|
Supplementary material to “Comparing Structurally Different Climate Models in a Paleoenvironmental Context”
24 May 2011
Alan M. Haywood, School of Earth and Environment, University of Leeds, Leeds, UK
Gilles Ramstein, Laboratoire des Sciences du Climat et de l'Environnement, Gif-sur-Yvette, France
Ayako Abe-Ouchi, University of Tokyo, Tokyo, Japan
Citation:
Haywood, A. M., G. Ramstein, and A. Abe-Ouchi (2011), Comparing structurally different climate models in a paleoenvironmental context, Eos Trans. AGU, 92(21), 180, doi:10.1029/2011EO210005. [Full Article (pdf)]
Paleoclimate Modelling Intercomparison Project Phase 3 Workshop; Kyoto, Japan, 5-10th December 2010.
Alan M. Haywood, School of Earth and Environment, University of Leeds, Leeds, UK; E-mail: earamh@leeds.ac.uk; Gilles Ramstein, Laboratoire des Sciences du Climat et l’Environnement, Gif-sur-Yvette, France; E-mail: gilles.ramstein@lsce.ipsl.fr; Ayako Abe-Ouchi, University of Tokyo, Tokyo, Japan, E-mail: abeouchi@aori.u-tokyo.ac.jp
The Palaeoclimate Modelling Intercomparison Project (PMIP), now in its third phase, co-ordinates a community of researchers who compare structurally different climate models in a palaeoenvironmental context. The project is endorsed by both WCRP/CLIVAR/WGCM and IGBP PAGES. Uncertainties highlighted in climate prediction involve aspects of the climate that have changed in response to different perturbations and on a wide range of time scales. PMIP’s mission is to assess whether these changes are consistent with current theories of the climate system. The project provides a unique opportunity to evaluate models using observed changes of large magnitude provided by Earth history.
At its recent workshop in Kyoto, sponsored by JSPS (Japan Society of Promoting Science), the University of Tokyo and JAMSTEC (Japan Agency For Marine-Earth Science and Technology), more than 100 scientists, including atmospheric scientists, oceanographers, and palaeoclimatologists from both the observational and modelling communities gathered to review the PMIP3 experimental design and progress on PMIP3 simulations, and to identify barriers to progress which could be overcome to ensure that results are published in time to contribute to the next Intergovernmental Panel on Climate Change Report.
An important aspect of PMIP3 discussed in detail at the meeting is benchmarking the three time periods are considered as part of the Phase 5 Coupled Modelling Intercomparison Project (CMIP5) within Tier 1 (the Last Glacial Maximum 21,000 years ago, and mid-Holocene 6,000 years ago) and Tier 2 (Last Millennium). Analyses of these simulations will address issues such as climate sensitivity, and how well current models reproduce the hydrological cycle and major feedbacks, as well as interannual to multi-decadal variability. PMIP will also be assessing carbon-cycle modeling, through its daughter project PCMIP (PaleoCarbon Modelling Intercomparison Project). PCMIP focuses on simulations of the Last Glacial Maximum and the Last Millennium.
PMIP has always been active in promoting large-scale data synthesis, and has been preparing both climate and carbon-cycle benchmark data sets in preparation for model evaluation within the framework of CMIP5. At the workshop, a new global data set of gridded quantitative climate reconstructions for the mid-Holocene and the Last Glacial Maximum was unveiled, as well as new global data sets of vegetation distribution, fire regimes and peatland carbon accumulation. These, and other large-scale data sets for model evaluation, will be made available via the PMIP website. The PMIP data synthesis group also identified “missing” data sets or ones that will require additional work in the coming year in order to be ready for AR5, the most vital of these being a synthesis of data documenting short-term (interannual to interdecadal) climate variability during the Last Glacial Maximum, the mid-Holocene and the last millennium. The PAGES2k initiative is expected to take the lead on the last-millennium synthesis.
CMIP5 simulations are not the only focus in PMIP. Snapshot-style and transient experiments will be run for the Holocene and Last Interglacial (~125,000 years ago) as well as snapshot-style simulations for the mid-Pliocene Warm Period (~ 3 million years ago) through strong links with a daughter project known as PlioMIP (Pliocene Modelling Intercomparison Project). PMIP3 also includes an ‘ensemble of opportunity’ for the greenhouse climates of the Eocene (~ 50 million years ago; EoMIP). These various experiments will allow PMIP3 to compare the differences between, and ability of, models to predict climate change driven by insolation (orbit driven e.g. mid-Holocene), low (LGM) intermediate (mid-Pliocene) and high (Eocene) CO2 forcing, as well as more rapid forcing mechanisms such as iceberg discharges and freshwater hosing (e.g. the 8.2 Kyr event and Heinrich Event H1). These experiments and model data comparisons will provide new estimates of forcing thresholds that influence polar amplification and the relationship between climate-ice-sheet and sea-level under different climate states. The transient experiments combining Earth System Models of Intermediate Complexity (EMICS) and Earth System Models (ESMs) represent an important step towards a better understanding of the dynamics and temporal response of the different components of the climate system.
In previous iterations of PMIP a single experimental design for each time interval was defined. This remains a central pillar of PMIP methodology but the meeting defined a number of additional modelling studies that will explore uncertainties derived from boundary condition forcing. This is perhaps best expressed within the newly published methodology for the Last Millennium and for the Last Glacial Maximum. This new expansion in methodology is consistent with PMIPs dual role as a means to 1) compare models and 2) to improve our basic understanding of past climates. Specific ideas to explore the uncertainty in boundary condition forcing included ice sheet reconstructions for the LGM and volcanic forcings for the Last Millennium.
Finally, workshop participants were active in trying to broaden our understanding of the total uncertainty in the palaeoenvironmental data used to evaluate models. These issues were broken down into three discreet challenges of error decomposition/robustness estimation and process understanding.