Abstract. We construct a carbon cycle box model to process observed or inferred
geochemical evidence from modern and paleo settings. The [simple carbon
project] model v1.0 (SCP-M) combines a modern understanding of the ocean
circulation regime with the Earth's carbon cycle. SCP-M estimates the
concentrations of a range of elements within the carbon cycle by simulating
ocean circulation, biological, chemical, atmospheric and terrestrial
carbon cycle processes. The model is capable of reproducing both paleo and
modern observations and aligns with CMIP5 model projections. SCP-M's fast
run time, simplified layout and matrix structure render it a flexible and
easy-to-use tool for paleo and modern carbon cycle simulations. The ease of
data integration also enables model–data optimisations. Limitations of the
model include the prescription of many fluxes and an ocean-basin-averaged
topology, which may not be applicable to more detailed simulations. In this paper we demonstrate SCP-M's application primarily with an analysis of
the carbon cycle transition from the Last Glacial Maximum (LGM) to the Holocene and also
with the modern carbon cycle under the influence of anthropogenic CO2
emissions. We conduct an atmospheric and ocean multi-proxy model–data
parameter optimisation for the LGM and late Holocene periods using the
growing pool of published paleo atmosphere and ocean data for CO2,
δ13C, Δ14C and the carbonate ion proxy. The results
provide strong evidence for an ocean-wide physical mechanism to deliver the
LGM-to-Holocene carbon cycle transition. Alongside ancillary changes in ocean
temperature, volume, salinity, sea-ice cover and atmospheric radiocarbon
production rate, changes in global overturning circulation and, to a lesser
extent, Atlantic meridional overturning circulation can drive the observed
LGM and late Holocene signals in atmospheric CO2, δ13C,
Δ14C, and the oceanic distribution of δ13C,
Δ14C and the carbonate ion proxy. Further work is needed on the analysis
and processing of ocean proxy data to improve confidence in these
modelling results.