© Crown copyright 2006 Page 1
The Cloud Feedback ModelIntercomparison Project (CFMIP)
Progress and future plansMark Webb, Keith Williams, Mark Ringer, Alejandro Bodas
Salcedo, Cath Senior (Hadley Centre)
Tomoo Ogura, Yoko Tsushima, (NIES, JAMSTEC Japan)
Sandrine Bony, Majolaine Chiriaco (IPSL/LMD)
and CFMIP contributors (BMRC,GFDL,UIUC,MPI,NCAR)
CERES/GERB meeting October 2006
© Crown copyright 2006 Page 2
Modelling and Prediction of Climate variability and change
CFMIP: Cloud Feedback Model Inter-comparison Project
• Set up by Bryant McAvaney (BMRC), Herve Le Treut (LMD)
• WCRP Working Group on Coupled Modelling (WGCM)
• Systematic intercomparison of cloud feedbacks in GCMs
• +/-2K atmosphere only and 2xCO2 ‘slab’ experiments
• Aim to identify key uncertainties
• Link climate feedbacks to cloud observations
• ISCCP simulator required (Klein & Jakob, Webb et al)
• Now have data for 13 GCM versions from 8 groups
• Website shows data available, publications, plans, etc.
• http://www.cfmip.net
© Crown copyright 2006 Page 3
Modelling and Prediction of Climate variability and change
CFMIP: Website http://www.cfmip.net
© Crown copyright 2006 Page 4
Comparison of +/- 2K and slab model experimentsRinger et al, GRL 2006
Cess experimentscapture the spread incloud feedback fromslab experiments.
Offset due tosuppressed clear-skyfeedbacks in slab vsCess
Values are globalmean changes in NET,SW and LW CRF perdegree of warming(Wm-2K-1)
© Crown copyright 2006 Page 5
Changes in ISCCP cloud types slab vs +/-2KRinger et al, 2006
Values areglobal meanchange in eachcloud type perdegree ofwarming (% / K)
High top
Mid-level
top
Low top
Thin Medium Thick
Ringer et al, 2006
© Crown copyright 2006 Page 6
Bony and Dufresne GRL 2005
Cloud radiative forcing (CRF)climate response in verticalvelocity bins over tropicaloceans (30N-30S) from 15coupled climate models8 higher sensitivity and7 lower sensitivity
Net CRF spread largest insubsidence regions,suggesting low clouds are‘at the heart’ of cloudfeedback uncertainties
© Crown copyright 2006 Page 7
CFMIP cloudfeedback “classes”
Webb et al 2006
Models with positive lowcloud feedback over largerareas have higher sensitivity
© Crown copyright 2006 Page 8
Webb et al Climate Dynamics 2006 – CFMIP models
LW cloud feedback (W/m2/K)SW cloud feedback (W/m2//K)Net cloud feedback (W/m2/K)
LW cloud feedback (W/m2/K)SW cloud feedback (W/m2//K)Net cloud feedback (W/m2/K)
Areas withsmall LWcloud feedbacksexplain 59% ofthe NET cloudfeedbackensemblevariance
Cloud feedbacksin these areasare indeeddominated byreductions inlow level cloudamount (shownwith ISCCPsimulator)
© Crown copyright 2006 Page 9
Tsushima et al.,2006
Cloud liquid(2xCO2 -1xCO2 kg/kg) Cloud ice (1xCO2 kg/kg)
‐90 ‐60 ‐30 0 30 60 90 ‐90 ‐60 ‐30 0 30 60 90
0
200
400
600
800
1000
0
200
400
600
800
1000
0
200
400
600
800
1000
0
200
400
600
800
1000
0
200
400
600
800
1000
[hPa]
-2e-5 0 2e-5 0 2e-5
Clim
ate
sens
itivi
ty
6.3℃
4.0℃
3.6℃
2.9℃
2.3℃
0
200
400
600
800
1000
0
200
400
600
800
1000
0
200
400
600
800
1000
0
200
400
600
800
1000
0
200
400
600
800
1000
[hPa]
MIROC-HS
MIROC-LS
HadSM4
GFDL
UIUC
© Crown copyright 2006 Page 10
Williams et al Climate Dynamics 2006 (CFMIP)
RMS-differences of present-day variabilitycomposites against observations for 10CFMIP/CMIP model versions. The fivemodels with smallest RMS errors tend to havehigher climate sensitivities. (Consistent withBony & Dufresne 2005)
1.01.01.91.51.1
0.91.11.01.11.2
ERBE/NCEPLW
2.12.41.72.23.4
1.21.31.51.81.7
ERBE/ERASW
1.11.01.61.41.3
1.01.21.01.21.1
ERBE/ERALW
2.33.02.22.83.7
1.61.31.92.12.2
ERBE/NCEPSW
1.61.91.92.02.4
1.21.21.41.51.5
AvgRMS
3.0K
2.3-3.5K
3.8K
2.9-4.4K
AvgClimateSens.&Range
© Crown copyright 2006 Page 11
Williams and Tselioudis (Climate Dyn in revision)
ISCCP observational cloud cluster regimes (20N-20S)
© Crown copyright 2006 Page 12
!!!===
""+"+"="nclusters
i
ii
nclusters
i
ii
nclusters
i
ii CRFRFOCRFRFORFOCRFCRF
111
In the cloud regime framework, the mean change in cloud radiative forcing can be thought of as havingcontributions from:
•A change in the RFO (Relative Frequency of Occurrence) of the regime
•A change in the CRF (Cloud Radiative Forcing) within the regime (i.e. a change in the tau-CTP spaceoccupied by the cluster/development of different clusters).
Williams and Tselioudis
© Crown copyright 2006 Page 13
Modelling and Prediction of Climate variability and change
CFMIP Phase I continues…
Daily cloud diagnostics to be hosted by PCMDI
To be made available as a community resource
UKMO, MIROC, MPI, NCAR data currently in transit
IPSL, Env Canada promised later this year
Will allow many ISCCP cloud studies to be applied to a representative selection of climate models
Will become part of standard IPCC diagnostic protocol by time of AR5 (agreed by WGCM Sep 06)
© Crown copyright 2006 Page 14
CFMIP Phase II – looking further ahead
Understandingof modelled cloud-climate
feedback mechanisms
Evaluation of model clouds
using observations
Assessment ofcloud-climate
feedbacks
Co-ordinators: Mark Webb, Sandrine Bony, Rob ColmanProject advisor: Bryant McAvaney
Main objective : A better assessment of modelled cloud-climate feedbacks for IPCC AR5
© Crown copyright 2006 Page 15
Modelling and Prediction of Climate variability and change
CFMIP Phase II – planned approach
Develop improved cloud diagnostic techniques in climate models :
- CFMIP CloudSat/CALIPSO simulator - Cloud water budget / tendency terms - 3 hourly data at key locations (ARM sites, GPCI )
Explore the sensitivities of cloud feedbacks to differing modelassumptions using idealised climate change experiments
Demonstrate the application of these techniques to theunderstanding and evaluation of cloud climate feedbacks via pilotstudies
Organise a systematic cloud feedback model comparison with thenext generation of climate models (ideally by embedding suitablecloud diagnostics in the AR5 experimental protocol.)
© Crown copyright 2006 Page 16
Modelling and Prediction of Climate variability and change
CFMIP CloudSat/CALIPSO simulator (C3S)
This is a modular cloud simulator framework which willallow a number of cloud simulator modules to beplugged into climate models via a standard interface.
This is currently under development in collaboration withvarious groups:
- Hadley Centre (Alejandro Bodas-Salcedo, Mark Webb, Mark Ringer) - LLNL (Steve Klein, Yuying Zhang) - LMD/IPSL (Marjolaine Chiriaco, Sandrine Bony) - CSU (Johnny Lyo, John Haynes, Graeme Stephens) - PNNL ( Roger Marchand )
© Crown copyright 2006 Page 17
C3S/CloudSat comparison with UK NWP model(Alejandro Bodas-Salcedo)
B
A
AB
Transect trough a mature extra-tropical systemStrong signal from ice clouds
Strong signal from precipitation
Cloud and precip not present in obs
© Crown copyright 2006 Page 18
ACTSIM LIDAR comparison with GLAS / ICESat data
(M. Chiriaco LMD/IPSL)
Lidar signal simulatedfrom LMDZ outputs
Lidar signal observedfrom GLAS spatial lidar
latitude
z, k
mz,
km
signal
Evaluation of the vertical structure of the atmosphere in models,
at global scale
Indicates excessive reflectivities from cloud ice in this climate model
© Crown copyright 2006 Page 19
Cloud water budget / process diagnostics(Tomoo Ogura NIES Japan)
=!
+!
t
ice)Qc(liqCondensation + Precipitation + Ice-fall-in + Ice-fall-out
+ Advection + +Cumulus Conv.
(1) (2)
(3) (4) (5)
Source term
Qc response
- Transient response of terms (1)…(5) to CO2 doubling is monitored every year.
- Terms showing positive correlation with cloud water response contributeto the cloud water variation.
mixing adjustment
response(1)or…(5)
Qc increase related to thesource term
Qc decrease related to thesource term
© Crown copyright 2006 Page 20
[Cloud water vs ]
90S 60S 30S EQ 30N 60N 90N90S 60S 30S EQ 30N 60N 90N1.0
0.6
0.2sigma
1.0
0.6
0.2
ΔCloud water (2xco2 - 1xco2, DJF)
-1 0 1
cloud decrease increase
(5)Dry conv. adj.(4)Cumulus(3) Advection(2) Ice fall in+out(1)Condens-PrecipCorrelation coefficient (when positive)
contour=3e-6 [kg/kg]
Cloud water budget diagnosis in MIROC(Tomoo Ogura NIES Japan)
Decrease in mid-low level sub-tropical cloud water correlates with decrease in large-scale condensation
© Crown copyright 2006 Page 21
GCSS/WGNE Pacific Cross-section Intercomparison (GPCI)
180 190 200 210 220 230 240Longitude
0
10
20
30
40
0.05
0.15
0.25
0.35
0.45
0.55
0.65
0.75
0.85
0.95
ISCCP total cloud cover Sea Surface Temperature
GPCI is a working group of the GEWEX Cloud System Study (GCSS)
Models and data are analyzed along a Pacific Cross section fromStratocumulus, to Cumulus and to deep convection
Period: June-July-August 1998 and 2003Time resolution: 0, 3, 6, 9, 12, 15, 18, 21 UTC
-1 2 5 8 11 14 17 20 23 26 29 32 35288
290
292
294
296
298
300
302
SS
T (
K)
latitude (degrees)
AM2 ARPEGE CAM 3.0 GSM0412 HadGAM RAC GME
JJA 1998
Joao Teixeira [email protected]
© Crown copyright 2006 Page 22
-1 2 5 8 11 14 17 20 23 26 29 32 35
latitude (degrees)
cam_av_cling.98000099 (g/Kg)
1000
900
800
700
600
500
400
300
200
100
pre
ssure
(hP
a)
-1 2 5 8 11 14 17 20 23 26 29 32 35
latitude (degrees)
hadgam_av_cling.98000099 (g/Kg)
1000
900
800
700
600
500
400
300
200
100
pre
ssure
(hP
a)
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.06
0.08
0.1
0.15
0.2
0.3
g/kg
Mean GPCI liquid water cross section - JJA98
from three climate models
liquid water
Too shallow -> fog Is this too muchliquid water?
How deep shouldthe PBL be..?
Joao Teixeira [email protected]
© Crown copyright 2006 Page 23
Mean diurnal cycle: ISCCP cloud cover
peak values of Sc cloudcover around 32-35 N
peak values of mid/highclouds close to ITCZ
Diurnal cycle: max in(early) morning local time
Joao Teixeira [email protected]
© Crown copyright 2006 Page 24
0 3 6 9 12 15 18 21
hours (UTC)
am_maxlcc01700_lathv_98000099 (%)
-1
2
5
8
11
14
17
20
23
26
29
32
35
latit
ude
(deg
rees
)
0 3 6 9 12 15 18 21
hours (UTC)
gsm_maxlcc01700_lathv_98000099 (%)
-1
2
5
8
11
14
17
20
23
26
29
32
35
latit
ude
(deg
rees
)
0 3 6 9 12 15 18 21
hours (UTC)
rac_maxlcc01700_lathv_98000099 (%)
-1
2
5
8
11
14
17
20
23
26
29
32
35
latit
ude
(deg
rees
)
JJA98 mean diurnal cycle: model low cloud cover
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
peak values too far tothe south (around 26 N)
realistic diurnalcycle: morning max
Joao Teixeira [email protected]
© Crown copyright 2006 Page 25
Modelling and Prediction of Climate variability and change
Proposed point diagnostics for CFMIP Phase II
- 3 hourly instantaneous model data- 10-20 years of data to give stable statistics- on a grid of locations covering the GPCI and TWP- additional key locations (e.g. ARM sites, high latitudes)- in AMIP and idealised climate change experimentsThe aim is to align climate model diagnostics with thosein use in GCSS/ARM to encourage a wider group ofpeople to examine and criticise cloud feedbacks inclimate modelsFor example this would give insight into the impact ofdiurnal cycle errors on cloud-climate feedbacks
© Crown copyright 2006 Page 26
Modelling and Prediction of Climate variability and change
CFMIP Phase II physics sensitivity tests
Sensitivity tests are proposed using the idealised climatechange experiments developed by Brian Soden:
i.e. AMIP and AMIP + composite CMIP SST anomaly
The aim is to quantify the impact of certain differences inmodel formulation on cloud-climate feedbacks byimplementing consistent treatments across models
Possible examples include: - fixed liquid cloud water content and radiative properties - consistent precipitation & mixed phase partitioning - consistent boundary layer resolution - consistent simple shallow convection scheme?
- suppression of interactive cloud radiative properties- simplified mixed-phase feedback processes- simplified clouds and precipitation- consistent treatment of shallow convection
Also possible tests of the effects of differing deep convective entrainment/detrainment on climate sensitivity
New process diagnostics will aid the interpretation of these experiments
© Crown copyright 2006 Page 27
Modelling and Prediction of Climate variability and change
CFMIP Phase II timescales
Concrete project proposal by Jan 2007
Aim for endorsement by WGCM and GEWEX SSG in early 2007
Joint CFMIP/ENSEMBLES meeting Paris April 2007
Development of diagnostics / pilot studies 2007-2008
Systematic model inter-comparison with new model versions 2008- (preferably as part of AR5 models)
© Crown copyright 2006 Page 28
Modelling and Prediction of Climate variability and change
Use of CERES/GERB products in CFMIP PII
CFMIP Phase I – mostly ISCCP/ERBE/ISCCP_FD
All of the following will benefit CFMIP Phase II:
CERES SRBAVG GEO, SYN/AVG/ZAVG products CERES/MODIS cloud retrievals CloudSat/CALIPSO/CERES merged products Surface + ATM + TOA products (e.g. SARB) Diurnal cycle GEO/GERB data
The barriers are often in bringing the sampling and statistical summaries from satellite products and GCM diagnostics into line – e.g. providing ISCCP D1-like tau-Pc histograms
© Crown copyright 2006 Page 30
Cleaner separation of SW feedbacksTaylor et al (in revision)
APRP (ApproximatePartial RadiativePerturbation) methodfor separation of SWcloud and non-cloudfeedbacks.
Validated against fullGFDL PRPcalculations courtesy
of B. Soden.
Similar to Yokohataet al method butwith subtledifferences.
© Crown copyright 2006 Page 31
Taylor et al SW APRP on CFMIP model data
Michel Crucifix, KarlTaylor, Mark Webb
APRP method givesa more positiveSW cloud feedbackand larger spreadthan CRF method
This is because itmakes a cleanerseparation betweensurface albedo andcloud feedbacks
Wm-2K-1
(Mean, SD)
© Crown copyright 2006 Page 32
Summary
A number of studies now point to low cloud feedbacks beinga key uncertainty in climate models
Daily cloud/radiation/ISCCP simulator diagnostics fromCFMIP Phase I will be available from PCMDI by the end ofthe year
Reductions in uncertainty due to cloud feedbacks will requirelinks to observations and also understanding and criticism offeedback mechanisms in models
CFMIP Phase II is an opportunity to align diagnostics in arange of models with those in use in GCSS/ARM/CPT
We hope to formalise our plans by the end of this year.
© Crown copyright 2006 Page 33
HadSM3 Cloud Response along GCSS Pacific Cross Section transect
Mid-level cloud categories excluded.
SW CRF response along Pacifictransect in CFMIP and AR4 slabmodels. Cloud feedbacks are typicallysmaller than model biases. No clearrelationship between bias and feedback
© Crown copyright 2006 Page 34
HadSM3 Cloud Response along GCSS Pacific Cross Section transect(Negative low cloud feedback case)
Mid-level cloud categories excluded.
© Crown copyright 2006 Page 35
Bony and Dufresne (2005)
Interannual sensitivity ofCRF to SST in verticalvelocity bins over tropicaloceans (30N-30S) from 15coupled climate models8 higher sensitivity and7 lower sensitivity andobserved (ISCCP FD)
Models underpredict lowcloud sensitivity, but highersensitivity models less so.
© Crown copyright 2006 Page 36
Low cloud response in HadSM3 transition regions
Deep and shallow convection weaken in the warmerclimate (consistent with Held and Soden 2006 )
Shallow convection typically detrains into two model layersin present day, but one level in the warmer climate
If a certain amount of water vapour is detrained into asingle layer it will moisten that layer more than would bethe case if it was spread over two layers
May explain why HadSM3 stratiform cloud fractionincreases with weakening shallow convection
Hence the negative low cloud feedback in HadSM3 maybe due to poor vertical resolution and so not credible
© Crown copyright 2006 Page 37
HadGSM1 Cloud Response along GCSS Pacific Cross Section transect
© Crown copyright 2006 Page 38
Requires: more cloud diagnostics from GCMs + enhanced scrutiny
CFMIP-II:➔Encourage the analysis of cloud feedback processes by a wider community !
- make cloud diagnostics more easily accessible to the community (daily clouddiagnostics from CFMIP-1 to be available from PCMDI)
- strengthen the link between CMIP and CFMIP (e.g. by increasing the number ofcloud diagnostics in the outputs of coupled models, by running the ISCCP simulator)
➔Organize climate physics sensitivity experiments + consistent implementation ofsimplified physics (e.g. mixed-phase cloud feedback)
Understandingof cloud feedbacks
Evaluationof cloud fields
Assessment ofclimate changecloud feedbacks
© Crown copyright 2006 Page 39
How may we use our physical understanding of climate change cloud feedbacks and theavailable model-data comparisons to define a “metrics” for cloud feedbacks?
CFMIP-II:➔Explore relationships between cloud evaluation tests and cloud-climate feedbacksbased on a wide diversity of diagnostics and approaches + a large number of GCMs
➔Discuss the issue during a joint CFMIP/ENSEMBLES workshop in Paris (11-13 April 2007).
Understandingof cloud feedbacks
Evaluationof cloud fields
Assessment ofclimate changecloud feedbacks
© Crown copyright 2006 Page 40
Modelling and Prediction of Climate variability and change
CFMIP Phase II
CFMIP Phase II will aim to reduce uncertainty in cloudfeedbacks and climate sensitivity by developing further linksbetween feedbacks and observations and improving ourunderstanding of cloud feedback mechanisms by:
1. Developing better cloud diagnostics for models: - CloudSat/CALIPSO simulator - GCSS Pacific Cross Section (diurnal cycle, …) - Cloud budget/tendency diagnostics
2. Applying 2. Exploring sensitivities of feedbacks to model physics e.g. low clouds, convective entrainment
3. Collaboration with Gewex/GCSS community - GCSS Pacific Cross Section (diurnal cycle, …)
© Crown copyright 2006 Page 41
Modelling and Prediction of Climate variability and change
Alternative experimental setups
Cess +/-2K fixed season experiments are not a quantitative guide to coupled model feedbacks - no seasonal cycle - no high latitude amplification of warming
Alternative options include: - continuing use of mixed layer experiments
- re-running sections of AR4 coupled experiments with extra diagnostics - ‘patterned SST perturbation’ experiments as developed by Brian Soden (possibly AMIP+)
© Crown copyright 2006 Page 42
Modelling and Prediction of Climate variability and change
Proposed new diagnostics for CFMIP Phase II
CFMIP Cloudsat/CALIPSO Simulator (C3S)
Sub-timestep information- Cloud condensate budget terms- Physics increments for temperature, humidity, etc
Detailed diagnostics at key locations as used in GCSS/ARM studies (e.g GPCI)
© Crown copyright 2006 Page 43
European Cloud System study (EUROCS)Pacific Cross Section
Siebesma et al 2004
Pacific TransectCalifornia/trades/ITCZ
9 regional, NWP and climatemodels compared, JJA 1998
Models too little cloud in Scregions
( +60 W/m2 SW error)Too much in trade cumulus areas ( -60 W/m2 SW error )
Total Cloud Amount (%)
Net SW TOA (Wm-2)
© Crown copyright 2006 Page 44
CFMIP Phase IIUnderstandingcloud feedback mechanisms
Evaluating model clouds/feedbacks
using observations
Assessment ofcloud-climate
feedbacks
Develop better cloud diagnostics for climate models:
High frequency instantaneous diagnostics alongWGNE/GCSS GPCI and at ARM sites
CFMIP CloudSat/CALIPSO Simulator (C3S)
© Crown copyright 2006 Page 45
Requires: process-level studies + model-data comparisons + new satellite data
➔Embed GCSS (e.g. high-frequency diagnostics at ARM sites, GEWEX Pacific Cross-Section; evaluation of some specific cloud processes)
➔Development of a CFMIP CloudSat/CALIPSO Simulator (C3S) (including eventuallyan A-train orbital simulator) -> will favor interactions with obs people!
➔We would like the AR5 models to be run with the ISCCP/radar/lidar simulator, and theCFMIP cloud diagnostics to be included into the list of standard outputs
Understandingof cloud feedbacks
Evaluationof model clouds
Assessment ofcloud climate
feedbacks
© Crown copyright 2006 Page 46
Reducing uncertainty by understanding feedbacks
1/ Try to understand how physical cloud climatefeedback processes are operating in models
2/ Ask ‘Is this behaviour credible?’
3/ Develop new schemes with more credible cloud-climate feedback behaviour in mind
4/ Differences between model feedbacks may reduce inthe longer term
5/ Can help to focus attention on key physicalprocesses for cloud feedbacks
© Crown copyright 2006 Page 47
Modelling and Prediction of Climate variability and change
Further links between feedbacks and observables
New diagnostics will provide opportunities for new links
e.g. CloudSat/CALIPSO data may constrain models with strong mixed-phase cloud feedbacks due to excessive amounts of cloud ice
This will be an ongoing area of research, and the focus of a joint ENSEMBLES/CFMIP meeting in Paris 11th-13th April 2007
© Crown copyright 2006 Page 48
CFMIP CloudSat/CALIPSO simulator(Alejandro Bodas-Salcedo)
The simulator consists of 5steps:
1. Orbital simulation2. Sampling3. Preprocessing4. Subgrid sampling of cloud
overlaps5. Radar reflectivity calculated
using code provided by MattRogers (CSU)
Simulated CloudSat reflectivities from UKMO forecast model
Outputs:
- Reflectivity from clouds and precipitation (without attenuation)
- Total reflectivity, accounting for attenuation by gases, clouds and precipitation
- Products obtained from inputs at gridbox and subgrid scales
© Crown copyright 2006 Page 49
ACTSIM
Lidar Equation
•Optical properties:P(π,z), Qsca(z), Qabs(z)(532 nm)
•Lidar calibration
•Multiple scattering
ex. lidarprofile (log)
ex. particlesconcentration profile
(/m3)
ex. water contentprofile (g/m3)
snow
ice
liqui
d
ice
ACTSIM CALIPSO simulator
– Marjolaine Chiriaco (LMD/IPSL)
© Crown copyright 2006 Page 50
Modelling and Prediction of Climate variability and change
CFMIP Phase II C3S inter-comparison
Initially we will provide an A-train orbital simulator to allow climate modellers to save model cloud variables co-located with CloudSat/CALIPSO overpasses
Initially sampled data would be submitted to CFMIP and both simulators run centrally
As the approach matures we plan to integrate this package with the ISCCP simulator so that it can be run in-line as part of the model development cycle