气候变化对湖库水质的影响是目前广泛关注但比较难于解决的问题。本项目的研究思路为“统计模型发掘数据规律，水生态模型验证统计模型从机理上揭示影响机制”。遵循“定量贡献-阐明机制-应对变化”的研究主线，以于桥水库及其流域为例，依据30年的气候要素和水质历史数据，考虑人类活动对水质的影响，利用非参数回归等统计模型定量分析气候变化和人类活动对水质的贡献；基于扩展型卡尔曼滤波法剔除人为因素干扰，通过水动力-水质-水生态模型从机理上模拟验证统计模型，阐明气候变化对水库水质直接和间接影响机制；基于CMIP5多模式的CORDEX降尺度气象数据，形成气候情景不确定性方案组（3+3+1），模拟预测中、远期气候情景和极端气候对水质影响程度，提出保障供水安全的气候变化应对建议。项目为阐明气候变化对湖库水质的影响机制提供理论依据，在制定适应对策和用水安全方面具有重要的科学意义。

Climate change is an important factor affecting water quality in lakes and reservoirs. The proposing of the research is that data mining discovers underlying rules by statistical model; to calibrate the statistical model, the mechanisms of impacts of climate change on water quality are illustrated by a coupled water quality - macrophytes model..The Yuqiao Reservoir as a case study for the shallow reservoir of water source region, the objectives are the following: (1) to quantify the impacts of different climate indicators on water quality with quantitative assessment method; (2) to analyze the influence mechanisms of climate change on water quality with the coupled water quality - macrophytes model; (3) to predict the regional impacts of climate scenarios uncertainty on water quality in medium-long term..Considering the effects of human activity and climate change on water quality, non-parametric regression is performed to explore potential key climate drivers of water quality using historical climate indicators and water quality parameters datasets in 30 years. Anthropogenic factors are separated from synthetical impacts by extended kalman filter method. The coupled water quality - macrophytes model calibrates the statistical model to illustrate the influence mechanisms. Climate scenarios (RCP8.5, RCP4.5, RCP2.6) in 2020-2030 and 2030-2050 and extreme weather are conducted based on the COordinated Regional climate Downscaling EXperiment (CORDEX)-East Asia climate data..The study will provide theoretic foundation for mechanisms of climate change affect on water quality and will be of great significance for guaranteeing water security in China.