Dongyu Zheng

The Toarcian Oceanic Anoxic Event (T-OAE, ~183 Ma) was a major hyperthermal event in the Phanerozoic, characterised by massive carbon release and widespread organic carbon burial. However, the source(s), flux, and sequestration pathways of carbon remain uncertain. Here, we assimilate globally representative carbon isotope and atmospheric CO ₂ records into an Earth system model to decipher the carbon cycle during the T-OAE. Our simulations indicate the release of ~18,000 Pg C of predominantly magmatic carbon through the T-OAE, though with an input of carbon from thermogenic and/or biogenic sources at the onset of the event. We show that at least 6800 Pg of organic carbon was buried during the event, which, combined with silicate weathering-driven excess carbon consumption (~6000 Pg C), helped the Earth’s climate system to recover from large-scale carbon release. Our results provide quantitative constraints on the tempo, magnitude, and recovery mechanisms of carbon-cycle perturbations during extreme greenhouse intervals.

The Carboniferous presents a period of dynamic changes in climate, tectonics and ecology. Plant evolution and expansion during this time allowed for large-scale coal deposit formation and enhanced biological input within biogeochemical cycles which exerted a stronger influence on Earth's climate. Using fossil data and vegetation–climate modelling, we explore the gradual changes of vegetation as they change from globally uniform ecosystems to fragmented and distinct provinces. This paper also discusses the potential missing link between vegetation strategies and localized environmental conditions, and climate modelling over the Carboniferous, which may explain the increase in organic sediment and coal forest extent despite lower global temperatures. As changes in palaeogeography and palaeoclimate created drier and more seasonal environments, the trajectory of plant evolution favoured seasonality- and drought-tolerant plants, bringing us closer to present-day floral distributions.

Knowing the past temperature of the Earth is crucial for understanding the mechanisms driving climate change and biosphere evolution, but there is significant debate about the range of past temperature variation. Previous interpretations, largely based on oxygen isotope records, suggest that global temperature has generally declined over the past 539 million years, but substantial uncertainties persist. In this study, we introduce an independent estimate for long-term Phanerozoic temperature trends based on a large database of chemical weathering indices from siliciclastic sedimentary rocks, globally upscaled using a state-of-the-art general circulation paleoclimate model. Our results imply that Phanerozoic global temperatures remained within 10-30 °C, and that Paleozoic oceans had comparable temperatures to Mesozoic and Cenozoic oceans, in contrast to previous work suggesting that they were anomalously hot. This finding supports the idea that negative feedback processes, such as silicate weathering, have maintained long-term global average temperatures within a relatively tight range, contributing to the continued long-term evolution of the biosphere.

Identification of complex geochemical signatures of basalts from various tectonic settings using trace elements contents is still challenging due to uncertainties in existing classification diagrams and machine‐learning attempts. To address this, we trained a machine‐learning model using a random forest classifier on trace element concentrations of a global basalt‐dataset, categorized by types of tectonic plate boundary—destructive or constructive—at which they formed. Achieving an accuracy exceeding 98%, the model efficiently extracts the distinctive characteristics from each basalt type. When applied to the Bangong‐Nujiang suture zone in central Tibet, the model reveals that the basalts exhibited features of both boundary types prior to 108–107 Ma before transitioning to solely destructive characteristics. This transition is likely to be caused by the detachment of a descending oceanic slab, aligning with existing geological evidence. This case study highlights the promising potential of machine learning models, trained on simplified basalt types, in more accurately tracing lithospheric evolution.

Databases of 3D paleoclimate model simulations are increasingly used within global biogeochemical models for the Phanerozoic Eon. This improves the accuracy of the surface processes within the biogeochemical models, but the approach is limited by the availability of large numbers of paleoclimate simulations at different pCO₂ levels and for different continental configurations. In this paper we apply the Frame Interpolation for Large Motion (FILM) deep learning method to a set of Phanerozoic paleoclimate model simulations to upscale their time resolution from one model run every ∼25 million years to one model run every 1 million years (Myr). Testing the method on a 5 Myr time-resolution set of continental configurations and paleoclimates confirms the accuracy of our approach when reconstructing intermediate frames from configurations separated by up to 40 Myr. We then apply the method to upscale the paleoclimate data structure in the SCION climate-biogeochemical model. The interpolated surface temperature and runoff are reasonable and present a logical progression between the original key frames. When updated to use the high-time-resolution climate data structure, the SCION model predicts climate shifts that were not present in the original model outputs due to its previous use of widely spaced datasets and simple linear interpolation. We conclude that a time resolution of ∼10 Myr in Phanerozoic paleoclimate simulations is likely sufficient for investigating the long-term carbon cycle and that deep learning methods may be critical in attaining this time resolution at reasonable computational expense, as well as for developing new fully continuous methods in which 3D continental processes are able to translate over a moving continental surface in deep time. However, the efficacy of deep learning methods in interpolating runoff data, compared to that of paleogeography and temperature, is diminished by the heterogeneous distribution of runoff. Consequently, interpolated climates must be confirmed by running a paleoclimate model if scientific conclusions are to be based directly on them.

Databases of 3D paleoclimate model simulations are increasingly used within global biogeochemical models for the Phanerozoic Eon. This improves the accuracy of the surface processes within the biogeochemical models, but the approach is limited by the availability of large numbers of paleoclimate simulations at different pCO₂ levels and for different continental configurations. In this paper we apply the Frame Interpolation for Large Motion (FILM) Deep Learning method to a set of paleoclimate model simulations to upscale their time resolution from one model run every ~25 million years to one model run every 1 million year (Myr). Testing the method on a 5 Myr time-resolution set of continental configurations confirms the accuracy of our approach when reconstructing intermediate frames from configurations separated by up to 40 Myrs. We then apply the method to upscale the paleoclimate datastructure in the SCION climate-biogeochemical model and demonstrate that upscaled outputs for global distributions of surface temperature and runoff follow a logical progression between the original keyframes. When updated to use the high-time-resolution climate datastructure, the SCION model predicts climate shifts that were not present in the original model outputs due to its previous use of wide-spaced datasets and simple linear interpolation. We conclude that a time resolution of ~10 Myr in paleoclimate simulations is likely sufficient for investigating the long-term carbon cycle, and that Deep Learning methods may be critical in attaining this time-resolution at a reasonable computational expense, as well as for developing new fully-continuous methods in which 3D continental processes—such as species distribution—are able to translate over a moving continental surface in deep time. Nonetheless, the efficacy of Deep Learning methods in interpolating runoff data, compared to that of paleogeography and temperature, is diminished by the heterogeneous distribution of runoff. Consequently, interpolated climates should be confirmed by running a paleoclimate model for any sound scientific conclusions.