早在1965年，Lorenz采用奇异向量方法研究大气可预报性。奇异向量可表征线性动力增长最快的初始扰动，但关于奇异向量对其特征矩阵的范数选取是否敏感，仍存在争议。为了让大家了解可预报性研究中最常用的奇异向量的最新理论研究进展，特邀美国佛罗里达州立大学Jon Ahlquist副教授来校访问交流，欢迎大家踊跃参加！
报告时间：5月8日（周三）下午3:00
报告地点：气象楼423报告厅
报告题目：Almost Anything Can Be the Leading Singular Vector
主持人：虞越越 教授
摘要：Lorenz (1965) pioneered the use of singular vectors to study atmospheric predictability. Singular vectors represent initial perturbations that grow as much as possible in a finite time according to linearized dynamics. That is, we seek vector x that maximizes ||Lx|| / ||x||, where Lx represents the linearized growth of the initial condition x during some specified time. The norm used to measure the initial size of x, i.e., ||x||, does not have to be the same norm as that used to measure the size of x after linearized growth, i.e., ||Lx||. The question has been asked, how sensitive is the singular vector x to the norms used? Kuang (JAS, 2004, 2005) concluded that: there may not be as much norm-related ambiguity in problems such as designing targeted observations or ensemble forecasts, as is often assigned to them. In this study, we show that the situation is very different. Specifically, given any perturbation x such that Lx is not 0, we can design initial and final norms so that x grows more than any other perturbation with whatever growth factor you want.
气象灾害教育部重点实验室
气象灾害预报预警与评估协同创新中心
大气科学学院
2019年5月7日
