A coupled general circulation model for the Late Jurassic including fully interactive carbon cycling

The climatology of a coupled atmosphere - ocean (including sea ice) general circulation model for the Late Jurassic epoch (Kimmeridgian stage) is presented. The simulation framework used is the FAMOUS climate model [Jones et al, Climate Dynamics 25, 189-204 (2005)], which is a reduced resolution configuration of the UK Met Office model HadCM3 [Pope et al, Climate Dynamics 16, 123-46 (2000)]. In order to enable computation of carbon fluxes through the Earth System, fully interactive terrestrial and oceanic carbon cycle modules are added to FAMOUS. These include temporally evolving vegetation on land and populations of zooplankton, phytoplankton and nitrogenous nutrients in the ocean. The Kimmeridgian was a time of significantly enhanced carbon dioxide concentrations in the atmosphere (roughly four times preindustrial) and as such is a useful test bed for "paleocalibration" of a future climate perturbed by anthropogenic emissions of greenhouse gases [Barron et al, Paleoceanography 10 (5) 953-962 (1995) for example]. From a geological perspective, the Kimmeridgian was also a time of significant laying down of hydrocarbon reserves (particularly in the North Sea) and thus the inclusion of a fully interactive carbon cycle in FAMOUS enables the study of the dysoxic (low oxygen) and circulatory conditions relevant to their formation and preservation. The parameter space of both the terrestrial and oceanic carbon cycles was explored using the Latin Hypercube method [Mckay, Proceedings of the 24th conference on winter simulation, ACM Press, Arlington, Virginia, 57-564 (1992)], which enables efficient yet rigorous sampling of multiple covarying parameters. These parameters were validated using present day observations of meteorological, vegetative and biological parameters since the data available for the Jurassic itself is relatively scarce. To remove subjective bias in the validation process, the "Arcsine Mielke" skill score was used [Watterson, Int. J. Climatology, 16, 379-391 (1996)]. This metric uses a combination of spatial correlation and magnitude to unify geographical and numerical variations between simulated and observed fields.