三江源国家公园是青藏高原生态屏障的核心单元，准确评估其土壤碳氮特征是区域生态功能认知和分区管理的重要基础。基于54个样点调查数据，结合高程、坡度、坡向及2000-2018年的年均气温、降水、归一化植被指数等生态因子，采用增强回归树模型研究了三江源国家公园表层（0-30 cm）土壤有机碳（SOC）、全氮（TN）密度的空间格局和等级区划及储量特征。结果表明三江源国家公园SOC密度和TN密度分别为（5.41±3.12） kg/m2（平均值±标准差，下同）和（0.57±0.27） kg/m2，其空间变异均主要受降水和归一化植被指数影响。澜沧江源园区和黄河源园区SOC和TN密度分别为（9.39±0.89） kg/m2和（0.92±0.09） kg/m2、（8.26±2.33） kg/m2和（0.80±0.20） kg/m2，约为长江源园区的2倍。SOC和TN密度等级在澜沧江源园区呈现出中心高周围低的特征，在黄河源园区和长江源园区分别表现出从北到南和从东南到西北逐渐降低的格局。三江源国家公园SOC储量和TN储量分别为0.60 Pg和0.06 Pg，其中澜沧江源园区、黄河源园区、长江源园区的储量占比分别约为20%、20%和60%。三江源国家公园SOC储量和TN储量均主要集中在高寒草甸和高寒草原，二者累计占比约为90%，是区域生态功能的主要载体和管理规划的重点对象。研究结果可为三江源国家公园的功能评估和分区管理提供参考依据。
The Qinghai-Tibetan Plateau sequestrates large amounts of soil carbon and plays an important role in regional ecological security barrier and global climate change. The Sanjiangyuan National Park lies on the hinterland of the Qinghai-Tibetan Plateau and has been officially established in October 2021. Accurate assessment of the density and stock of soil carbon and nitrogen is a prerequisite for scientific recognition of its ecological function and zoning management. However, such knowledge based on field surveys is currently lacking. Based on 54 field soil samples' data, combined with ecological variables such as elevation, slope, aspect, mean annual air temperature, mean annual precipitation, and mean annual normalized difference vegetation index (NDVI) from 2000 to 2018, the boosted regression trees models, one of up-to-date machine learning logarithms, were employed to explore the spatial pattern and classifications of the topsoil (0-30 cm) organic carbon (SOC) and total nitrogen (TN) density in the Sanjiangyuan National Park. The results showed that the density of SOC and TN were (5.41±3.12) kg/m2 (Mean±SD, the same below) and (0.57±0.27) kg/m2, respectively, and they spatially correlated very well. The spatial variations in the density of SOC and TN were both predominated by precipitation and NDVI, having accumulatively relative contributions of more than 70%. The density of SOC and TN spatially were related to precipitation and NDVI with saturated S-shaped responses. The density of SOC and TN in the Lantsang Park and the Huanghe Park were (9.39±0.89) kg/m2 and (0.92±0.09) kg/m2, (8.26±2.33) kg/m2 and (0.80±0.20) kg/m2, respectively, which were approximately twice than those of the Yangtze Park. The classifications on the density of SOC and TN have been conducted and the rank order in the Lantsang Park decreased gradually from the centers to the surrounding areas, while they decreased from the north to the south and from the southeast to the northwest in the Huanghe Park and the Yangtze Park, respectively. The stocks of SOC and TN in the Sanjiangyuan National Park were 0.60 Pg and 0.06 Pg, respectively, of which in the Lantsang Park, the Huanghe Park, and the Yangtze Park were approximately 20%, 20%, and 60%, respectively. The stocks of SOC and TN were mainly concentrated in alpine meadows and alpine steppes, which accounted for about 90% of the total reservoir. The alpine meadows and alpine steppes are thus key carriers of carbon sequestration capacity and should be taken into high priority considerations in the Sanjiangyuan National Park. These results can provide important references for the functional evaluation and zoning management of the Sanjiangyuan National Park.