东北黑土区土壤侵蚀面广量大，严重危害着我国的粮食安全和生态安全。面对如此严峻的环境问题，探索泥沙输移规律，掌握土壤侵蚀的动态变化，可以为历史水土流失的准确评价提供依据，从而指导未来水土流失治理措施的合理规划和布设，使水土流失防治工作更具针对性。 泥沙来源研究对流域泥沙平衡计算、河流泥沙减控、土壤侵蚀预报模型验证和水土保持效益评价都具有非常重要的意义，对生态建设与保护由事后治理向事前预防的战略性转变具有重要的实际指导价值。泥沙来源指纹示踪法在避免涉及土壤侵蚀机理的前提下，通过指纹直接联系源区土壤和侵蚀泥沙，定量描述潜在泥沙源区对流域产沙的相对泥沙贡献比例。虽然我国研究泥沙来源指纹示踪法的时间较短，但发展非常迅速，已成功地应用到一些地区。然而，至今针对东北黑土区的泥沙来源研究甚少，还基本处于空白。 本研究以东北黑土区的典型小流域及其内部微小集水区为研究对象，以土壤剖面分布为依据将流域的泥沙源区划分为表土和底土（侵蚀沟），再按土地利用类型将表土分为耕地和非耕地（草地和林地）。利用人工混合泥沙、沟底的沉积泥沙和流域出口处水库的泥沙沉积剖面，分别做了以下研究：泥沙来源定量模型的准确度评估、不同指纹的泥沙来源示踪能力评价、定量小流域及其内部微小集水区潜在源区的泥沙贡献比例、融雪侵蚀与降雨侵蚀的泥沙来源对比分析。获得的主要结论如下： （1）Walling-Collins模型与Bayesian模型的泥沙贡献比例定量准确度较高，平均绝对误差＜8%，且两种模型的准确度无显著差异。而DFA模型的表现很不理想，平均绝对误差＞18%，很难准确地定量泥沙贡献比例，需要进一步改进，其准确度与前两种模型有显著差异（p < 0.01）。 （2）复合指纹、同位素指纹和单指纹定量泥沙贡献比例的结果大体一致，研究区表土和侵蚀沟的泥沙贡献各约占50%。尤其，137Cs单指纹的泥沙源区判别能力高达100%，定量标准误差和95%置信水平的置信区间范围最小，其泥沙贡献比例与基于复合指纹的定量结果最为接近。 （3）1976–2016年间，耕地和侵蚀沟是小流域的主要泥沙源区，对流域出口处水库分别贡献50.6%和42.9%的泥沙。耕地的泥沙贡献比例总体呈减小的趋势，而侵蚀沟的泥沙贡献比例有增加的趋势。非耕地仅贡献6.5%的泥沙，且变化趋势不明显。比较两个不同时期的泥沙贡献比例，发现1999–2016年耕地的泥沙贡献比例较1976–1998年减少了5.7%，而侵蚀沟的泥沙贡献比例增加了2.8%。耕地和侵蚀沟对水库不同位置的泥沙贡献比例有显著差异（p < 0.01），较多来源于侵蚀沟的泥沙沉积于水库入口及周围，而较多来源于耕地的泥沙沉积于水库内部及水坝附近。1976–2016年，水库的年均泥沙沉积速率约为1.2 cm yr-1（10.7 kg m-2 yr-1），1999–2016年的水库年均泥沙沉积速率较1976–1998年降低了约43.7%。 （4）1974–2004年间，微小集水区出口处水库淤积的泥沙中，耕地和侵蚀沟分别贡献47.9%和42.0%，非耕地（林地和草地）的平均泥沙贡献比例仅为10.1%。来源于耕地的泥沙贡献比例随时间呈显著减小的趋势（p < 0.01），而来源于侵蚀沟的泥沙贡献比例则呈显著增大的趋势（p < 0.01）。与小流域尺度相比，两个空间尺度三种源区的泥沙贡献比例差异不大。其中，微小集水区的非耕地泥沙贡献比例较小流域大3.6%，而耕地和侵蚀沟的泥沙贡献比例较小流域小2.7%和0.9%。根据210Pb测年法的结果，发现1980–1997年的微小集水区年均产沙量较1974–1979年减少了约60%。1998年特大洪水引起了严重的土壤侵蚀，导致1999–2004年的年均产沙量较1980–1996年增加25%，但较1974–1979年降低了约53%。 （5）降雨侵蚀与融雪侵蚀的泥沙来源有显著差异（p < 0.01）。降雨侵蚀的泥沙主要来源于耕地，其次是侵蚀沟和非耕地，而融雪侵蚀中约70%的泥沙来源于侵蚀沟，其次为耕地和非耕地。降雨侵蚀和融雪侵蚀的径流量均与耕地和侵蚀沟的泥沙贡献比例显著相关（p <0.01）。降雨侵蚀的径流量与耕地的泥沙贡献比例呈负相关关系，与侵蚀沟呈正相关关系。而融雪侵蚀的径流量与耕地泥沙贡献比例呈正相关关系，与侵蚀沟呈负相关关系。这些相关性表明，融雪侵蚀过程中侵蚀沟的泥沙贡献比例远大于耕地和非耕地的主要原因，可能不是径流对沟壁和沟底的冲刷作用所致，而是春季解冻期冻融作用引起的沟壁坍塌和沟头溯源侵蚀增大了侵蚀沟的产沙量。

The area and amount of soil and water loss is very large, and soil erosion is very serious in black soil region of northwast China. In the face of such severe environmental problems, we need explore the pattern of sediment transport and understand the dynamic changes of soil erosion to accurately evaluate the past, reasonably plan the future, and prevent soil from erosion. Research on provenence of sediment is of great significance for calculation of basin sediment balance, control of river sediment reduction, verification of soil erosion prediction model, and evaluation of soil and water conservation. It also has important practical guiding value for strategic change of Eco-build from post-control to pre-prevention. Under the premise of avoiding the mechanism of soil ersion, sediment fingerprinting quantitatively describes the relative sediment contribution of potential sources by directly contacting the soil and sediment. Although China`s research on sediment fingerprinting has a short time, it has developed rapidly and has been successfully applied to some areas. However, to date, there has been little research on sediment provenence in the black soil region of Northeast China, and it is still basically blank. In this study, the typical small watershed and its inner tiny catchment in the black soil region of Northeast China were selected as research area. According to the vertical distribution of soil ersion, potential sources of the basin were divided into topsoil and subsoil (gully). The topsoil was divided into cultivated land and non-cultivated land (forest and grass land) further based on land use. Using the sediment collected from gully bottom, reservoir, exit of the basin and the artificial mixed sediment, the following studies were carried out: Accuracy assessment of the quantitative models in predicting sediment contributions; Tracing capacity evaluation of fingerprints; Characteristics of sediment contributions of potential sources in the small watershed; Combined with the 210Pb-dating technique to invert the dynamic changes of soil erosion and sediment contribution in the tiny catchment; Differences of sediment contributions of snowmelt erosion and rainfall erosion. The main conclusions are as follows: (1) The sedimeng contribution results from composite fingerprints, radionuclides and single fingerprints were similar. For the deposited sediment, the topsoil and gully contributed about half in the black soil region of Northeast China. Especially, the source identification ability of 137Cs was up to 100%. Its confidence interval ranges of quantitative standard error and 95% confidence level were the smallest and similar to the quantitative results based on composite fingerprints. (2) Accuracy of Walling-Collins model and Bayesian model were more robust in predicting sediment sourece contributions with mean absolute error < 8%, and the accuracy of the two models was not significantly different (p >0.01). In contrast, the DFA model performed the worst in prrdicting sediment sourece contributions with mean absolute error > 18%. It means that the DFA model is not able to predict sediment source contributions accurately. The accuracy of the DFA model was significantly different (p < 0.01) with Walling-Collins model and Bayesian model. (3) 1976–2016, as the main sediment sources in small watershed, cultivated land and gully contributed 50.6% and 42.9% sediment to reservoir. The sediment contribution of cultivated land decreased with time, while the sediment contribution of gully increased. The non-cultivated land only contributed 6.5% of sediment and the trend was ont clear. Comparing the sediment contributions of two different periods, it was found that the sediment contribution of cultivated land in 1999–2016 was 5.7% lower than that in 1976–1998, and the sediment contribution of gully increased by 2.8%. The sediment contributions of internal and external reservoir were significantly different (p < 0.01). The sediment deposited in external reservoir was primary from gully, while the internal sediment was dominated by cultivated land. For the entire time series (1976–2016), the sedimentation rate was 1.2 cm yr-1 (10.7 kg m-2 yr-1). The annual average sedimentation rate of reservoir in 1999–2016 reduced by approximately 43.7% compared to 1976–1998. (4) The cultivated land and gully were both important sediment sources during 1974–2004, contributing 47.9% and 42.0% of the sediment, while only 10.1% of sediment was from non-coltivated land. There was a general trend for cultivated land to decrease (p < 0.01) and for gully erosion to increase (p < 0.01) over the period 1974–2004. Compared with the small watershed, the sediment contributions of the three sediment sources at two spatial scales are not much different. Among them, the contribution of non-cultivated in the tiny catchment is larger than that in the small watershed about 3.6%, while the sediment contributions of cultivated land and gully are slightly smaller than that of the small watershed about 2.7% and 0.9%. Based on the results of the 210Pb-dating technique, it was found that the annual average sediment yield of tiny catchment in 1980–1997 reduced by aboud 60% compared to 1974–1979. After 1998, the sediment yields remained 25% higher than those associated with the period 1980–1996 and this has been tentatively ascribed to the e?ects of the major ?ood in 1998 in activating erosion. However, it was about 53% lower than in 1974–1979. (5) The sediment contrbutions of potential sources between snowmelt erosion and rainfall erosion were significantly different (p < 0.01). The sediment eroded by rainfall was maily from cultivated land, followed by gully and non-cultivated land, while 70% of the sediment eroded by snowmelt was from gully, followed by cultivated land and non-cultivated land. The runoff of rainfall and snowmelt erosion was significantly correlated with the sediment contributions of cultivated land and gully (p < 0.01). There was a negative correlation between runoff of rainfall and sediment contributions of cultivated land, and a positive correlation with gully. However, the runoff of snowmelt was positively correlated with sediment contribution of cultivated land, and negatively correlated with gully. Above correlations indicated the reason that much more sediment contribution of gully than those of cultivated land and non-cultivated land in snowmelt ersion was not scouring effect on wall and bottom of gully, but the collapse of gully caused by freezing and thawing in the spring thawing period.