21st April 2026
Courtney Quinn
Rate-induced tipping: searching for the invisible threshold
Identifying and studying rate-induced tipping (r-tipping) remains a challenge in many applications as it requires knowing the tipping threshold of a system as well as identifying when a trajectory will cross that threshold. For low-dimensional systems this threshold is sometimes directly computable as an unstable invariant object. As dimension increases, however, this threshold becomes increasingly difficult to track under various forcing scenarios. In this talk I will review how I became interested in r-tipping and how I am attempting to understand the complexity of it in higher dimensional systems. I will discuss some new results, including examples of r-tipping in delay-differential equations and r-tipping without basin instability. Finally, I will end with an outlook on avenues for identification in general high-dimensional systems.
The methodologies are here applied to a data set from a simulation of AMOC collapse with a complex climate model, actually a freshwater hosing experiment with the FAMOUS GCM. The AMOC on-state is found to lose stability via a subcritical Hopf bifurcation; however, the transition to the off-state occurs far ahead of the bifurcation point. The early collapse can be explained by a combination of rate-induced and noise-induced tipping.
28th April 2026
Frank Kwasniok
Data-driven anticipation and prediction of Atlantic Meridional Overturning Circulation collapse using non-autonomous spatio-temporal dynamical modelling
Data-driven methodologies for identifying, anticipating and predicting critical transitions in high-dimensional model or observational data sets are introduced, based on explicit non-stationary low-order modelling of the tipping dynamics, allowing for dynamical understanding of the underlying tipping mechanism and genuine prediction of the future system state by extrapolation. A set of spatial modes carrying the tipping dynamics are identified and a stochastic model of appropriate complexity is estimated in the subspace spanned by these modes. Analysis of the reconstructed dynamics provides information on the proximity to a bifurcation point and the type of the impending bifurcation. In a first step, we focus on linear stability analysis and derive non-autonomous dynamic and optimal mode decompositions (non-aut-DMD, non-aut-OMD) as extensions of the stationary DMD and OMD. In a second step, we estimate nonlinear stochastic low-order models. Different competing tipping mechanisms can be compared and assessed using likelihood inference and information criteria. The method allows to quantify the likelihood or risk of a critical transition at some point in the future having observed a certain amount of data up to present.
26th May 2026
Henk Dijkstra
AdvanTip Webinar: Towards Early Warning Systems for AMOC and SPG tipping
The AMOC and SPG are considered to be separate, but connected, tipping systems which both may undergo transitions due to global warming. A main challenge is the design of an Early Warning System (EWS) of such transitions using available (and possibly future new) observations. Here, I will present initial ideas on the design of these EWSs (addressing issues such as the warning signal and optimal observation locations) and on their testing. The ideas will be illustrated using results from a hierarchy of ocean models.
9th June 2026
Jan Saynisch-Wagner
Enhancing machine learning performance under changing conditions such as tipping.
On the one hand, neural networks increasingly are employed in Earth and climate sciences. On the other hand, many Earth systems show increased and accelerated changes with more to be expected in the future. As neural networks often show decreased performance out of their training data realm, these two developments do not work together very well. In the talk, we present a method to enhance neural network performance when training data and application data are not very similar, e.g., so called out of distribution problems. We introduce and discuss this method with nonlinear use cases from the climate sciences, which include temporal, spatial and cross-domain extrapolations of neural networks and AMOC tipping as well. The method consists of three main steps:
- Retrain the neural network towards reasonable subsets of the training data set and note down the resulting weight anomalies.
- Choose reasonable predictors and derive a regression between the predictors and the weight anomalies.
- Extrapolate the weights, and thereby the neural network, to the application data.
We present successful applications, discuss the downsides of the approach and possible ways for further improvement.