4 th WOAP Meeting 29 March 2010 K. E. Taylor The Working Group on Coupled Modeling Karl E. Taylor Program for Climate Model Diagnosis and Intercomparison

4th WOAP Meeting29 March 2010

K. E. Taylor

The Working Group on Coupled Modeling

Karl E. Taylor

Program for Climate Model Diagnosis and Intercomparison (PCMDI)

with credit and thanks to Sandrine Bony

Presented to the Fourth WCRP Observation and Assimilation Panel Meeting

Hamburg, Germany

29 March 2010

PCMDI4th WOAP Meeting29 March 2010 K. E. Taylor

Who are the WGCM?

Members

Sandrine Bony (co-chair)

Gerald Meehl (co-chair)

Veronica Eyring

Marco Giorgetta

David Karoly

M. Kimoto

Corinne Le Quéré

Natalie Mahowald

Catherine Senior

Bin Wang

Ex-Officio Members

Gokhan Danabasoglu

Helge Drange

Greg Flato

Filippo Giorgi

John Mitchell

Ronald Stouffer

Karl Taylor


What is the WGCM’s mission?

• Foster the development of coupled climate models (and now ESMs)

• Coordinate model experiments and inter-comparisons to:

better understand natural climate variability

predict the climate response to natural & anthropogenic perturbations

Promote and facilitate model evaluation and diagnosis of shortcomings

A balance among:Predicting – Evaluating - Understanding


Major challenges

• How can we improve our confidence in climate models?

• How can we assess the credibility of model projections ?


K. E. Taylor

LES modelsCloud Resolving Models

Single ColumnModels

3D-Climate ModelsNWP Models

High resolution global models(global CRM, MMF)

Global observational datasets

Field campaigns & instrumented sites

How can we gauge and gain confidence in GCMs projections?(1) Bottom-Up approach : evaluate and improve the physical basis of climate models

through large-scale and process-scale evaluations

Analysis & Understandingclimate change

Model projections


K. E. Taylor

LES modelsCloud Resolving Models

Single ColumnModels

3D-Climate ModelsNWP Models

High resolution global models(global CRM, MMF)

Global observational datasets

Field campaigns & instrumented sites

How can we gauge and gain confidence in GCMs projections?(1) Bottom-Up approach : evaluate and improve the physical basis of climate models

through large-scale and process-scale evaluations

(2) Top-Down approach : understand the models' results & identify critical processes to provide guidance for specific observational tests/process

studies and model improvements

Analysis & Understandingclimate change

Model projections


The second approach has traditionally dominated WGCM’s interests, but this is changing.

• Current activities focus on understanding why model projections differ

e.g., quantifying the strength of individual feedbacks across models.

• Evaluate climate models over a wide range of scales and phenomena

From weather to paleo

From global to regional

From individual physical processes to climate (across all physical and biogeochemical components)

• Explore how model formulation and present-day model performance translate to reliability of climate projections

Perhaps the biggest challenge of all.


CMIP3 transformed climate science by enabling community-wide participation in the analysis of model output.

0

100

200

300

400

500

600

700

800

900

1000

1100

12009/26/2009

daily raterunning 31-day mean

date

gig

abyte

s p

er

day

WCRP CMIP3 Archive Download Rates

• 35 Tbytes of model

output stored at PCMDI

• More than 765 TB

downloaded

• More than 3,000 users

• More than 550 publications

Aug 2009Jan 2007(AR4 WGI)

Nov 2004

Courtesy of Bob Drach (PCMDI)


What are prospect for CMIP5?

• Better understand robust and uncertain aspects of climate change

Enable quantification of strengths of major feedbacks

• Include carbon cycle component (ESM’s)

• Better meet the needs of the “impacts” community

A more comprehensive set of model output

• Provide information needed to assess adaptation and mitigation strategies

• Coordinate/integrate across the modeling community:

CMIP includes portions of: C4MIP, PMIP, AMIP, CFMIP, Aqua-planet


K. E. Taylor

CMIP5: Three Suites of Experiments

“Long-Term”(century & longer)

TIER 1

TIER 2

CORE

evaluation& projection

diagnosis

“Near-Term”(decadal prediction)

(initialized ocean state)

hindcasts & forecasts

CORE

TIER 1

TIER 2

TIER 1

AMIP

“time-slice”

CORE

Atmosphere-Only

(for computationally demanding and NWP models)

TIER 1

TIER 2Taylor et al. 2008,

http://cmip-pcmdi.llnl.gov/cmip5


K. E. Taylor

Example: CMIP5 long-term suite of experiments

An important focus is model evaluation and understanding...

D & A

ense

mble

s

Control, AMIP, & 20 C

RCP4.5, RCP8.5

natural-only,

GHG-only

individual

forcing

RCP2.X, RCP6

extend

RC

P4.5 to

2300 extend RCP8.5 &

RCP2.X to 2300

ense

mbl

e of

ab

rupt

4xC

O2

5-yr

run

saqua

planet

(clouds) uniform

ΔSST

(clouds)

Mid

-Hol

ocen

e &

LG

Mlast

m

illen

nium

E-driven RCP8.5

E-driven control & 20 C

patterned

ΔSST

(clouds)

aerosol forcing

ca. 2000

AC&C4

(chemistry)

1%/yr CO2 (140 yrs)

abrupt 4XCO2 (150 yrs)

fixed SST with 1x & 4xCO2

radiation code sees 1xCO

2 (1%/yr or 20C+RCP4.5)carbon cycle sees 1XCO

2 (1%/yr or 20C+RCP4.5)Understanding

Model Evaluatio

n

ClimateProjections

Green subset is for coupled carbon-cycle climate models only

ensembles: AMIP & 20 C


K. E. Taylor

CMIP now involves many WCRP/IGBP partners

Example: CMIP5 long-term suite of experiments

Detection-Attribution(IDAG)

Paleo(PMIP, IGBP-

PAGES)

Cloud andmoist processes

(CFMIP-GCSSWGNE)

Carbon-climate feedbacks(C4MIP, IGBP-AIMES)

Integrated AssessmentConsortium (IAM),

connection to WG-III

+ Satellite simulators& process

diagnostics (CFMIP-GCSS)

Chemistry, aerosols(SPARC, AC&C,

CCMVal, aerocom)


WGCM (with others) promotes a variety of community-wide activities to advance climate modeling

• Mentioned already:

SPARC & IGBP/IGAC (CCMVal, AeroCom..): chemistry & aerosols

WGCM & IGBP/PAGES (PMIP): paleoclimate

WGCM, GCSS, and WGNE (CFMIP): clouds and cloud feedbacks

WGCM, IGBP/AIMES (C4MIP): carbon cycle

IDAG: detection and attribution studies

• WGNE/WGCM (Transpose-AMIP): evaluation of climate models in NWP mode

• CLIVAR WGSIP, WGOMD : seasonal to interannual prediction, ocean

• TFRCD (CORDEX) : regional

• GEWEX GCSS (GPCI) : processes

• WGNE/WGCM Metrics panel

• CF metadata conventions for archiving and sharing climate data

• WGCM endorsed demonstration study (GeoMIP): geo-engineering


WGCM’s Challenge to WOAP

• Push to make observations as easily available and useable as climate model output generated by CMIP.

Provide easy access and guidance on quality and limitations

Following CMIP3, store observational data in a standard way and make it available through a common portal.

• The WGCM has endorsed a NASA/JPL pilot initiative to provide satellite data in a form useful to CMIP5 (Joao Texeira, Duane Waliser, Jerry Potter, S Boland).

• A similar NOAA initiative may be launched soon, and this in the work plans ISENES (a European project to provide infrastructure support for climate model research)

• These new projects were all partially inspired by CMIP

The WGCM would like to see similar efforts undertaken by other providers of satellite and in-situ observations.


WGCM’s Challenge to WOAP

• Provide encouragement and possibly a framework whereby

the observational and climate modeling community would interact to identify observational data sets useful in model evaluation

The highest priority would be to consider the CMIP output -– variables, temporal/spatial sampling, time-periods.

• Establish guidelines for

Metadata that will facilitate search and discovery.

Formats and metadata that will facilitate analysis (as provided by the CF metadata standard)

• Develop a strategy for making multiple datasets developed for this purpose accessible in a way that parallels the CMIP model output archive.



List of CMIP5 output fields

• http://cmip-pcmdi.llnl.gov/cmip5/output_req.html

• Domains:

Atmosphere (including aerosols)

Ocean (including carbon cycle variables)

Land surface (including carbon cycle variables)

Cryosphere

• Temporal sampling

Annual

Monthly

Daily (including max., min. & mean surf. Air T, precip. humidity, surf. wind, PSL; many more 2-d & 3-d fields for 1950-2005)

6-hourly

3-hourly (including from 1950-2005)

~ half-hour (but not globally)

http://cmip-pcmdi.llnl.gov/cmip5/output_req.html


Model output characteristics

• Specified template for filenames and directory structure

• Additional metadata

modeling_realm

tracking_id

model_id

creation_date

Forcing

initialization_method,

physics_version

• Output may be on native grid, rather than longitude-latitude cartesian


K. E. Taylor

CMIP5 Decadal Prediction Experiments

additional predictions Initialized in

‘01, ’02, ’03 … ’09100-yr “control”

& 1% CO

2

prediction with 2010 Pinatubo-

like eruption

alternative initialization strategies

atmos. chem

istry

&/or aerosols &/or

regional air quality

AMIP

incre

ase

ense

mble

size

s

from

O(3

) to

O(10)

mem

bers

hind

cast

s with

out

volc

anoe

s

30-year hindcast and prediction ensembles: initialized 1960, 1980 &

2005

10-year hindcast & prediction ensembles: initialized 1960, 1965, …,

2005


K. E. Taylor

CMIP5 Atmosphere-Only Experiments(targeted for computationally demanding and NWP models)

AMIP(1979-2008)

AMIP

ensemble

AMIP

SSTs

with 4

XCO 2aqua planet

(clouds)

uniform ΔSST

(clouds)

patterned ΔSST

(clouds)

future “time-slice”(2026-2035)

Future “time-slice”

ensemble

Documents

4 th WOAP Meeting 29 March 2010 K. E. Taylor The Working Group on Coupled Modeling Karl E. Taylor Program for Climate Model Diagnosis and Intercomparison