Joint age-proxy system modelling

In the development of most paleoclimate records, there are two fundamental pursuits: 1) understanding what happened, and 2) understanding when it happened. Typically, these are separate efforts that rely on independent sets of observations. However, in many systems, especially those where flux is used to infer climatic or environmental variability, the evidence for climatic change and the evidence for geochronology are interdependent. By considering the age-proxy system as a single problem, there is considerable potential for an improved understanding of both geochronology and climate.

Here we present initial results on joint age-proxy system modelling at Lake Bosumtwi, Ghana (6.5°N, 1.4°W). Lake Bosumtwi is a small (8 km diameter), 1.07 Ma meteor impact crater lake. Geochronology at Lake Bosumtwi is well constrained for the past 50 kyr by discontinuous varved sections and more than 100 14C ages. Beyond the limit of radiocarbon, age control is limited to handful of optically stimulated luminescence, U-series, and a fission-track estimates of the impact date, resulting in gaps between age control of up to 500 kyr. Lake Bosumtwi is particularly well-suited to joint age-proxy system modelling because 1) the geometry of the basin makes sedimentation rates in the lake extremely sensitive to lake level, changing by several orders of magnitude from lake highstands to lowstands, and 2) a combination of age-dependent (mass accumulation) and age-independent (elemental abundance, total organic carbon, wet bulk density) indicators of lake level. We jointly solve for age and lake level over the past 1 Myr at Lake Bosumtwi by iteratively searching for modeled histories of sedimentation rate and lake level that are internally consistent and agree optimally with independent estimates of sediment age and relative lake level. Initial results indicate that the ensemble of joint solutions greatly reduces both age and lake level uncertainty, especially in the intervals of poor age control.