新安江水库（也称千岛湖）是我国长三角地区的重要水源，维系千岛湖的新安江流域是我国不可多得的优质水源地与生态屏障。1970s以来，受气候变化与人类活动影响，新安江流域的水资源与水环境发生变化，威胁到区域供水安全。开展变化条件下新安江流域的水文与水质效应研究具有重要的科学意义与应用价值。本文以新安江流域上游（街口站以上）为研究区，基于多源观测数据分析了水量水质变化及其影响因素；构建了分布式水文与生物地球化学过程耦合的数值模型，模拟了过去40年流域水-沙-氮磷演变过程；解析了坡面径流、土壤侵蚀和氮磷负荷的时空特征及其影响因素，揭示了流域水量与氮磷物质平衡关系及河道水质变化规律；基于未来气候模式结果，采用机器学习算法预估了未来变化情景，通过数值模拟预估了未来50年流域径流与水质变化。主要结论如下：（1）站点观测数据分析结果显示，研究区在1970~2019年的气温以0.012~0.036 ℃/yr的速率上升，降雨日平均雨强以每年0.002~0.090 mm/day速率增加；日流量的极大与极小值分别发生在6月中旬和10月下旬~11月下旬；年内的径流涨落逆转次数和低流量脉冲次数每年分别增加0.38~0.55次和0.11~0.37次；河道总氮（TN）浓度在枯水期较高，总磷（TP）浓度在汛期较高。（2）数值模拟结果分析表明，研究区在1980~2019年间TN和TP的坡面污染负荷分别为3219.2 kg/km2/yr和293.3 kg/km2/yr，入河系数分别为 0.40和0.38，其中，汛期TN和TP的入河量分别占年总入河量的65%和63%，河道TN和TP滞留系数分别为0.86和0.89；TN负荷强度的年际变化受径流、土壤侵蚀和流域污染输入等因素影响，其空间分布与土地利用具有较好的一致性；TP负荷强度的年际变化主要受降雨-径流过程影响，其空间分布与径流一致；大气氮沉降对TN入河量的贡献率高达65%~71%，说明大气沉降污染是该水源地的主要污染源。（3）基于SSP1-2.6、SSP2-4.5和SSP5-8.5情景的预估结果显示，2021~2070年研究区径流将减小0.7%~5.9%，SSP1-2.6和SSP2-4.5情景下流域出口TN和TP负荷将分别减少19.9%~24.5%和0.3%~4.6%，SSP5-8.5情景下TN与TP负荷将分别增加4.5%和7.7%，说明绿色发展对维持新安江水源地优良水质的重要性。以上研究成果可为新安江水源地科学应对气候变化与管理水土资源，促进高品质水源地水质保护提供依据。
The Xin ‘an River Reservoir is an important water source in the Yangtze River Delta region of China. Since the 1970s, the water resources and water environment in the Xin‘an River basin have been greatly influenced by the climate change and human activities, threatening the safety of the regional water supply. Therefore, research on the runoff, sediment, nitrogen, and phosphorus dynamics in the upper Xin’an River basin under changing environment is of great scientific significance and application value.The upper Xin‘an River basin (above Jiekou station) is the study area of this research. Based on the multi-source observation data, this study analyzed the characteristics of upstream water quantity and quality and its influencing factors and established a distributed process-based model to simulate the hydrological processes together with the nonpoint source pollutants transport and transform processes in the upper Xin’an River basin (XRB). Through numerical simulation, long-term spatial (1 km×1 km) variations in the runoff, sediment, nitrogen and phosphorus loads in the past 40 years were reproduced, and the spatiotemporal changes in the runoff and water quality in the next 50 years were estimated. The following main conclusions have been reached:First, according to the analysis of observation data, the annual-mean air temperature in the XRB increased at a rate of 0.012~0.036 ℃/yr. The average rainfall at rainfall days increased at a rate of 0.002~0.090 mm/day per year. The maximum and minimum daily discharge occurred in mid-June and late October to late November respectively. The number of discharge fluctuation reversal and low discharge pulses increased by 0.38~0.55 times/yr and 0.11~0.37 times/yr, respectively. River total nitrogen (TN) concentration was higher in winter and spring, and the concentration of total phosphorus (TP) was higher in the flood season.Second, based on the numerical simulation, the long-term average annual TN and annual TP loads were 3219.2 kg/km2 and 293.3 kg/km2, respectively. And the proportional coefficients of TN and TP loaded from hillslopes into the rivers were 0.40 and 0.38, respectively. The TN and TP load during flood season from April to July accounted for 65% and 63% of the annual total. The temporal variation of annual TN load was significantly correlated with that of annual runoff depth, soil erosion and TN input, while TP load variation was significantly correlated with annual runoff depth and soil erosion. The long-term average TN load showed similar spatial patterns with the land use and the TP load showed similar spatial patterns with runoff depth, respectively. The TN load from atmospheric deposition accounted for 65%~71% of the total inputs, which implied that the water pollution caused by atmospheric deposition pollution in the study headwater region cannot be ignored.Third, based on the GCM output, this study used machine learning algorithms to estimate the changes in land use types and pollution sources under the shared socioeconomic paths. Through numerical simulations, the upstream discharge reduction of SSP1-2.6, SSP2-4.5, and SSP5-8.5 during 2021~2070 is projected to be 0.7%~5.9%. TN (TP) loads of SSP1-2.6 and SSP2-4.5 are predicted to decrease by 19.9%~24.5% (0.3%~4.6%). TN and TP loads are predicted to increase by 4.5% and 7.7% in SSP5-8.5. It is clarified that green development is an important way to maintain excellent water quality in the upper Xin‘an River basin.The results of this study can provide a basis for the scientific response to climate change and the management of soil and water resources, and protection of high-quality in the study area.
