Toward the Inference of Deglacial Ocean Dynamics from the Spatial Pattern of LGM-to-Modern d13C and d18O Change

Abstract:

Estimates of the whole-ocean d13C change between the Last Glacial Maximum (LGM) and the modern-day are converging to values of about 0.4 per mil, and are of great use in partitioning land versus ocean contributions to the deglacial carbon cycle. To determine which specific oceanic processes are at play, however, knowledge of the spatial pattern of LGM-to-modern d13C and d18O change is critical. Spatial maps have mostly focused on Atlantic d13C, with less progress for d18O and the Pacific and Indian sectors, due to the concentration of sediment-core observations in the Atlantic and the difficulty in making meaningful maps from sparse data. Here, we demonstrate that a state estimation (or data assimilation) method based on recently compiled data and a simple kinematic ocean model simultaneously produces reasonable results for: 1) global maps of d13C and d18O, 2) uncertainty in the estimated properties, and 3) oceanic water-mass geometry. The observations include benthic d13C and d18O data from 493 marine sediment cores that were collected from the scientific literature and NOAA, PANGEA, and Delphi databases. The kinematic model permits each data point to have influence both up- and downstream along a water-mass pathway, typically allowing a larger geographical range than a statistical interpolation method. No assumption regarding the state of the circulation is necessary, and the modern-day circulation need not be assumed to be representative of the LGM. With this method, meridional (or other) sections can be compared between ocean basins. Furthermore, the internally-consistent d18O and d13C maps are used to determine the LGM-to-modern spatial changes that are robust given the uncertainty and sparsity of data. Rather than simply focus on property maps, we suggest that the link between observations and circulation changes (as reflected by the paths that water travels), points the way toward dynamical processes that must be explained. A particular application of our result is the geographic constraint of possible unobserved reservoirs of d13C or radiocarbon and calculation of their potential impact on the global chain of events during the deglaciation.