Multivariate Climate Projections: More Accurate Equilibrium Estimations & Evolution of Climate Feedbacks

Robbin Bastiaansen
Utrecht University
Tuesday, April 13, 2021 - 12:20pm to 1:20pm
See abstract for Zoom info

When the climate system is forced by e.g. changes in atmospheric CO2, it responds to this change on multiple time scales, showing responses on time scales ranging from very short to very long. It is clear that the behavior over these time scales can be very different: for instance, as ice melts, the ice-albedo feedback becomes less and less important. Predominantly used climate projection methods, however, typically do not adequately take such state changes into account and are univariate: they only consider the global mean surface temperature -- assuming everything else is just linearly correlated to that. In this talk, I will show multivariate estimation techniques that are capable of tracking these state changes by incorporating additional observables into the analysis directly. This has two important advantages. First, such methods are better equipped to provide projections for the longer time scales (for instance, estimations of equilibrium climate sensitivity). Second, it makes it possible to estimate other observables directly -- without making assumptions on their relation to the global mean surface temperature -- which leads to better quantitative insights in how precisely the climate will change in the future (for instance, how climate feedback processes might change over time).Zoom link (provide email address to receive link):