CMIP5 Coupled Model Intercomparison Project
WCRP    World Climate Research Programme

CMIP5 - Modeling Info - Setting up CMIP5 Experiment, Getting Started



This “Setting up CMIP5 Experiment, Getting Started” guide for CMIP5 modeling groups is a synopsis of information provided in A Summary of the CMIP5 Experimental Design by K.E. Taylor, R.J. Stouffer, and G.A. Meehl.

Note that the full collection of coordinated CMIP5 modeling experiments extends substantially beyond what is anticipated for the IPCC’s Fifth Assessment of Climate Change (AR5), scheduled for publication in 2013 (see IPCC/CMIP5 AR5 Timetable). At the most basic level, the CMIP5 experiments are divided between Decadal Hindcasts and Predictions and “long-term” climate simulations.


Decadal Hindcasts and Predictions climate experiments include

  • Decadal hind-casts and predictions by coupled ocean-atmosphere models initialized from observations (Table 1);
  • “Time-slice” experiments (Table 2) with specification of SSTs and sea ice that allow the use of atmosphere-only models.

Long-term (century-scale) climate experiments include

  • Baseline paleoclimatic, preindustrial (pre-1850), and historical (post-1850) climate simulations (Table 3);
  • Future projections of climates associated with different atmospheric greenhouse gas (GHG) scenarios specified by several “representative concentration pathways” (RCPs) (Table 4);
  • Coupled carbon-climate [i.e. “Earth System Model (ESM)”] simulations of historical and future climates (Table 5);
  • Experiments for diagnosing the model representations of selected climate mechanisms (Table 6);
  • Simulations designed to facilitate climate-change detection and attribution studies (Table 7).

Depending on available resources and scientific interests, a modeling group may choose to perform experiments in only one of these categories; but some groups may be able to implement at least subsets of both the short-term and long-term experiments. For either type of climate simulation, all participating modeling groups are expected to first perform specified “core” experiments. Then, depending on the group’s scientific priorities, model runs of supplemental “tier 1” experiments that examine specific aspect of the core experiments, and “tier 2” experiments that delve deeper into these aspects are also recommended.


Recommended climatic forcings


It is scientifically advisable that the CMIP5 experiments employ common climatic forcings and other boundary conditions (BCs) as much as feasible. Links to CMIP5-endorsed data sets are listed on an accompanying web page (Model Forcings web page) for model specifications of


  • Solar forcing
  • Ozone concentrations
  • GHG and other anthropogenic gas and aerosol concentrations/emissions
  • Historical land-use data
  • SST/sea ice boundary conditions etc. for Atmospheric Model Intercomparison Project (AMIP) and Cloud Forcing Model Intercomparison Project (CFMIP) protocols

These data sets cover (or will shortly be extended to cover) at least the period 1850-2100 relevant for the CMIP5 preindustrial, historical postindustrial, and future scenario experiments. (Note, however, the absence of recommended data sets for specification of historical concentrations of volcanic or other natural aerosols.)


Guidelines on the climatic forcings and other BCs, as well as the initial conditions (ICs) that are specific to each of the CMIP5 short-term or long-term experiments are as follows. These experiments are structured as in the Taylor et al. experimental design document, which should be consulted for further implementation details, as needed.



Table 1. Decadal prediction experiments

 

CORE

 

Decadal Prediction Experiments 1.1, 1.2

ICs: Initialize using modeler choices of observation-based data for oceans and atmosphere.

BCs: Specify forcings as in Expt 3.2 for those experiments that end in 2005, and as in Expt 4.1 for those experiments that extend beyond 2005 (e.g. Expt 1.1-I).

 

TIER 1

 

Decadal Prediction Experiments 1.1-E, 1.2-E, 1.1-I

ICs: Initialize using modeler choices of observation-based data for oceans and atmosphere.

BCs: Specify forcings as in Expt 3.2 for those experiments that end in 2005, and as in Expt 4.1 for those experiments that extend beyond 2005 (e.g. Expt 1.1-I).

 

Expt 3.3 (AMIP run: minimum length 1979 thru 2008)—see 3.3 in Table 3 below

 

Expt 3.1-S (Shortened preindustrial control run)—see 3.1-S in Table 3 below

 

Expt 6.1-S (Short idealized 1%/yr run)—see 6.1 in Table 6 below

 

Expt 1.3 (Hindcasts without volcanoes)

ICs: Initialize as in Expt 1.1 etc.

BCs: Specify forcings as in Expt 3.2 but without effects of volcanic aerosols.

 

Expt 1.4 (Predictions with 2010 Pinatubo-like eruption)

ICs: Initialize as in Expt 1.1 etc. near the end of 2005.

BCs: Specify forcings as near the end of Expt 3.2 through year 2005, and as in Expt 4.1 after 2005, but with Pinatubo volcanic forcing (from ~ year 1991 of Expt 3.2) added instead in year 2010.

 

Experiment 1.5 (Initialization with alternative strategies)

ICs: Initialize using modeler choices of observation-based data for oceans and atmosphere.

BCs: Specify forcings as in Expt 3.2 for those experiments that end in 2005, and as in Expt 4.1 for those experiments that extend beyond 2005 (e.g. Expt 1.1-I).

 

Expt 1.6 (Decadal prediction run with more complete atmospheric chemistry)

ICs: Initialize as in Expt 1.1 etc near the end of 2005.

BCs: Specify forcings as in Expt 3.2, but with more complete atmospheric chemistry (short-lived species and pollutants--to be supplied later).



Table 2. “Time-slice” experiments

 

TIER 1


Expt 3.3 (AMIP run: minimum length 1979 thru 2008)—see 3.3 in Table 3 below

 

Expt 2.1 (Future “time-slice” experiment, years 2026-2035)

ICs: Two options are allowed. In option 1, initialize the atmosphere and land as in year 2026 of Expt 4.1 (RCP4.5 scenario). In option 2, initialize the atmosphere and land as in year 2026 of a 30-year RCP4.5 scenario run starting in the year 2005 (i.e., the Expt 1.2 ensemble member initialized in 2005).

BCs: In option 1, specify SST and sea-ice concentration anomalies relative to their AMIP Expt 3.3 values according to the climatological differences of years 2026-2035 of the coupled ocean-atmosphere Expt 4.1 and years 1979-2008 of the coupled historical Expt 3.2. In option 2, specify SST and sea-ice concentration anomalies according to the climatological differences of years 2026-2035 of the selected Expt 1.2 ensemble member and a climatology for years 1979-2008 based on a subset of the years covered in the Expt 1.1 series of 10-year simulations. For both options, the total run length is 10 years.

 

TIER2


Expt 3.3-E (AMIP ensemble) --link to 3.3-E in Table 3 below

 

Expt 2.1-E (Future “time-slice experiment ensemble)

ICs: Follow one of the options of Expt 2.1, but with different ICs imposed on the atmosphere, sea ice, ocean, and possibly also the land, so as to generate an ensemble of at least 3 decadal-length runs.

BCs: Specify SST and sea-ice concentration anomalies as in option 1 or option 2 of Expt 2.1 to generate an ensemble of at least 3 decadal-length runs. The anomalies should be taken from more than one pair of coupled atmosphere-ocean model runs, if available.

 

Expt 6.5 (Cloud response to an imposed 4xCO2 [Hansen-style diagnosis of “fast” climate system responses])—see 6.5 in Table 6 below

 

Expt 6.6 (Cloud response to an imposed change in SST pattern)—see 6.6 in Table 6 below

 

Expt 6.7a (Aqua-planet: control run)—see 6.7a in Table 6 below

 

Expt 6.7b (Aqua-planet: cloud response to an imposed 4xCO2 [Hansen-style diagnosis])—see 6.7b in Table 6 below

 

Expt 6.7c (Aqua-planet: cloud response to an imposed uniform change in SST)—see 6.7c in Table 6 below

 

Expt 6.8 (Cloud response to an imposed uniform change in SST)—see 6.8 in Table 6 below.

 


Table 3. Baseline experiments

 

CORE


Expt 3.1 (Preindustrial control run)

ICs: Not mandated. It is recommended that the run start ~ 1750 (gaseous emissions/concentrations data begin in 1765) or ~ 1850 (beginning of ozone data) and continue for 500 years after completion of model spin-up.

BCs: Choose non-evolving pre-industrial forcings appropriate for the chosen start year.

Expt 3.1-S (Shortened preindustrial control run)

ICs: As in Expt 3.1.

BCs: As in Expt 3.1, but with the model running for only 100 years after spin-up.

 

3.2 (Historical run: minimum length 1850 thru 2005)

ICs: Initialize from the preindustrial control run. If starting this run ~ 1750 and the model’s RCP (future scenario) runs are to go out to year 2300, this mandates initializing from a point at least 550 years before the end of the control run; if starting this run at 1850 and the RCP runs are to go to year 2100, this mandates initializing from a point 250 years before the end of the control run.

BCs: As in Expt 3.1 but with BCs changing appropriately over the chosen time period (~1750-2005 or 1850-2005). If possible, dynamic vegetation (if it is a model feature) should be ‘turned off’.

Expt 3.3 (AMIP run: minimum length 1979 thru 2008)

ICs: Initialize the model atmosphere several months prior to January 1979 from some “reasonable” state. The method for initializing snow cover and soil moisture is not mandated.

BCs: As in Expt 3.2, but with SSTs and sea ice extents also imposed from PCMDI’s observation-based data sets.

 

TIER 1


Expt 3.2-E (Historical ensemble)

ICs: Same as Expt 3.2, except initialize the model from different points in the preindustrial control run.

BCs: Same as Expt 3.2.

Expt 3.3-E (AMIP ensemble)

ICs: Initialize the model at the same starting point as in Expt 3.3, but with different atmospheric states, and possibly also different land states, resulting in an ensemble of at least 3 members.

BCs: Same as in Expt 3.3.

 

Expts 3.4 (Mid-Holocene paleoclimatic run) and 3.5 (Last Glacial Maximum paleoclimatic run)

See the PMIP website (http://pmip.Isce.ipsl.fr/) for specification of ICs and BCs.

 

TIER 2


Expt 3.6 (Last Millenium paleoclimatic run)

See the PMIP website (http://pmip.Isce.ipsl.fr/) for specification of ICs and BCs.


Table 4. Future climate projections forced by RCP concentrations

 

CORE


Expts 4.1 and 4.2 (RCP4.5 and RCP8.5 over years 2006-2100)

ICs: Initialize from the end state of Expt 3.2.

BCs: Specified as in Expt 3.1, but changing appropriately over the years 2006-2100, and with volcanic and other natural aerosols zeroed. (Radiative forcing stabilizes at ~4.5 W m-2 after 2100 in Expt 4.1 and at ~ 8.5 W m-2 in Expt 4.2. If the model is an ESM that predicts CO2 emissions, save CO2 surface fluxes in order to calculate allowable emissions.)

 

TIER 1


Expts 4.3 and 4.4 (RCP2.X and RCP6 over years 2006-2100)

ICs: Initialize from the end state of Expt 3.2.

BCs: Specified as in Expt 3.1, but changing appropriately over years 2006-2100, and with volcanic and other natural aerosols zeroed. (Radiative forcing stabilizes at ~2 W m-2 near year 2100 in Expt 4.3 and at ~ 6 W m-2 after year 2100 in Expt 4.4. )

Expts 4.1-L (RCP4.5 extended through year 2300)

ICs: Initialize from end state of Expt 4.1 and run through year 2300.

BCs: Specified as in Expt 4.1, but changing appropriately over the extended time period.

 

TIER 2


Expts 4.2-L (RCP8.5 extended through year 2300)

ICs: Initialize from the end state of Expt 4.2 and run through year 2300.

BCs: As in Expt 4.2, but changing appropriately over the extended time period.

Expts 4.3-L (RCP2.X extended through year 2300)

ICs: Initialize from the end state of Expt 4.3 and run through year 2300

BCs: As in Expt 4.3, but changing appropriately over the extended time period.



Table 5. Earth System Model (ESM) experiments

 

CORE


Expt 5.1 (ESM pre-industrial control run)

ICs: As in Expt 3.1 (Pre-industrial control run), but with atmospheric CO2 concentrations determined by the model itself from emissions data at the starting year (either ~1765 or 1850).

BCs: As in Expt 3.1 (link), but with atmospheric CO2 concentrations determined by the model itself from the emissions data at the starting year (either ~1765 or 1850).

Expt 5.2 (ESM historical 1850-2005 run)

ICs: As in Expt 3.2 .

BCs: As in v, but prescribe evolving anthropogenic CO2 emissions, rather than concentrations.

Expt 5.3 (ESM RCP8.5 experiment 2006-2100)

ICs: Initialize from the end of Expt 5.2 and continue into the future to year 2100.

BCs: As in Expt 4.2, but prescribe evolving CO2 emissions (rather than concentrations) through year 2100.

 

TIER 1

Expt 5.4 (ESM carbon-climate feedback experiment—increasing CO2 without physical climate change)

ICs: Two options are allowed. If choosing option 1, spin off from the preindustrial control Expt 3.1 at the same point as in Expt 6.1. In option 2, spin off from Expt 3.1 at the same point as Expt 3.2.

BCs: In option 1, impose the same constant atmospheric CO2 concentrations as in Expt 3.1 on the radiation code, but allow the carbon cycle to respond to a prescribed 1% per year increase of atmospheric CO2 (as in Expt 6.1), and run for a total of 140 years. In option 2, treat the radiation code as in option 1, but allow the carbon cycle to respond to increasing historical atmospheric CO2 concentrations (as in Expt 3.2), and to increasing future concentrations (as in the Expt 4.1 RCP4.5 scenario), for a total run of 251 years.

 

TIER 2


Expt 5.5 (ESM carbon-climate feedback experiment—physical climate change without increasing CO2)

ICs: The same two options as in Expt 5.4 are allowed.

BCs: In option 1, impose the constant atmospheric CO2 concentrations (as in the preindustrial control Expt 3.1) on the carbon cycle, but allow the radiation code to respond to a 1% per year increase of atmospheric CO2 (as in Expt 6.1), and run for a total of 140 years. In option 2, treat the carbon cycle as in option 1, but allow the radiation code to respond to increasing historical atmospheric CO2 concentrations (as in Expt 3.2), and to increasing future concentrations (as in Expt 4.1), for a total run of 251 years.




Table 6. Diagnostic experiments

 

CORE


Expt 6.1 (Idealized 1%/yr run to 4xCO2)

ICs: Initialize from a point in Expt 3.1.

BCs: Prescribe a 1% per year increase in CO2 concentrations up to 4 x CO2 levels (140-year run).

Expt 6.1-S (Short idealized 1%/yr run)

ICs: Initialize at year 20 of Expt 3.1-S.

BCs: Prescribe a 1% per year increase in CO2 concentrations as in Expt 6.1, but run for only 80 years.

Expt 6.2a (Baseline for prescribed SST experiments 6.2b, 6.4a,b)

ICs: Initialize atmosphere-only model from Expt 3.1.

BCs: Prescribe SSTs and sea ice, but with other BCs consistent with the climatology of Expt 3.1, for a run length of at least 30 years.

Expt 6.2b (Perturbed run for Hansen-style diagnosis of “fast” climate system responses to 4xCO2)

ICs: As in Expt 6.2a.

BCs: As in Expt 6.2a, but prescribe 4xCO2 concentrations for a run length of at least 30 years.

Expt 6.3 (Gregory-style diagnosis of “slow” climate system responses to 4xCO2)

ICs: As in Expt 6.2a.

BCs: As in Expt 6.2a, but prescribe an instantaneous quadrupling of CO2 concentrations (relative to preindustrial control Expt 3.1) that then are held fixed for a run length of 150 years

 

TIER 1


Expt 6.3-E (Ensemble of runs to improve estimate of Gregory-style “fast” climate response)

ICs: As in Expt 6.3, but initialize each ensemble member from a different month of the year.

BCs: As in Expt 6.3, but terminate each run after 5 years.

Expt 6.4a (Hansen-style diagnosis of “fast” climate system responses to all anthropogenic aerosols for year 2000)

ICs: As in Expt 6.2a.

BCs: As in Expt 6.2a, but with all anthropogenic aerosols consistent with conditions in the year 2000 of Expt 3.2 for a run length at least 30 years.

Expt 6.4b (Hansen-style diagnosis of “fast” climate system responses to sulfate aerosols alone for year 2000)

ICs: As in Expt 6.2a.

BCs: As in Expt 6.2a, but with sulfate aerosols consistent with conditions in the year 2000 of Expt 3.2 for a run length of at least 30 years.

Expt 6.5 (Cloud response to an imposed 4xCO2 [Hansen-style diagnosis of “fast” climate system responses])

ICs: As in AMIP Expt 3.3.

BCs: Consistent with CFMIP requirements (i.e. inactivated carbon cycle and prescribed observationally based vegetation characteristics), specify BCs as in Expt 3.3, but with only the radiation code “seeing” the quadrupled CO2 concentrations for a run length of 30 years.

Expt 6.6 (Cloud response to an imposed change in SST pattern)

ICs: As in Expt 3.3.

BCs: Consistent with CFMIP requirements, BCs are prescribed as in Expt 3.3, but with a CFMIP-provided patterned SST perturbation added to the AMIP SSTs for 1979-2008 (a run length of 30 years).

Expt 6.7a (Aqua-planet: control run)

ICs: Initialize with an atmospheric state that is consistent with the imposed zonally symmetric SST BCs.

BCs: Consistent with CFMIP requirements, impose zonally symmetric SSTs on an Earth without continents, sea ice, or active carbon cycle, and under a Sun fixed in a perpetual equinoctial position; then run for 5 years.

Expt 6.7b (Aqua-planet: cloud response to an imposed 4xCO2 [Hansen-style diagnosis])

ICs: As in Expt 6.7a.

BCs: As in Expt 6.7a, but with 4xCO2 atmospheric concentrations imposed.

Expt 6.7c (Aqua-planet: cloud response to an imposed uniform change in SST)

ICs: As in Expt 6.7a.

BCs: As in Expt 6.7a, but with a uniform +4 K SST perturbation added.

 

TIER 2


Expt 6.8 (Cloud response to an imposed uniform change in SST)

ICs: As in Expt 3.3.

BCs: Consistent with CFMIP requirements, specify BCs as in AMIP Expt 3.3, but with a uniform +4 K perturbation added to the SSTs for a run over years 1979-2008.

 



Table 7. Experiments for climate change detection and attribution studies

 

TIER 1


Expt 7.1 (Natural-only aerosol experiment)

ICs: As in Expt 3.1.

BCs: As in Expt 3.1, but with volcanoes and solar variability evolving as in historical Expt 3.2, for a run length of 156 years.

Expt 7.2 (GHG-only experiments)

ICs: As in Expt 3.1.

BCs: As in Expt 3.1, but with GHGs evolving as in Expt 3.2, for a run length of 156 years.

Expt 7.3 (Other individual forcing experiments)

ICs: As in Expt 3.1.

BCs: As in Expt 3.1, but with another individual forcing (e.g. only land use changes or anthropogenic aerosols or volcanic aerosols) evolving as in Expt 3.2, for a run length of 156 years.

 

TIER 2


Expts 7.1-E, 7.2-E, 7.3-E (Individual forcing ensembles)

ICs: As in Expt 3.1, but with each ensemble member initialized from different points in this preindustrial control run.

BCs: Impose BCs as in Expts 7.1-7.3, but for multi-member ensembles with run lengths of 156 years.




[Document prepared by T. Philips; April 2010.]