A multi-model assessment of the early last deglaciation (PMIP4 LDv1): The meltwater paradox reigns supreme

Abstract. Transient simulations of the last deglaciation have been increasingly performed to better understand the processes leading to both the overall deglacial climate trajectory as well as the centennial- to decadal- scale climate variations prevalent during deglaciations. The Paleoclimate Modelling Intercomparison Project (PMIP) has provided a framework for an internationally coordinated effort in simulating the last deglaciation (~20 – 11 ka BP) whilst encompassing a broad range of models. Here, we present a multi-model intercomparison of 17 simulations of the early part of the last deglaciation (~20 – 15 ka BP) from nine different climate models spanning a range of model complexities and uncertain boundary conditions/forcings.

A main contrasting element between the simulations is the method by which groups implement freshwater fluxes from the melting ice sheets and how this forcing then impacts ocean circulation and surface climate. We find that the choice of meltwater scenario heavily impacts the deglacial climate evolution, but the response of each model depends largely on the sensitivity of the model to the freshwater forcing as well as to other aspects of the experimental design (e.g., CO₂ forcing or ice sheet reconstruction). There is agreement throughout the ensemble that warming begins in the high latitudes associated with increasing insolation and delayed warming in the tropics aligned with the later increases in atmospheric CO₂ concentration. The delay in this warming in the tropics is dependent on the timescale of the CO₂ reconstruction used by the modelling group. Simulations with freshwater forcings greater than 0.1 Sverdrup (Sv) after 18 ka BP experience delayed warming in the North Atlantic, whereas simulations with smaller freshwater forcings begin deglaciating sooner. All simulations show a strong correlation between North Atlantic temperatures, atmospheric CO₂ concentrations, and the AMOC. In simulations with a freshwater forcing greater than 0.1 Sv, North Atlantic temperatures correlate strongly with changes in the AMOC. Simulations with a smaller freshwater forcing show stronger correlations with atmospheric CO₂. This indicates that the amount of meltwater strongly controls the climate trajectory of the deglaciation. Comparing multiple simulations run by the same model demonstrate model biases by showing similar surface climate spatial patterns despite the use of different ice sheet reconstructions and/or meltwater flux scenarios. Simulations run with different models, but similar boundary conditions, have provided insight into the sensitivity of individual models to particular forcings, such as the amount freshwater forcing, which has been highly debated in previous studies.

This debate has stemmed from the so-called ‘meltwater paradox’ that exists in choosing how much meltwater to input into simulations of the last deglaciation (i.e., large and geologically inconsistent meltwater forcings that successfully produce abrupt climate events versus glaciologically realistic meltwater fluxes that do not). The results of this research highlight how important this decision is.