Session: Cross-cutting Group 2 (Paleovar, Past to future, Data assimilation)
Author: Peter Hopcroft / email@example.com / University of Bristol
Co-author: Alistair Sellar, Met Office, UK;
Dan Lunt, University of Bristol;
Paul Valdes, University of Bristol;
Anna Harper, University of Exeter;
Emma Stone, University of Bristol;
Paleoclimate simulations are usually performed using coupled general circulation models or Earth System models that have released to the academic community. This can often mean that the performance of a given model for a given palaeo-epoch does not have much influence on the development of that model, and does not therefore influence decisions about model structure, parameterisations or tuning. Here we report work on palaeoclimate simulations of the last glacial maximum and mid-Holocene that have been performed in parallel with the development of the United Kingdom Earth System Model (UKESM). This model will form a major part of the UK’s contribution to CMIP6 and is being developed jointly by the UK Met Office and UK academic community.
We present the first palaeoclimate simulations with the UK Met Office’s Global Atmosphere model version 7 (GA7) in an Earth System configuration that includes dynamic vegetation, aerosols and chemistry. GA7 is a 3D semi-Lagrangian, non-hydrostatic, fully compressible general circulation model with a horizontal resolution of 1.875x1.25 degrees and with 85 vertical levels, 35 of which are in the stratosphere. GA7 includes a prognostic rather than diagnostic cloud scheme, and a new dynamical core which has improved the simulation of climate variability. A size-resolving aerosol scheme allows for a more realistic simulation of the indirect aerosol-cloud interactions. The land surface scheme (JULES) uses 9 natural plant functional types and simulates vegetation interactively using an updated version of TRIFFID.
In comparison with the most recent pollen-based LGM vegetation reconstructions for northern Eurasia, GA7 significantly overestimates bare soil coverage. Results from a parallel 100-member perturbed parameter ensemble of land surface-only LGM simulations together with a statistical emulator, allow us to identify underlying reasons for this bias and to make suitable parameter changes that lead to improved vegetation coverage, thereby potentially improving the fully coupled Earth System model. Together with previous work using an older version of the UK Met Office Earth System model (HadGEM2-ES), vegetation coverage in mid- to high-latitudes at the LGM appears to provide effective bounds on the temperature dependence of vegetation productivity, and the associated surface feedbacks from snow cover.