This study examines the climate response to different magnitudes of Atlantic Meridional Overturning Circulation (AMOC) weakening under glacial conditions using the Australian Earth System Model. A potential threshold is identified between linear AMOC slowdown and nonlinear shutdown, driven by a critical change in ocean heat loss that induces a nonlinear atmospheric response governed by energetic constraints.

We compare pollen-based reconstructions of Northern Hemisphere tree cover over the last 20 000 years with simulations from the Max-Planck-Institute Earth System Model (MPI-ESM). The model captures broad forest trends but misses key regional patterns and the mid-Holocene forest peak. Testing climate drivers reveals mismatches in how temperature, water, and CO₂ control forests, pointing to structural limits and the need for improved vegetation processes in models.

We introduce a new climate proxy based on plant DNA preserved in lake sediments. Validated with a large surface sample dataset and applied to a sediment record, this method provides more accurate and robust reconstructions of past climate change than traditional vegetation proxies like pollen, likely due to a higher taxonomic resolution and more localized signal.

This high-resolution lake sediment record reveals 230 floods over 8,000 years in Western Norway. Flood regimes shifted from rainfall to snowmelt dominance around 3,100 years ago, with the last 500 years being exceptionally flood-rich. These findings show how climate and human activity shape flood risk and underscore the need for long-term archives to guide adaptation under future warming.

During the mid-Miocene (~16 to 14 million years ago), Earth was much warmer with CO₂ levels similar to what is expected for our future. Here, we perform simulations with a new geography of the mid-Miocene, which include changes in ice coverage, land and sea distribution, solar energy and CO₂ concentrations. Despite high CO₂ concentrations, our mid-Miocene is too cold compared to geological reconstructions, but shows great potential to understand future climate change and its sensitivity to CO₂.

To better understand climate drivers during a warm period ~200,000 years ago, we reconstructed global surface temperature using 132 marine records placed on a common timeline. We found strong hemispheric differences and showed that the warmest phase occurred despite lower CO₂ levels. In addition, we suggest that extreme orbital forcing temporarily shifted the climate into the older 41,000-year rhythm, resulting in a hybrid warm period shaped by both 100,000-year and 41,000-year climate dynamics.