120 likes | 229 Views
A tasty new flavour of the GENIE earth-system model, applied to paleoclimates. D. Lunt (1) , A. Yool (2) , R.Marsh (2) , P.Valdes (1) , and the GENIE team. Bristol Research Initiative for the Dynamic Global Environment (BRIDGE), Bristol University, UK Southampton Oceanographic Centre, UK.
E N D
A tasty new flavour of the GENIE earth-system model, applied to paleoclimates D. Lunt (1), A. Yool (2), R.Marsh(2), P.Valdes(1) , and the GENIE team • Bristol Research Initiative for the Dynamic Global Environment (BRIDGE), Bristol University, UK • Southampton Oceanographic Centre, UK • Introduction and Motivation • The GENIE modelling framework • Results • Tuning the IGCM atmosphere • Ensemble of Last Glacial Maximum simulations • Moisture flux experiments with Goldstein • 4) Conclusions and future work d.j.lunt@bristol.ac.uk www.genie.ac.uk
RESULTS(1) – Tuning the IGCM atmosphere Methodology: (1) Identify tuneable parameters (29) . (2) Carry out an initial control simulation. (3) For each of the 29 tuneable parameters, carry out 4 simulations, 2+ve pertubations, 2-ve pertubations. (4) Calculate a skill score for every simulation (based on mean error cf NCEP temp. and precip., DJF and JJA). (5) Create a new control simulation by applying the best 5 pertubations to the previous control. (6) Repeat from (2)
RESULTS(1) – Tuning the IGCM atmosphere Results: Most improving parameters: 1st iteration: humidity at which large scale-clouds form. 2nd iteration: bulk aerodynamic coefficient for L,H Other improving parameters: e.g. roughness length of ocean, typical timescale for convection. Also, turning off the convection scheme improves precip!
RESULTS(1) – Tuning the IGCM atmosphere IGCM untuned IGCM tuned DJF surface temperature - NCEP 11% DJF 17% JJA HadAM3
RESULTS(1) – Tuning the IGCM atmosphere IGCM untuned IGCM tuned JJA precip - NCEP 0.1% DJF 10% JJA HadAM3
RESULTS(2) – Ensemble of Last Glacial Maximum simulations • Is the LGM or the modern climate more sensitive to uncertainties in internal model parameters? • 59-member ensemble of LGM and modern simulation pairs. Each of 58 pairs represent a +ve or –ve anomaly of one of 29 parameters. Plus control pair. • PMIP boundary conditions except ICE5G ice-sheets for LGM. • ‘Relative sensitivity’, RS, parameters calculated for each pair RS=log( abs( (LGM-LGMcont)/(PD-PDcont) ) ) RS +ve means LGM more sensitive than modern RS –ve means modern more sensitive than LGM RS calculated for 28 parameters, NH/SH DJF/JJA temp/precip +ve/-ve = 448 values.
RESULTS(2) – Ensemble of Last Glacial Maximum simulations 212 modern vs. 162 LGM. In particular, SH winter temperatures.
RESULTS(3) – Moisture flux experiments with Goldstein Decreasing flux, Inreasing flux Lohman (LGM) Oort (modern) IGCM (LGM) Miller+Rusell (LGM)
FUTURE WORK • To complete the coupling of the IGCM atmosphere to Goldstein ocean to MOSES/TRIFFID land surface to GLIMMER ice-sheet. • Integrate Tiedtke convection scheme into IGCM. • Tune the modern model components with EnKF scheme. • Long-term transient simulations (glacial/interglacial cycle) • Short-timescale experiments, e.g. 8.2 kyrBP holocene cooling event.
CONCLUSIONS • IGCM has been tuned with sledgehammer approach. Temperature responds well, precip not much improved – need to replace with new scheme. • Modern atmosphere is more sensitive to internal model parameters than the LGM. • Goldstein ocean sensitive to LGM Atlantic-Pacific freshwater flux. Coupling to IGCM will allow an investigation of the ocean’s response to interannual, decadal, and centennial variations in this flux. • Many experiments/investigations to be carried out soon!