Steven Lade (Postdoctoral Researcher, Max Planck Institute for the Physics of Complex Systems, Dresden, Germany; Participant in the 2009 Complex Systems Summer School)
Abstract: Critical transitions, here defined as sudden changes of state, occur in many systems in nature and society, such as ecology, physiology, climate, and economies. Given the often catastrophic nature of these transitions, some warning of these transitions is highly desirable. Over the last decade a number of such early warning signals have been proposed based on simple analyses of time series data, for example an increasing variance or increasing autocorrelation.
These methods however can be limited by the amount of data they require. In this talk I will propose a new method that significantly reduces the amount of data required. It is based on combining multiple types of time series data with system-specific structural knowledge through the framework of a generalised model. I apply the method to three ecological examples, including the simulated collapse of a fishery.