基于Noah-MP模型多参数化方案的积雪过程集合模拟与同化研究

积雪能够对全球变化做出迅速响应，是研究气候变化的重要变量。如何有效利用现有积雪模型和观测资料对积雪的时空分布进行合理的估计是当前研究的热点内容。为了进一步提高雪深集合模拟的精度，本项目拟以北疆地区多源数据（站点雪深、中国积雪反照率数据、AMSR-2亮度温度）为基础，开展雪深多参数化方案集合模拟与同化研究。首先，利用Noah-MP模型在区域尺度上开展雪深多参数化方案集合模拟，量化雪深模拟结果对参数化方案的敏感性；其次，利用贝叶斯模型平均方法实现多种组合方案的雪深预报，减少模型结构的不确定性。最后，利用数据同化技术提升模型的雪深模拟性能，构建联合多参数化方案集合模拟与多观测资料的积雪数据同化系统，获得准确的雪深预报结果。本项研究可为雪深模拟提供新的思路，为开展融雪径流模拟、积雪水资源利用研究奠定基础。

Snow can respond to global changes rapidly and is an important variable in the study of climate change. How to effectively use the existing snow model and observational data to estimate the temporal and spatial distribution of snow is a hot topic in current research. In order to improve the accuracy of ensemble simulated snow depth, this proposed research will carry out snow depth multi-parameterization ensemble simulation and data assimilation based on multi-source data in northern Xinjiang (snow depth in-situ, China snow albedo data product, AMSR brightness temperature). First, the Noah-MP model will be used to carry out snow depth multi-parameterization ensemble simulation on regional scale to quantify the sensitivity of simulated snow depth to parameterizations; secondly, the Bayesian model averaging method will be used to achieve snow depth prediction of multiple combination schemes, and the uncertainty of model structure will be reduced in this way. Finally, data assimilation technology will be used to improve the snow depth simulation performance of the model, and a snow data assimilation system will be configured by combining multi-parameterization ensemble simulation and multiple observation data to obtain accuracy snow depth prediction. The expected research can provide a new idea for snow depth simulation and will lay foundation for carrying out snowmelt runoff simulation and snow water utilization research.