Early warning signals and abrupt transitions
in climate
Bedartha Goswami
Cluster of Excellence “Machine Learning”
University of Tübingen (Tübingen), Germany
Abstract
Early warning signals (EWS) can in principle allow society to anticipate potentially devas-
tating and practically irreversible abrupt critical transitions in climate systems. Several
tipping elements of the climate systems such as the Amazon rainforest, the Greenland ice
sheet, and the Indian monsoon are already identified as being at risk of abruptly tipping
to undesirable states. However, there is no clear consensus on how to estimate EWS from
observed climatic time series. Moreover, in the context of paleoclimate, early warning
analyses that take dating uncertainties into account are rather uncommon. Here, we present
an overview of the state-of-the-art of EWS in climate systems and potential pitfalls of a
purely data-driven approach to estimate EWS. Last, we present two ways in which complex
networks can help to detect EWS in climate systems: one based on recurrence networks of
paleoclimate time series, and the other based on climate networks constructed from extreme
rainfall time series.