Semanas Franco‑Uspianas 2025

Sessão 3: Harnessing proxy data uncertainties to improve past climate reconstructions in the tropics

Resumo: Pollen-based quantitative climate reconstructions are essential for advancing our understanding of past environmental changes across various spatial and temporal scales, as well as for evaluating climate simulations. However, these reconstructions are primarily found in extratropical North America and Eurasia, with very few quantifications available in the tropics. In the tropics, sparse pollen record networks and limitations in existing reconstruction techniques hinder comprehensive large-scale analyses. My work addresses this global data imbalance by focusing on modelling tropical pollen data. To achieve this, I first developed a statistical approach – the CREST (Climate REconstruction SofTware) method – that overcomes the main issues associated with the most widely used techniques. Subsequently, I created a pantropical calibration dataset, leveraging the growth of public datasets to enable reconstructing climate from most tropical regions. With this, I aim to integrate extensive networks of tropical pollen records, characterise regional spatiotemporal modes of climate variability, and reveal the primary drivers of tropical climate changes at millennial to orbital timescales. In this presentation, I will showcase the approach with a 270,000-year-long multivariate reconstruction using the high-resolution pollen record from Laguna Fúquene in Colombia. I will also present preliminary reconstructions from the Colônia record.