Abstract:
The decadal climate variability observed since the early 20th century arises from an interplay between internal climate processes and external forcings. Disentangling these components is essential for accurate decadal climate forecasting, improving our ability to monitor and mitigate anthropogenic climate change. Despite significant progress, the relative contributions of internal and externally forced variability remain a subject of active debate within the scientific community. This presentation will focus on methods using climate model simulations to differentiate between internal and forced variability. A method using convolutional neural networks will be presented to recognize in observations the forced spatio-temporal variability generated by climate models. The results will be compared to that of other methods. The associated uncertainty regarding the main mode of decadal variability, such as the Atlantic Multidecadal Variability or the Pacific Decadal Oscillation will be discussed.
Bio:
Guillaume Gastineau graduated from Sorbonne University, Paris, France, and earned his Ph.D. from the same institution in 2008. He pursued postdoctoral research at the University of Miami, Florida, USA, before becoming an Assistant Professor at Sorbonne University’s LOCEAN laboratory, part of the Institut Pierre-Simon Laplace (IPSL), in Paris in 2011. His research explores mechanisms of climate variability using long-term observations, climate models, and statistical methodologies. Dr. Gastineau’s findings have been featured in journals such as Nature Communications, Journal of Climate, Climate Dynamics, and Journal of Advances in Modeling Earth Systems.
