文献类型：期刊文献

中文题名：基于FY-3/MWRI数据的中国2010年—2021年地表冻融数据集

英文题名：Near-surface soil freeze/thaw status datasets of China based onFY-3/MWRI data from 2010 to 2021

作者：王健[1] 蒋玲梅[2] 武胜利[3] 张成[2] 陈尔学[1] 赵磊[1] 范亚雄[1] 孙凌[3] 张鹏[3]

第一作者：王健

机构：[1]中国林业科学研究院资源信息研究所林木资源高效生产全国重点实验室国家林业和草原局林业遥感与信息技术重点实验室,北京100091;[2]北京师范大学地理科学学部遥感与数字地球全国重点实验室,北京100875;[3]国家卫星气象中心,北京100081

年份：2025

卷号：29

期号：4

起止页码：844-856

中文期刊名：遥感学报

外文期刊名：NATIONAL REMOTE SENSING BULLETIN

收录：;EI(收录号：20251818359413);北大核心:【北大核心2023】；

基金：国家重点研发计划(编号:2022YFF0801302,2021YFB3900104);风云卫星应用先行计划(编号:FY-APP-2022.0305)。

语种：中文

中文关键词：地表冻融数据集;FY-3B;FY-3D;微波成像仪;动态冻融判别式算法

外文关键词：near-surface soil freeze/thaw status datasets;FY-3B;FY-3D;microwave radiation imager;dynamic near-surface soil freeze/thaw detection algorithm

分类号：TP701;P2

摘要：地表冻融是水循环和碳循环等系统中的重要变量,准确掌握地表冻融状态及其时空变化对水文过程、气候变化、生态学等研究有重要意义。现有冻融产品在地形、气候、土壤条件复杂的大区域存在表现不稳定的问题,且精度尚不能满足实际应用需求。因此,针对当前冻融产品存在的问题,为获取高精度中国的地表冻融数据,本研究利用再定标的风云3B(FY-3B)和风云3D(FY-3D)微波成像仪数据,以利用结合边缘检测算法和冻融判别式算法发展得到的动态冻融判别式算法为主算法,基于季节阈值算法为辅助算法制备中国地区2010年—2021年的地表冻融状态数据集,并利用分布在青藏高原、东北根河和华北塞罕坝地区的地面观测站点实测5 cm土壤温度数据验证此冻融数据集的整体精度。数据集精度验证结果显示本研究生产的地表冻融数据集在不同季节以及不同气候区精度表现稳定,该冻融数据集的整体精度在86%以上。在对此数据集进行充分验证的基础上,基于2011年—2020年的10年中国地表冻融数据进行地表冻融的时空变化分析,探究地表冻融变化与植被净初级生产力NPP(Net Primary Productivity)和植被总初级生产力GPP(Gross Primary Productivity)之间的相关性。结果发现:植被NPP和GPP与地表年融化首日和年冻结天数呈负相关,决定系数在0.52—0.72;地表融化开始日期越早,年冻结天数越少,植被年NPP和GPP越高,表明本数据集在评估植被生态系统碳储量及气候变化方面具有一定的潜力。此外,本研究构建的数据集还可为大区域土壤侵蚀、气候变化、水文过程等研究提供高精度的地表冻融状态数据。
Near-surface soil freeze/thaw(F/T)state is an important variable in water cycle and carbon cycle systems.Accurately obtaining the F/T state of near-surface soil and its spatial and temporal changes is important for research on hydrological processes,climate change,and ecology.The main existing F/T products based on passive microwave remote sensing data are unstable in large scale with relatively complex topography,climate,and soil conditions,and the accuracy has yet to meet the requirements of applications.The microwave radiation imager(MWRI)carried on China’s FY-3 satellite can acquire passive microwave remote sensing data and is currently less used in near-surface soil F/T monitoring.To address the problems of the existing near-surface soil F/T products,this study presented the nearsurface soil F/T dataset of China from 2010 to 2021 based on the FY-3/MWRI data.Method The algorithms used to obtain the dataset consist of a primary and a secondary algorithm,i.e.,the dynamic near-surface soil F/T detection algorithm and the seasonal threshold algorithm,respectively.The dynamic F/T detection algorithm is developed based on the union of soil F/T discriminant algorithm and edge detection algorithm and performs stably at large scales.To avoid evident F/T misclassification,corrected ERA5-Land temperature data were first used to identify areas that are not subject to F/T cycles before generating the near-surface soil F/T dataset.To reduce the effect of precipitation and water bodies on the accuracy of F/T dataset,precipitation is labeled using GPM precipitation data,and water bodies are labeled using land cover data(GlobeLand30-2010).Finally,the daily near-surface FY-3B(2010-2019)and FY-3D(2017-2021)F/T datasets consisting of daytime(ascending orbit)and nighttime(descending orbit)are presented.The in situ 5 cm soil temperature data obtained from the Qinghai-Tibetan Plateau,the Genhe watershed in Northeastern China,and the Saihanba area in Northern China were used to evaluate the FY-3B and FY-3D F/T datasets.The accuracy of the near-surface soil F/T dataset presented in this study is stable across seasons and climate zones and performs best when compared with the other existing passive microwave remote sensing F/T products.The overall accuracy of the presented F/T dataset is more than 86%.By analyzing the spatial and temporal variations of near-surface soil F/T from 2011 to 2020 based on the presented dataset,we found that the annual thaw onset was delaying,the annual frozen onset was advancing,and the annual frozen days was increasing during the 10-year period over the Qinghai-Tibetan Plateau,whereas no considerable change was observed over other regions.The vegetation Net Primary Productivity(NPP)and Gross Primary Productivity(GPP)were negatively correlated with the land surface annual thaw onset date and annual frozen days,with the coefficient of determination ranging from 0.52 to 0.72.The earlier the date of land surface thawing and the fewer the annual frozen days,the higher the annual NPP/GPP.These analyses demonstrated the potential application of this presented F/T dataset in studies of climate change,vegetation biomass,and vegetation carbon stocks.The dataset is stored in H5 file format and can be downloaded at DOI:10.11888/Crvos.tpdc.300445.