Streamflow data assimilation for numerical weather prediction models

数值天气预报（NWP）以及定量降水预报（QPF）在很大程度上是一个初始值和边值问题。夏季大陆降水预报的技巧取决于土壤湿度和其他陆地表面状态的初始化。虽然Deutscher weterdienst （DWD）在Local Model （LM）中实现了一种吸收2m空气温度并调整土壤状态以满足观测到的大气状态的方法，但该方法的土壤湿度并不一定反映实际情况。该项目的目的是研究径流数据同化的潜力，以便更好地估计根区土壤水分含量的水平和垂直分布，特别是在植被茂密的地区，从而有助于改善定量降水预报的初始条件。一个观测算子，它定义了单个网格框对空间和时间集成的水流观测的贡献，将基于模型集合导出。该项目的结果将为DWD和科学界提供对河流数据在改善土壤湿度方面的潜力的估计，以及对与水平衡有关的陆地表面模型误差的量化。在LM-TERRA中建立路线模型不仅有利于DWD，也有利于洪水预报服务。

Numerical weather prediction (NWP) and therefore quantitative precipitation forecast (QPF) is to a large extent an initial and boundary value problem. Skill in summertime continental precipitation prediction depends an the initialization of soil moisture and other land surface states. Though the Deutscher Wetterdienst (DWD) implemented a method into the Local Model (LM) that assimilates 2m-air temperature and adjusts the soil state to meet the observed atmospheric state, the soil moisture of this method does not necessarily reflect reality. The objective of this project is to investigate the potential of streamflow data assimilation to obtain a better estimate of the horizontal and vertical distribution of the root zone soil moisture content especially in areas of dense vegetation to contribute to an improvement of the initial condition for quantitative precipitation forecast. An observation operator, which defines the contribution of an individual grid box to the spatially and temporally integrated streamflow observation will be derived based an model ensembles. The results of this project will provide the DWD and scientific community with an estimate of the potential of the streamflow data in soil moisture improvements and a quantification of the land surface model error with respect to the water balance. A routing model within the LM-TERRA will benefit not only the DWD but also the flood forecasting services.