Dr Manuel Chevalier - Harnessing proxy data uncertainties to improve climate reconstructions in the tropics

Pollen-based quantitative climate reconstructions are fundamental to better understanding past environmental changes across various spatial and temporal scales and evaluating climate simulations. However, these currently mainly exist in extratropical North America and Eurasia, and very few quantifications exist in the tropics, where sparse pollen record networks and limitations in existing reconstruction techniques hinder performing comprehensive large-scale reconstructions. My work addresses this global data imbalance by focusing on the tropics. Having developed a novel statistical approach – the CREST (Climate REconstruction SofTware) method – that circumvents the main issues of the most-used techniques, I seek to analyse and integrate extensive networks of tropical pollen records, characterise regional spatiotemporal modes of climate variability, and unveil the main drivers of tropical climate changes at the millennial to orbital timescales. In this presentation, I will showcase the approach with a multivariate reconstruction using a 270,000-year-long, high-resolution pollen record from Laguna Fùquene in Colombia.