Narrowing Uncertainties of Climate Projections using Data-driven methods

PIERRE TANDEO; PIERRE LE BRAS; FLORIAN SEVELLEC; PIERRE AILLIOT; JUAN RUIZ

Sapporo

Workshop; 6th Workshop on Mesoscale Modeling; 2023

Climate simulations are available from different numerical models and gathered within theCoupled Model Intercomparison Project (CMIP, see Eyring et al. 2016). Each of these modelshas its own strengths and weaknesses, which impact its ability to predict future climate.In this presentation, we focus on a specific climate index, which has a major influence on theclimate of Europe and North America, the Atlantic Meridional Overturning Circulation(AMOC). All CMIP6 future projections show a slowdown of this circulation in the comingdecades (see Lyu et al. 2020). This decrease is not homogeneous: several models predict adramatic reduction of the AMOC while others predict small changes in the future.The goal of this presentation is to improve the projections of the AMOC in the next decadesby reducing their spread. To do this, we investigate new data-driven methods to weightclimate simulations, assuming that some climate models are better than others at describingthis climate index.