Developing 2-way feedback nested-domain LETKF data assimilation system and application to high-resolution typhoon analyses and forecasts

这个项目在第二个财政年度中期结束，因为首席调查员搬到了日本以外。在本项目为期一年零四个月的时间里，在线嵌套域LETKF在Scale-LETKF系统中得到了新的实现。与Scale-RM模式中现有的在线嵌套域功能相结合，Scale-LETKF已经能够在一个在线框架中运行数据同化循环和具有多个嵌套域的扩展集合预报。该设计在计算和人工方面都是高效的。利用这个新的在线嵌套功能进行了2015年9月关东-东北暴雨过程的数据同化实验。与Lien等人的旧的离线缩减实验相比。(2017，Sola)，由于进一步在内部高分辨率区域进行数据同化以及模式预测步骤中两个区域的频繁耦合，在线嵌套域实验导致了预报技能的提高。主要研究者还与人合著了Honda等人的两篇论文。(2018a，b，MWR，JGR)用于台风资料同化研究的SCALE-LETKF系统。本项目开发的所有源代码已在https://github.com/gylien/scale-letkf，上公开，并获得麻省理工学院的许可。由于本项目的提前结束，具有多个嵌套区域的双向反馈LETKF分析仍在开发中。首席调查员也许能够继续开展这方面的工作，并在他的下一个职位上结束工作。

This project ended in the middle of the second fiscal year because the principal investigator moved outside of Japan. During the one-year and four-month period of this project, the online nested-domain LETKF has been newly implemented in the SCALE-LETKF system. Combined with the existing online nested-domain function in the SCALE-RM model, the SCALE-LETKF has become capable of running both data assimilation cycles and extended ensemble forecasts with multiple nested domains in an online framework. The design is efficient in both the computational and human-effort aspects.The data assimilation experiments with the September 2015 Kanto-Tohoku Heavy Rainfall case were conducted using this new online nesting function. Compared with the old offline downscaling experiment in Lien et al. (2017, SOLA), the online nested-domain experiment led to improved forecast skill, due to further performing data assimilation in the inner high-resolution domain and frequent coupling of the two domains in the model forecast steps.The principal investigator also coauthored two papers by Honda et al. (2018a,b, MWR, JGR) for typhoon data assimilation studies using this SCALE-LETKF system.All source code developed in this project is publicly available at https://github.com/gylien/scale-letkf, with an MIT license.Due to early termination of this project, the 2-way feedback LETKF analysis with multiple nested domains is still under development. The principal investigator may be able to continue working on it and concluding the work in his next position.