An Overview of CMIP5 and the Experiment Design

Please download to get full document.

View again

of 14
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information Report



Views: 2 | Pages: 14

Extension: PDF | Download: 0

Related documents
An Overview of CMIP5 and the Experiment Design
  AFFILIATIONS:   T AYLOR  —Lawrence Livermore National Laboratory, Livermore, California; S TOUFFER  —NOAA/Geophysical Fluid Dynamics Laboratory, Princeton, New Jersey; M EEHL  — National Center for Atmospheric Research, Boulder, Colorado CORRESPONDING AUTHOR:  Karl E. Taylor, PCMDI, Lawrence Livermore National Laboratory, P.O. Box 808, L-103, Livermore, CA 94550E-mail: The abstract for this article can be found in this issue, following the table of contents. DOI:10.1175/BAMS-D-11-00094.1In final form 28 September 2011©2012 American Meteorological Society The fifth phase of the Climate Model Intercomparison Project (CMIP5), now underway, promises to produce a freely available state-of-the-art multimodel dataset designed to advance our knowledge of climate variability and climate change. A  t a September 2008 meeting involving 20 climate modeling groups from around the world, the World Climate Research Programme’s (WCRP) Working Group on Coupled Modelling (WGCM), with input from the International Geosphere–Biosphere Programme’s (IGBP) Analysis, Integration and Modeling of the Earth System (AIMES) project, agreed to promote a new set of coordinated climate model experiments. These ex-periments comprise the fifth phase of the Coupled Model Intercomparison Project (CMIP5). The WGCM’s endorsement of CMIP5 followed a plan-ning stage involving extensive community input (Meehl and Hibbard 2007; Hibbard et al. 2007) that led to a consensus proposal to perform a suite of climate simulations that focus on major gaps in understanding of past and future climate changes. CMIP5 will notably provide a multimodel context for 1) assessing the mechanisms responsible for model differences in poorly understood feedbacks associated with the carbon cycle and with clouds; 2) examining climate “predictability” and exploring the predictive capabilities of forecast systems on decadal time scales; and, more generally, 3) determining why similarly forced models produce a range of responses. It is ex-pected that some of the scientific questions that arose during preparation of the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4) will through CMIP5 be addressed in time for evaluation in the Fifth Assessment Report (AR5, scheduled for publication in late 2013). The enhanced set of historical and paleoclimate simulations and the expanded set of model output called for by CMIP5 promise to offer new opportunities for more detailed model evaluation. The four CMIP5 scenario runs, which provide a range of simulated climate futures (characterizing the next few decades to centuries), can be used as the basis for exploring climate change impacts and policy issues of considerable interest and relevance to society.CMIP5 builds on the successes of earlier phases of CMIP (see Meehl et al. 2000, 2005). In phase 3 (ca. 2004–present), for example, climate model output was for the first time released almost immediately upon completion of the runs so that scientists outside the AN OVERVIEW OF CMIP5 AND THE EXPERIMENT DESIGN BY  K ARL  E. T AYLOR , R ONALD  J. S TOUFFER , AND  G ERALD  A. M EEHL 485 APRIL 2012AMERICAN METEOROLOGICAL SOCIETY |  modeling groups could provide a more timely and comprehensive analysis of the results. This unprec-edented openness ushered in a “new era” in climate change research (Meehl et al. 2007). The CMIP3 multimodel dataset provided the basis for hundreds of peer-reviewed papers and played a prominent role in the IPCC’s AR4 assessment of climate variability and climate change. During phase 4 of CMIP (Meehl et al. 2007), additional simulations were performed that could be used to separate anthropogenic and natural influences on twentieth-century climate.The ongoing CMIP activities are organized by the WGCM, which represents the modeling groups. As part of the planning process, the WGCM received substantial input from potential users of the model output, some of whom are outside the traditional climate research community (e.g., scientists studying climate change impacts and policy makers). The experiments comprising CMIP5 were proposed, dis-cussed, and prioritized by climate modelers working closely with other climate scientists and the biogeo-chemistry community. Figure 1 shows the interna-tional organizations that have a formal interest in CMIP. The WCRP, through the WGCM, coordinates CMIP. The climate research based on CMIP is performed by a broad climate research community, and results of that research can inform major assessment activities, such as the ongoing IPCC process.The CMIP5 simulations were planned knowing that resource limitations would have to be care-fully considered. Clearly, not all possible experiments of interest could be included. Nevertheless, the integrated set of CMIP5 simu-lations attempt to address major priorities of several different com-munities and incorporates some of the ideas and suggestions of many individuals and from a number of workshops and meetings. 1  These workshops involved scientists with a wide range of interests, including climate modeling, biogeochemistry modeling, integrated assessment modeling, climate change impacts, climate analysis, climate processes, and climate observations.With input from these various groups, CMIP5 provides a framework for coordinated climate change experimentation that over the next several years (and well beyond the scheduled publication date of the IPCC AR5) promises to yield new insights about the climate system and the processes responsible for cli-mate change and variability. More than 20 modeling groups are performing CMIP5 simulations using more than 50 models. CMIP5 is not meant to be comprehensive or exclusive. Rather, various groups and interested parties are developing additional experiments that build on or augment the CMIP5 experiments. For example, the Coordinated Regional Downscaling Experiment (CORDEX), after applying a variety of methods, will produce high-resolution “downscaled” climate data based on the CMIP5 simulations (Jones et al. 2011; see also http://wcrp ). An entirely different group of scientists plans to carry out a set of Geoengineering Model Intercomparison Project (GeoMIP) experiments (Kravitz et al. 2011), which, F IG .  1. The relationship of CMIP5 to organizations established to coordinate climate research activities internationally and to the IPCC, the modeling centers, and the climate research community. 1 Notable contributions came from an Aspen Global Change Institute workshop (July 2006), a joint WGCM–AIMES meeting (September 2006), a Snowmass Energy Modeling Forum (July 2007), an IPCC Expert Meeting on New Scenarios (Noordwijkerhout, the Netherlands, in September 2007), an International Detection and Attribution Group (IDAG) meeting (Boulder, Colorado, in January 2008), WGCM meetings (Hamburg, Germany, in September 2007; Paris, France, in September 2008), a Working Group on Numerical Experimentation (WGNE) meeting (Montreal, Quebec, Canada, in November 2008), and individuals who have commented on various versions of Taylor et al. (2009). 486 APRIL 2012 |  building on two of the idealized CMIP5 experi-ments, explores the effects of possible geoengineering approaches to mitigate climate change.This paper is intended to provide an overview of CMIP5. The information from Taylor et al. (2009), which specifies the experiment design, is summa-rized and distilled in a form more suitable for a wide audience; the earlier document can be consulted by those seeking further details. This paper includes an introduction to the CMIP5 experiments, a description on how CMIP5 builds on and goes beyond the previ-ous phases of CMIP, information on how to access CMIP5 model output, an introductory discussion of issues relevant to the interpretation of CMIP5 results, and a summary. THE CMIP5 EXPERIMENTS.  The CMIP5 strat-egy (Hibbard et al. 2007; Meehl and Hibbard 2007) includes two types of climate change modeling experi-ments: 1) long-term (century time scale) integrations and 2) near-term integrations (10–30 yr), also called decadal prediction experiments (Meehl et al. 2009). The long-term integrations are usually started from multicentury preindustrial control (quasi equilibrium) integrations, whereas the decadal prediction experi-ments are initialized with observed ocean and sea ice conditions. Both the long- and near-term experi-ments are integrated using atmosphere–ocean global climate models (AOGCMs), the “standard” models used in previous CMIP phases. Earth system models of intermediate complexity (EMICs; Petoukhov et al. 2005) can also be used to perform many of the CMIP5 experiments. The AOGCMs and EMICs respond to specified, time-varying concentrations of various atmospheric constituents (e.g., greenhouse gases) and include an interactive representation of the atmo-sphere, ocean, land, and sea ice. For the long-term sim-ulations, some of the AOGCMs will, for the first time in CMIP, be coupled to biogeochemical components that account for the important fluxes of carbon be-tween the ocean, atmosphere, and terrestrial biosphere carbon reservoirs, thereby “closing” the carbon cycle in the model. These models are called Earth system models (ESMs), and they have the capability of using time-evolving emissions of constituents from which concentrations can be computed interactively. They may in some cases also include interactive prognostic aerosol, chemistry, and dynamical vegetation com-ponents. Individual groups may choose to perform either the long-term or the near-term experiments with either AOGCMs or ESMs, or they may do various combinations of each. Application of the EMICs will be limited to the long-term experiments.CMIP5 also recognizes that some groups may wish to perform simulations with a higher resolution or a more complete treatment of atmospheric chemistry than is typical of AOGCMs or ESMs. In these models, computer resources may be insufficient to allow fully coupled simulations, so CMIP5 includes an option to perform so-called time-slice integrations of both the present-day climate [essentially following the Atmo-spheric Model Intercomparison Project (AMIP) pro-tocol first described by Gates (1992)] and the future climate (in particular, the decade 2026–35, which allows for direct comparison with the fully coupled experiments). In time-slice simulations of the future, projected changes in sea surface temperature (SST) and sea ice are obtained from a prior integration of a fully coupled AOGCM simulation. In comparison with previous CMIP phases, the time-slice option allows a wider range of modeling groups to participate in CMIP5 (e.g., in weather forecast centers). Some groups may choose to perform additional time-slice experiments for other periods (e.g., late twenty-first century).The long-term experiments directly build on the CMIP3 experiments but include additional runs to provide a more complete understanding of climate change and variability. The near-term prediction experiments, in contrast, are an entirely new addition to CMIP and therefore are considered more explor-atory in nature. In these simulations, the models will not only respond, as in the long-term runs, to climate forcing (e.g., increasing atmospheric CO 2  concentra-tion) but also potentially track to some degree the actual trajectory of climate change, including (within the currently unknown predictability limits of the climate system) the unforced component of climate evolution. Thus, in the near-term experiments CMIP5 models, as part of a forecast system, will attempt a full prediction of climate change, whereas in the long-term experiments the models will provide a projection of the “forced” responses of climate to changing atmospheric composition and land cover. In these long-term projections, the climate change will be obscured to some degree by natural “unforced”  variability that only rarely and by coincidence could be expected to match the observable, evolving climate trajectory.Because of the large numbers of simulations included in the CMIP5 framework, the integrations for both century and decadal time scales are divided (based in part on consensus prioritization) into a “core” set, and then one or two surrounding “tiers” (Figs. 2 and 3). Although a group may choose to perform only the long-term core or only the near-term core 487 APRIL 2012AMERICAN METEOROLOGICAL SOCIETY |  of experiments, in each case the complete set of core simulations is expected to be completed. The intent is to generate a sufficiently large set of runs to enable a systematic model intercomparison within each type of experiment and thereby produce a credible multimodel dataset for analysis. The core experiments (located in the innermost circle and shaded pink in Figs. 2 and 3) are critical for evaluating the models, and they provide high-interest information about future climate change as well as help identify reasons for differences in the projections. The tier 1 integrations (surrounding the core and shaded yellow) examine specific aspects of climate model forcing, response, and processes, and tier 2 integrations (shaded green) go deeper into those aspects. Thus, proceeding from core to tier 1 to tier 2 can be seen as a progression from basic to more specialized simulations, exploring multiple aspects of climate system projections and responses. For each suite of experiments, the modeling groups will per-form the core integrations first, followed by a selection of the tier 1 and tier 2 integrations, depending on their interests and available resources.For detailed specifications of all the experiments, the reader is referred to the experiment design document (Taylor et al. 2009), which can be obtained from the CMIP5 website ( /cmip5 ). Long-term experiments.  The core simulations within the suite of CMIP5 long-term experiments (Fig. 2) include an AMIP run, a coupled control run, and a “historical” run forced by observed atmospheric composition changes (reflecting both anthropogenic and natural sources) and, for the first time, including time-evolving land cover. The historical runs cover much of the industrial period (from the midnineteenth century to near present) and are sometimes referred to as “twentieth century” simulations. Within the core set of runs, there are also two future projection simula-tions forced with specified concentrations [referred to as “representative concentration pathways” (RCPs)], consistent with a high emissions scenario (RCP8.5) and a midrange mitigation emissions scenario (RCP4.5). For AOGCMs and EMICs that have been coupled to a carbon cycle model (i.e., for ESMs), there are control and historical simulations, and the high emissions scenario (RCP8.5). For this set of ESM runs, the time-evolving atmospheric concentration of CO 2 , rather than being specified, is calculated by the model.The CMIP5 projections of climate change are driven by concentration or emission scenarios con-sistent with the RCPs described in Moss et al. (2010). In contrast to the scenarios described in the IPCC “Special Report on Emissions Scenarios” (SRES) used for CMIP3, which did not include policy intervention, F IG .  2. Schematic summary of CMIP5 long-term experi-ments with tier 1 and tier 2 experiments organized around a central core. Green font indicates simulations to be performed only by models with carbon cycle representations. Experiments in the upper hemisphere are suitable either for comparison with observations or provide projections, whereas those in the lower hemisphere are either idealized or diagnostic in nature and aim to provide better understanding of the climate system and model behavior. F IG .  3. Schematic summary of CMIP5 decadal predic-tion integrations. 488 APRIL 2012 |  the RCPs are mitigation scenarios that assume policy actions will be taken to achieve certain emission targets. For CMIP5, four RCPs have been formulated that are based on a range of projections of future population growth, technological development, and societal responses. The labels for the RCPs provide a rough estimate of the radiative forcing in the year 2100 (relative to preindustrial conditions). For example, the radiative forcing in RCP8.5 increases throughout the twenty-first century before reaching a level of about 8.5 W m −2  at the end of the century. In addition to this “high” scenario, there are two intermediate scenarios, RCP4.5 and RCP6, and a low so-called peak-and-decay scenario, RCP2.6, in which radiative forcing reaches a maximum near the middle of the twenty-first century before decreasing to an eventual nominal level of 2.6 W m −2 .For the diagnostic core integrations (in the lower hemisphere of Fig. 2), CMIP5 calls for 1) calibration-type runs to diagnose a specific transient climate response (defined as the globally averaged tempera-ture change at the time of CO 2  doubling in a 1% yr − 1  CO 2  increase experiment; 2) an abrupt CO 2  increase experiment to estimate the equilibrium global mean temperature response to a quadrupling of CO 2  and to quantify both radiative forcing and some of the important feedbacks; and 3) fixed SST experiments to refine the estimates of forcing and help interpret differences in model response.The tier 1 and tier 2 experiments explore various aspects of the core experiments in further detail. For ESMs, there are two carbon cycle feedback experiments. In the first, climate change is suppressed (by specifying in all radiation code calculations a constant, preindustrial CO 2  concentration), so that the carbon cycle response only reflects changing CO 2  influences unrelated to climate change. In the second, the climate responds to CO 2  increases, but the CO 2  increase is hidden from the carbon cycle. Following an approach found useful in the Coupled Climate–Carbon Cycle Climate Model Intercomparison Project (C4MIP; Friedlingstein et al. 2006), the surface fluxes of CO 2  will be saved in these experiments and then compared with fluxes from the corresponding core experiment (in which the carbon cycle simultane-ously responds to both climate and CO 2  concentra-tion changes). From these fluxes, the strength of the carbon–climate feedback can be expressed in terms of a difference in allowable emissions or in airborne fraction.Some experiments included in CMIP5 were srcinally conceived as part of other model intercom-parison projects. These include the Cloud Feedback Model Intercomparison Project (CFMIP; ; see also Bony et al. 2011), the Paleoclimate Modelling Intercomparison Project (PMIP; ; see also Braconnot et al. 2011), and earlier CMIP experiments. Thus, in CMIP5 there is a suite of cloud feedback experi-ments, some paleoclimate experiments to study the response of the models under much different forcing, experiments for climate change detection/attribution studies with only natural forcing or only greenhouse gas (GHG) forcing (as well as some single-forcing experiments), twenty-first century runs with the other two RCPs (RCP2.6 and RCP6), and extensions of the future climate simulations out to year 2300. These twenty-second- and twenty-third-century por-tions of the projections differ from the twenty-first century segments in that the RCPs were extended without reference to specific underlying societal, technological, or population scenarios (Moss et al. 2010). There are also diagnostic experiments with ab-breviated abrupt 4XCO 2  integrations that should yield more accurate estimates of adjusted CO 2  forcing, an experiment to quantify the magnitude of the aerosol forcing, and a placeholder for an experiment focusing on atmospheric chemistry and climate [Atmospheric Chemistry and Climate Activity 4 (“AC&C4”)].Several of the CMIP5 experiments require speci-fication of concentrations or emissions of various atmospheric constituents (e.g., greenhouse gases and aerosols). The Integrated Assessment Modeling Consortium working with the AC&C community has provided the concentrations, emissions, and time-evolving land use changes that will be prescribed in some of the CMIP5 experiments (e.g., Lamarque et al. 2010). Near-term experiments (decadal prediction).  The near-term experiments have been formally organized through a new collaboration between the WGCM and the Working Group on Seasonal to Interannual Prediction (WGSIP). There are two sets of core near-term integrations, as indicated by Fig. 3. The first is a set of 10-yr hindcasts initialized from observed climate states near the years 1960, 1965, and every 5 yr to 2005. In these 10-yr simulations, it will be possible to assess the skill of the forecast system in predicting climate statistics for times when the initial climate state may exert some detectable influence. The other core integrations extend the 10-yr simulations initial-ized in 1960, 1980, and 2005 by an additional 20 yr, ending up with two 30-yr hindcasts, and one 30-yr prediction to the year 2035. At this somewhat longer time scale, the external forcing from increasing GHGs 489 APRIL 2012AMERICAN METEOROLOGICAL SOCIETY |
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!