Noah-MP模型中积雪模拟对参数化方案的敏感性评估

针对Noah-MP模型多参数化方案、模拟结果不确定性范围难以确定的特点,选取北疆地区具有代表性的阿勒泰站气象资料作为模型驱动数据,探讨了积雪对多参数化方案的敏感性。在不考虑模型参数和驱动数据不确定性的条件下,设计了集合数为13824的多参数化方案集合模拟试验。选用Natural selection方法对物理过程的敏感性进行分析,并在敏感性分析结果的基础上进一步讨论了模拟结果的不确定性。结果表明:积雪对地表热交换、雨雪分离、土壤温度底层边界条件和第一层积雪或土壤时间方案4个物理过程敏感;在不考虑驱动数据和模型参数不确定性的条件下,多参数化方案集合模拟试验中的不确定性主要来源于敏感物理过程。去除敏感物理过程中能够明显降低模拟性能的参数化方案后,集合模拟结果的不确定性大幅减小。最后,根据分析结果构建了该站雪深和雪水当量模拟的最优参数化方案组合。

Sensitivity Evaluation of Snow Simulation to Multi-parameterization Schemes in the Noah-MP Model

On account of the latest community Noah land surface model with multi-parameterization(NoahMP)schemes and its uncertainty breadth in simulation results being difficult to be determined,this study assessed the sensitivity of snow to physics options using meteorological data from the Altay Station in northern Xinjiang. The Noah-MP physics ensemble simulation with the total number of 13 824 was designed without the consideration of the uncertainties of forcing data and parameters. The natural selection approach was used to analyze the sensitivity of physical processes. Based on the results of sensitivity analysis,the uncertainty of ensemble simulation results was further discussed. The results showed that snow was sensitive to the physical processes of surface-layer exchange coefficient,partitioning precipitation into rainfall and snowfall,lower boundary condition of soil temperature,and first-layer snow or soil temperature time scheme;Uncertainties in multi-parameterization ensemble simulation experiments were mainly from sensitive physical processes under the condition of disregarding uncertainties of forcing data and parameters. After removing the parameterization schemes that notably reduced simulation performance in sensitive physical processes,the uncertainty breadth in ensemble simulations decreased significantly. Finally,an optimal combination group of parameterization schemes for this station was configured.