1. Ronald J Stouffer Karl Taylor, Jerry Meehl and many others
Overview of Coupled Model Intercomparison Project (CMIP) and IPCC AR5 Activities
Ronald J Stouffer
Karl Taylor, Jerry Meehl and many others
June 2009

2. Outline Background Overview of CMIP5/AR5 design Questions Review AR4
Initialization
Variability
Overview of CMIP5/AR5 design
Long term experiments
Near term experiments
Questions

3. Climate Model (AOGCM) Earth System Model (ESM)
An Earth System Model (ESM) closes the carbon cycle
Atmospheric circulation and radiation
Climate Model (AOGCM)
Sea Ice
Land physics
and hydrology
Ocean circulation
Atmospheric circulation and radiation
Allows Interactive CO2
Earth System Model (ESM)
Top – Typical AR4 model – physical climate model with atmosphere, ocean, land sea ice components
Bottom – Earth System Model – many definitions, for AR5 – closed carbon cycle (green boxes), no implications for fidelity of the atm chemistry. Our ESMs currently have no atmospheric chemistry in this generation model. See our plans for making CM3 an earth system model.
Sea Ice
Plant ecology and
land use
Ocean ecology and
Biogeochemistry
Land physics
and hydrology
Ocean circulation 3

4. CMIP3/AR4
About 18 groups using 24 models participated. More than 1000 papers written using CMIP3 database
Downloads continue at very high rate

5. Initialization: How is this done? Implications for you
Modelers make a long Pre-industrial control
Typically 1850 or 1860 conditions
Perturbation runs start from control
Model related to real years only through radiative forcing
Solar, volcanoes, human emissions, land use, etc.
Each ensemble member an equally likely outcome
Do not expect wiggles to match – model vs obs

6. Model vs. Observed Global Mean Surface Temperature Anomalies
Obs-Jones
Obs-GISS oK
Year

7. Global surface temperatureoC

8. US Surface TemperatureoC

9. DC Surface TemperatureoC

10. Natural Variability
Natural variability confounds or “hides” the global warming signal.
Smaller space scales and shorter time scales exhibit more natural variability
Signal to noise ratio much lower for smaller space and time scales
Temperature has a relative large signal to noise ratio, precipitation is much smaller S/N ratio

11. Human activities are very likely the cause of the warming of last 100 years.
Black line: temperature observation from thermometers.
Pink shade: Climate model simulations using all past radiative forcings.
Blue shade: Climate model simulation using only natural forcings (solar,
volcanoes).
IPCC WGI SPM 2007

12. Human activities are likely to be the cause of the warming over last 100 years on each continent.
IPCC WGI SPM 2007

13. CMIP5 Background Process started in 2005 Aspen meeting
Lots of groups with input
IPCC WG1, 2, 3 authors
Various WCRP and IGBP committees
Lots of individuals
Strategy adopted by WGCM in Paris in September 2008

14. Long Term Experiments Core Tier 1 Tier 2 Core: ≥1718 yrs
Tier 1: ≥1727 yrs
Tier 2: ≥2038 yrs
Core
Tier 1
Tier 2

15. Long Term Experiments Overview
AOGCM – control/historical/projections
Physical climate models
ESM – control/historical/projections
AOGCM + closed carbon cycle
Lots of runs for “understanding”

16. CMIP5 Long Term Experiments AOGCM Core (concentration driven)
Control
500+ years
Historical
~1850 to 2005
Projection
2006 to 2100
RCP4.5 (stabilization near 2100)
RCP8.5 (GHG continue to increase)
AMIP – Atmosphere-land model, obs SST + sea ice
1979 to 2008
1850
2005
2100

17. Runs for Understanding Core
AOGCM or ESM
1% CO2 increase to 2X (140 years)
TCR
4XCO2 switch-on (150 years)
Equilibrium climate sensitivity
Atmosphere-land
Hansen style
1XCO2
4XCO2

18. Long Term Experiments AOGCM Tier 1
Historical
More ensemble members
Natural, GHG-only forcing, other?
Projection
RCP2.X (GHG continue to decrease) 2006 to 2100
RCP6.0 (stabilization near 2100) 2006 to 2100
RCP to 2300
AMIP
PMIP – see PMIP protocol; AOGCM or ESM
6KBP
LGM

19. Long Term Experiments AOGCM Tier 2
Historical
Ensemble natural and GHG-only forcings
Ensemble individual forcings
Projection
RCP2.X 2101 to 2300
RCP to 2300
PMIP – see PMIP protocol; ESM or AOGCM
Last 1000 yrs (850 to 1850)

20. Long Term Experiments ESM Core
All AOGCM long term runs plus
Control with pCO2 free
500+ years
Historical – emission driven
~1850 to 2005
Projection – emission driven
RCP8.5 (2006 to 2100)

21. GFDL’s Long Term Experiment Plans
CM3 using GHG concentrations and aerosol emissions
Study aerosol-cloud interactions and impact on climate and climate change
ESM2M and ESM2G
Study carbon cycle feedbacks in climate change
Study impact of increasing carbon on ecosystems
Study impact of using differing ocean vertical coordinates on simulation and response

22. CMIP5 Near Term Experiments
Foci
Regional information
Short lived species
Decadal Prediction
Core
Tier 1
Core: yrs
Tier 1: ≥1700 yrs

23. Near Term - Core 10 year hindcasts 30 year forecasts
Initialized at 1960, 1965, 1970 …
3+ ensemble members
30 year forecasts
Initialized at 1960, 1980, 2005
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
2010
2030

24. Near Term – Tier 1 Increase ensemble size to 10+
Hindcast without volcanoes
Runs initialized in 200X
2001, 2002, 2003, 2004, 2006, 2007, 2008, …
Prediction with Pinatubo-like event in 2010

25. Near Term – Tier 1 Investigate role of short live species
Investigate alternative initialization methods
100 yr control and 1% run (if not doing long term experiments with same model)

26. Near Term – Tier 1 Time Slice Experiments
Time periods
AMIP ( ) and
Overlap with decadal prediction exps
High atmospheric resolution
Atmospheric chemistry experiments
Study regional impacts
Study extreme events

27. GFDL’s Near Term Experiment Plans
CM2.1 using Coupled Data Assimilation with ensemble filter techniques
Investigate predictability on decadal time scales
High resolution atmosphere-land-only models in time slice mode (25km)
Investigate regional climate change
Study extreme events (hurricanes)
Potentially use high resolution coupled models
Study importance of resolution on predictability
GFDL will use at least 2 different types of models for the near term AR5 experiments. One will be our AR4 model, CM2.1. We have developed an ocean data assimilation system based on that model version. Using output from the data assimilation system, we will initialized the model and run a number of the integrations.
We will also develop a very high atmosphere-only model. This model will be used to study extreme events like hurricanes and other regional climate issues. Much of the integration of this model will occur on the machine at Oak Ridge.

28. CMIP5/AR5 Data Serving Distribution data paradigm
Earth System Grid (ESG) software
PCMDI still a gateway to data
Several new gateways
BMRC, MPI, NCAR, GFDL, …
Archive expected to be more than 1 PB
CMIP panel oversight
Variable list

29. GFDL Data Serving
Node on PCMDI’s network (ESG) of data servers for CMIP5
Also provides an independent path to GFDL data
Currently serving 10’s of TB to external users
Potentially 100’s of CMIP5/AR5 TB available

30. Thank you Questions?

31. Runs for Understanding Tier 1
ESM-only
CO2 increases for bio, CO2 constant for radiation
1% or historical+RCP4.5
AOGCM or ESM
Ensemble 4XCO2 switch-on (5 years)
Atmosphere-land
CFMIP simulator – observed SSTs, sea ice
See CFMIP protocol
Aqua planet
Hansen
Future run with sulfate at year 2000

32. Runs for Understanding Tier 2
ESM-only
CO2 increases for bio, CO2 constant for radiation
1% or historical+RCP4.5
AOGCM or ESM
AC&C chemistry protocol
Atmosphere-land
CFMIP simulator – +4C increase SSTs
See CFMIP protocol