Quantitative evaluation of the roles of ocean chemistry and climate on ooid size across the Phanerozoic: Global versus local controls

Ooid size in natural systems can be predicted based on the expected equilibrium between precipitation and abrasion rates. Therefore, ooid size distributions may enable quantitative inferences about past ocean and climate conditions from sedimentology that complement insights from geochemistry. A few previous attempts have been made to compile information on ooid sizes across the Phanerozoic, but no systematic ooid size dataset is available and equilibrium models of ooid growth have not been explored comprehensively. In this study, convolutional neural network-based segmentation was used to automate ooid size measurement and provide a systematic sampling of ooid sizes spanning the Phanerozoic. This data set is coupled with Monte Carlo simulations to quantitatively explore the interplay between three testable physicochemical parameters (long-term average carbonate saturation state, bed shear velocity and intermittency of transport) in determining equilibrium ooid size. The numerical model shows that a typical-sized ooid (<= 1000 mu m) can grow under a wide range of parameter combinations whereas giant ooids (>2000 mu m) can only form under a more restricted set of circumstances. Furthermore, the model predicts that the formation of giant ooids is common when conditions favourable for rapid CaCO3 precipitation are achieved. This finding is consistent with the naturally rare occurrences of giant ooids, which developed under exceptional environmental conditions and are primarily associated with intervals of bimineralic seas and elevated sea-surface temperature (greenhouse or transitional climate). Whereas global controls could help explain secular trends in ooid size, significant spatial variability of ooid size in several time intervals further demonstrates the importance of local environmental parameters in the growth of ooids. Together, spatial and temporal variation in ooid size distributions can complement geochemical proxies in quantitatively reconstructing ancient tropical marine environments.

Automated summary

This study uses a convolutional neural network-based segmentation method to measure ooid sizes and provides a systematic sampling of ooid sizes throughout the Phanerozoic. The numerical model shows that typical-sized ooids can grow under a wide range of parameter combinations, while giant ooids can only form under specific conditions. The study also finds that the formation of giant ooids is associated with conditions favorable for rapid CaCO3 precipitation.