Komlan Midodzi Noukpoape (Inria MODAL) : Bayesian statistics for reconstructing past climates and environments

We propose a new method for estimating reference curves using nonparametric regression. Our approach is fully Bayesian and relies on cubic smoothing splines, making it well suited to a wide range of data. We apply the proposed method to sedimentary sequence data to produce age-depth curves, which are highly valuable in paleoecological and paleoenvironmental studies, particularly for reconstructing past climates and environments. We compare our approach with Bchron, OxCal, and Bacon, three age-depth modeling methods commonly used in the literature. The key difference between our approach and existing age-depth chronology methods lies in the model design: unlike these methods, ours makes no assumptions about the sediment accumulation rate.