Multimodel comparison of weathering fluxes during the last deglaciation

The global carbon cycle is a complex system with many drivers, including slow ones such as the chemical weathering of rocks. At long enough timescales, changes in weathering rates influence CO₂ consumption, but also the river loads of carbon, nutrients, and alkalinity. In particular, the global ocean inventory of alkalinity is a critical driver of carbon sequestration into the ocean. Thus, any transitory imbalance between the sources and sinks of alkalinity can lead to changes in ocean chemistry and impact atmospheric CO₂ concentration. During the last deglaciation (ca. 19-11 ka BP), the Earth’s climate transitioned from cold and arid to comparatively warmer and wetter conditions. Simultaneously, large ice sheets melted and led to a significant rise of sea level (ca. +120 m), which reduced the size of the exposed continental shelves. Loess deposits were also gradually eroded. These changes logically influenced the chemical weathering of rocks because weathering rates depend on climate variables (runoff and temperature), land-sea distribution and lithology. Some modelling studies and proxy reconstructions suggest little net changes over this period. Yet, the deglacial changes of weathering rates remain poorly constrained.

Most Earth System Models do not explicitly represent weathering and the consequent river fluxes. Moreover, the alkalinity inventory is often assumed constant in models, despite the fact that proxy data suggest an elevated total alkalinity at the Last Glacial Maximum (and the likely changes of its sources and sinks). These choices can potentially bias the model representation of the global carbon cycle, whose deglacial variations have been notoriously hard to simulate for decades. In this study, we calculate weathering fluxes of phosphorus and alkalinity (among others) using reconstructed lithological maps, and model results from transient runs of the last deglaciation and/or time-slice runs of the Last Glacial Maximum and pre-industrial period. To improve robustness, we compare the evolution and spatial distribution of weathering fluxes in different models. We demonstrate that while the increase of runoff during deglaciation enhances weathering, the rise of sea level and the erosion of loess deposits tend to have a counterbalancing effect on the river loads. Our model ensemble tends to show inconsistent deglacial changes of some river loads (e.g. for phosphorus), depending both on runoff biases and on the representation of land-sea distribution. Still, all models indicate a significant decrease of river alkalinity from the LGM to the pre-industrial. Using these findings, we discuss the implications of an explicit representation of weathering fluxes for the global carbon cycle in transient runs with Earth System Models.