Abstract
Over geological timescales, CO2 levels are determined by the operation of the long term carbon cycle, and it is generally thought that changes in atmospheric CO2 concentration have controlled variations in Earth's surface temperature over the Phanerozoic Eon. Here we compile independent estimates for global average surface temperature and atmospheric CO2 concentration, and compare these to the predictions of box models of the long term carbon cycle COPSE and GEOCARBSULF.We find a strong relationship between CO2 forcing and temperature from the proxy data, for times where data is available, and we find that current published models reproduce many aspects of CO2 change, but compare poorly to temperature estimates. Models are then modified in line with recent advances in understanding the tectonic controls on carbon cycle source and sink processes, with these changes constrained by modelling 87Sr/86Sr ratios. We estimate CO2 degassing rates from the lengths of subduction zones and rifts, add differential effects of erosion rates on the weathering of silicates and carbonates, and revise the relationship between global average temperature changes and the temperature change in key weathering zones.Under these modifications, models produce combined records of CO2 and temperature change that are reasonably in line with geological and geochemical proxies (e.g. central model predictions are within the proxy windows for >~75% of the time covered by data). However, whilst broad long-term changes are reconstructed, the models still do not adequately predict the timing of glacial periods. We show that the 87Sr/86Sr record is largely influenced by the weathering contributions of different lithologies, and is strongly controlled by erosion rates, rather than being a good indicator of overall silicate chemical weathering rates. We also confirm that a combination of increasing erosion rates and decreasing degassing rates over the Neogene can cause the observed cooling and Sr isotope changes without requiring an overall increase in silicate weathering rates.On the question of a source or sink dominated carbon cycle, we find that neither alone can adequately reconstruct the combination of CO2, temperature and strontium isotope dynamics over Phanerozoic time, necessitating a combination of changes to sources and sinks. Further progress in this field relies on >108 year dynamic spatial reconstructions of ancient tectonics, paleogeography and hydrology. Whilst this is a significant challenge, the latest reconstruction techniques, proxy records and modelling advances make this an achievable target.
Highlights
Atmospheric carbon dioxide appears to have been essential in the maintenance of habitable conditions throughout Earth history by providing additional radiative forcing under a less luminous ancient sun
We test current Earth system box models against these constraints, and those provided by the geological strontium isotope record, which responds to changes in the carbon cycle
Uncertainty about the operation of the Neoproterozoic carbon cycle currently prevents the GEOCARBSULF model being run beyond the Phanerozoic, as the isotope mass balance approach to calculating organic fluxes fails when the input carbonate δ13C is below the mantle value of ~−5‰, such as during the enigmatic Shuram excursion (e.g. Rothman et al, 2003)
Summary
Atmospheric carbon dioxide appears to have been essential in the maintenance of habitable conditions throughout Earth history by providing additional radiative forcing under a less luminous ancient sun. The most recent Earth system box models are powerful predictive tools, used to reconstruct changes in global biogeochemistry and climate for times when proxy estimates are either unavailable or unreliable (Arvidson et al, 2013; Royer et al, 2014; Mills et al, 2016; Lenton et al, 2018; Krause et al, 2018), or used as a framework in which to test hypotheses about processes driving climate or biosphere changes over geological timescales In this paper we return to the core predictions of Earth system box models, for atmospheric CO2 and global surface temperature. These are among the most -testable predictions, as a wealth of pCO2 estimates exist (Royer, 2014) and during the Cenozoic at least, surface temperature proxies are able to produce signals beyond the inherent uncertainties and climatic noise (e.g. Zachos et al, 2001; Hansen et al, 2013). We modify current models to take into account recent revisions of tectonic forcings, revise the strength of model feedbacks, and extend predictions back into the Neoproterozoic
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
