Southern Ocean Surface Temperatures and Cloud Biases in Climate Models Connected to the Representation of Glacial Deep Ocean Circulation

Simulating and reproducing the past Atlantic meridional overturning circulation (AMOC) with comprehen-sive climate models are essential to understanding past climate changes as well as to testing the ability of the models in sim-ulating different climates. At the Last Glacial Maximum (LGM), reconstructions show a shoaling of the AMOC compared to modern climate. However, almost all state-of-the-art climate models simulate a deeper LGM AMOC. Here, it is shown that this paleodata-model discrepancy is partly related to the climate model biases in modern sea surface temperatures (SST) over the Southern Ocean (70 degrees-45 degrees S). Analysis of model outputs from three phases of the Paleoclimate Model Inter -comparison Project shows that models with warm Southern Ocean SST biases tend to simulate a deepening of the LGM AMOC, while the opposite is observed in models with cold SST biases. As a result, a positive correlation of 0.41 is found between SST biases and LGM AMOC depth anomalies. Using sensitivity experiments with a climate model, we show, as an example, that changes in parameters associated with the fraction of cloud thermodynamic phase in a climate model re-duce the biases in the warm SST over the modern Southern Ocean. The less biased versions of the model then reproduce a colder Southern Ocean at the LGM, which increases formation of Antarctic Bottom Water and causes shoaling of the LGM AMOC, without affecting the LGM climate in other regions. The results highlight the importance of sea surface con-ditions and clouds over the Southern Ocean in simulating past and future global climates.