Understanding threshold-crossing events: New approaches to use computer models for time series analysis of complex systems

许多复杂的动力系统表现出跨越阈值的行为，即在根本不同的运行状态之间突然切换。这些翻转的原因往往鲜为人知，尽管它们对这些系统的动态具有根本的重要性。一个突出的例子是迄今无法解释的冰川气候中的大幅度(10-15K)转换，Dansgaard-Oeschger(DO)事件显示出明显的趋势，最好是在大约1470年的接近倍数的情况下再次发生。该项目的目的是开发和应用基于计算机模型的新技术，以改进对跨越阈值事件的时间序列分析。主要目的是了解是什么导致了冰川气候中的DO事件。提出了几种假设来解释这些气候变化及其特征的重现时间。但检验这些假设是非常重要的，需要新的概念性方法，例如基于替代数据的零假设检验，基于递归图方法，可能还基于复杂网络。我希望这个项目能够对自然气候变化的主要原因以及未来可能发生的大范围气候不稳定提供新的见解。此外，由于许多动力系统中存在跨越阈值的行为，该项目还可以为对远远超出地球科学的复杂系统的创新时间序列分析提供新的刺激。

Many complex dynamical systems exhibit threshold-crossing behaviour, i.e. sudden flips between fundamentally different states of operation. The cause of these flips is often poorly understood, despite their fundamental importance for the dynamics of these systems. A prominent example are so far unexplained, large-amplitude (10-15 K) switchings in glacial climate, the Dansgaard-Oeschger (DO) events, which show an apparent tendency to recur preferably in near multiples of about 1470 years.The aim of this project is to develop and apply new techniques, based on computer models, for improved time series analyses of threshold-crossing events. The main objective is to understand what caused DO events in glacial climate. Several hypotheses were proposed to explain these climate shifts and their characteristic recurrence time. But testing these hypotheses is highly non-trivial and requires new conceptual approaches, based e.g. on null-hypothesis testing with surrogate data, on recurrence plot methods and, possibly, on complex networks. I expect that this project could give new insight into main causes of natural climate variability and concerning the potential occurrence of large-amplitude climate instabilities in the future. Moreover, since threshold-crossing behaviour exists in many dynamical systems, this project could also give new stimulus to innovative time series analyses of complex systems far beyond geoscience.