度、恢复速率和稳定阶段。\n
算法原理：基于最小二乘优化求解断点位置及各分段斜率。", "ds_quality": "
为评估 LandTrendr 与 PWLF 的识别性能，本研究选取 20 个现场及影像判读确认的滑塌点与 20 个非滑塌对照点，共 40 个样本，构建混淆矩阵（Confusion Matrix）进行精度验证。\n
总体精度 Accuracy = 0.925\n
精确率 Precision = 1.000\n
召回率 Recall = 0.850\n
F1-score = 0.918\n
结果表明模型对热融滑塌扰动的识别具有高可靠性和较强泛化能力。Precision = 1.0 说明无误报；Recall = 0.85 表明多数滑塌事件能被有效识别，少量漏检与弱扰动或积雪/阴影干扰相关。", "ds_acq_start_time": "2010-01-01 00:00:00", "ds_acq_end_time": "2020-12-31 00:00:00", "ds_acq_place": "青藏高原中部楚玛尔河附近地区", "ds_acq_lon_east": 92.75, "ds_acq_lat_south": 35.016666666666666, "ds_acq_lon_west": 92.63333333333334, "ds_acq_lat_north": 33.1, "ds_acq_alt_low": 4457.0, "ds_acq_alt_high": 5477.0, "ds_share_type": "apply-access", "ds_total_size": 2874026144, "ds_files_count": 575, "ds_format": "GeoTIFF", "ds_space_res": "30米", "ds_time_res": "8天", "ds_coordinate": "WGS84", "ds_projection": "WGS 84 / UTM Zone 46N（EPSG:32646）", "ds_thumbnail": "8d68f800-fd0e-43bb-ac77-959edc0a1e43.png", "ds_thumb_from": 0, "ds_ref_way": "", "paper_ref_way": "", "ds_ref_instruction": "本数据集基于多源遥感数据融合结果，适用于多年冻土区地表扰动监测、滑坡灾害识别、生态变化分析等研究。LandTrendr 和 PWLF 方法对强烈扰动的 NDVI 信号识别效果最佳，但在积雪、阴影覆盖或沙化区域可能产生不稳定拟合，应结合高分辨率影像辅助判读。数据仅反映 NDVI 时间序列的扰动—恢复模式，不代表滑坡体物理位移过程，必要时需与 InSAR、GNSS 等实测位移数据结合使用。", "ds_from_station": null, "organization_id": "9de89acc-5714-4927-aba3-ac88067dff8a", "ds_serv_man": "康建芳", "ds_serv_phone": "0931-4967597", "ds_serv_mail": "ncdc@lzb.ac.cn", "doi_value": "", "subject_codes": [ "170.4510" ], "quality_level": 3, "publish_time": "2025-12-09 20:12:27", "last_updated": "2025-12-09 20:12:27", "protected": false, "protected_to": null, "lang": "zh", "cstr": "11738.11.NCDC.NIEER.DB7028.2025", "i18n": { "en": { "title": "Rapid Detection Dataset of Thermokarst Landslides Near the Chumaer River, Qinghai–Tibet Plateau (2010–2020)", "ds_format": "", "ds_source": "
The NDVI data comes from the \"NDVI Time Series Dataset with Spatial and Temporal Resolution (30 m, 8 d) of the Qinghai Tibet Plateau (2000-2020)\", DOI: https://doi.org/10.11888/Terre.tpdc.272681\n
The inventory data of thermal melting landslides comes from the National Data Center of the Qinghai Tibet Plateau's \"Distribution Data of Thermal Melting Landslides in Permafrost Regions of the Qinghai Tibet Plateau (2018-2020)\", DOI: https://doi.org/10.11888/Cryos.tpdc.300333\n
The independent verification data comes from the Comprehensive Investigation of Thermal Melting and Landslides along the Qinghai Tibet Engineering Corridor (2019), DOI: https://doi.org/10.11888/Cryos.tpdc.272672", "ds_quality": "
To evaluate the recognition performance of LandTrendr and PWLF, this study selected 20 landslide points confirmed by on-site and image interpretation and 20 non landslide control points, a total of 40 samples, and constructed a Confusion Matrix for accuracy verification.\n
Overall Accuracy=0.925\n
Precision=1.000\n
Recall rate=0.850\n
F1-score = 0.918\n
The results indicate that the model has high reliability and strong generalization ability in identifying thermal collapse disturbances. Precision=1.0 indicates no false positives; Recall=0.85 indicates that most landslide events can be effectively identified, and a small number of missed detections are related to weak disturbances or snow/shadow interference.", "ds_ref_way": "", "ds_abstract": "
Based on the 30 m/8 d NDVI time series dataset of the Qinghai Tibet Plateau (2000-2020), combined with the \"Remote Sensing Identification Data of Thermal Melting and Landslide in the Qinghai Tibet Engineering Corridor\" and high-resolution PlanetScope images, LandTrendr and Segmented Linear Function Fitting (PWLF) methods were applied to identify and analyze the NDVI mutation process, revealing the key time nodes, disturbance amplitudes, and sustained effects of thermal melting and landslide disasters in the vicinity of the Chumar River on the Qinghai Tibet Plateau from 2010 to 2020. Multi source remote sensing image changes were used to verify its geomorphic disturbance mechanism. This dataset is suitable for monitoring surface processes in permafrost regions, analyzing the temporal sequence of landslide disasters, and studying regional ecological changes.
", "ds_time_res": "8天", "ds_acq_place": "The area near the Chumar River in the central part of the Qinghai Tibet Plateau", "ds_space_res": "30米", "ds_projection": "", "ds_process_way": "（1）LandTrendr（Landsat-based Detection of Trends in Disturbance and Recovery）\nAlgorithm principle: Divide the NDVI time series into multiple linear segments, and identify the inflection points of disturbance and recovery stages through iterative optimization.\nApplication purpose: To identify the year of NDVI mutation, the magnitude of decline, and the trend of later recovery.\n
(2) PWLF Piecewise Linear Function Fitting\nPurpose: To perform fine linear segment fitting on the trend of NDVI changes before and after the inflection point, quantifying disturbance intensity, recovery rate, and stable stage.\n
Algorithm principle: Based on least squares optimization, solve the breakpoint position and slope of each segment.", "ds_ref_instruction": "This dataset is based on the fusion results of multi-source remote sensing data and is suitable for research on surface disturbance monitoring, landslide hazard identification, ecological change analysis, etc. in permafrost regions. The LandTrendr and PWLF methods have the best recognition performance for NDVI signals with strong disturbances, but may produce unstable fitting in areas with snow cover, shadow coverage, or desertification, and should be combined with high-resolution images for auxiliary interpretation. The data only reflects the disturbance recovery mode of NDVI time series and does not represent the physical displacement process of the landslide body. If necessary, it needs to be combined with measured displacement data such as InSAR and GNSS." } }, "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, "ds_topic_tags": [ "冻融滑坡", "速变感知", "Landtrendr", "PWLF" ], "ds_subject_tags": [ "自然地理学" ], "ds_class_tags": [], "ds_locus_tags": [ "青藏高原中部楚玛尔河附近地区" ], "ds_time_tags": [ 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020 ], "ds_contributors": [ { "true_name": "李艳", "email": "liyan1@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": "周增光", "email": "zhouzg@aircas.ac.cn", "work_for": "中国科学院空天信息创新研究院", "country": "中国" } ], "ds_meta_authors": [ { "true_name": "李艳", "email": "liyan1@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": "李艳", "email": "liyan1@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": "冰川" }