Seamless Weather and Climate PredictionSeamless Weather and Climate Prediction

(1)

Seamless Weather and Climate Prediction Seamless Weather and Climate Prediction

Jagadish Shukla

George Mason University (GMU), USA

Institute of Global Environment and Society (IGES)

“Revolution in Climate Prediction is Both Necessary and Possible”

Shukla, Hagedorn, Hoskins, Kinter, Marotzke, Miller, Palmer, and Slingo, BAMS 2009

Center of Ocean-Land- Atmosphere studies

Climate Test Bed Seminar Series 10 February 2009

(2)

Outline Outline

1. 1. Introduction: “Seamless” (WCRP) Introduction: “Seamless” (WCRP)

2. 2. Generalized Seamless Prediction Concept Generalized Seamless Prediction Concept 3. 3. Model Limitations and Successes Model Limitations and Successes

4. 4. Role of Tropical Convection/Heating Role of Tropical Convection/Heating 5. 5. Model Fidelity and Predictability Model Fidelity and Predictability

6. 6. World Modeling Summit for Climate Prediction World Modeling Summit for Climate Prediction

7. 7. Suggestions to Revolutionized Climate Prediction Suggestions to Revolutionized Climate Prediction

(3)

Evolution of the Concept of Seamless Prediction Evolution of the Concept of Seamless Prediction

in WCRP in WCRP

2002: In response to proposals by J. Shukla to launch the World Climate Experiment, and assess predictability of the climate system, the Joint

Scientific Committee (JSC) of WCRP (Hobart, Tasmania) established a Task Force on Predictability Assessment of the Climate System.

Members: B. Hoskins, J. Church, J. Shukla (Seamless Prediction concept introduced)

2004: JSC established a Talk Force on COPES

Members: R. Barry (CLiC), D. Carson (WCRP), B. Kirtman (TFSP), J. Matsumoto (CEOP), J. Mitchell (WGCM), K. Puri (WGNE), A. O’Neill (SPARC), J. Shukla

(JSC, WMP), P.K. Taylor, K. Trenberth (JSC, WOAP), M. Visbeck (CLIVAR), E.

Wood (GEWEX)

2005: WCRP/COPES strategic framework and WCRP Modeling Panel adopted the concept of seamless prediction as the organizing principle for WCRP

modeling.

COPES: Coordinated Observation and Prediction of the Earth System. (WCRP-123, WMO/TD-No. 1291, 2005, pp 1-59)

(4)

GEWEX 1988  SPARC

1992

WGNE WGCM WGSF

SOLAS 2001 ->

CLIVAR

1995 

CliC

²⁰⁰⁰



ObservationWCRP

&

Assmilation Panel

WCRP Modelling

Panel

Coordinated Observation and

Prediction of the Earth System

(5)

The WCRP Strategic Framework The WCRP Strategic Framework

2005-15 2005-15

AIM

To facilitate analysis & prediction of Earth system

variability & change for use in an increasing range of

practical applications of direct relevance, benefit & value to society

Coordinated Observation and Prediction of the Earth System

(WCRP-COPES)

(6)

Coordination of WCRP Modeling Coordination of WCRP Modeling

Activities Activities

WGNE, TFSP, GMPP

Intra-Seasonal Prediction

(1-30 days)

WGSIP, TFSP, GMPP, CliC, SPARC, WGOMD

Seasonal Prediction

(1-100 days)

WGNE

Weather Prediction

(1-10 days)

WGCM

Climate Change Prediction

(1-100,000 days)

WGCM, WGOMD

Decadal Prediction

(1-10,000 days)

WGSIP, TFSP, WGCM, WGOMD

Interannual Prediction

(1-1,000 days)

(7)

Seamless Prediction Problem Seamless Prediction Problem

1. There is now a new perspective of a continuum of prediction problems, with a blurring of the distinction

between shorter-term predictions and longer-term climate projections. Increasingly, decadal and century-long

climate projection will become an initial-value

problem requiring knowledge of the current observed state of the atmosphere, the oceans, cryosphere, and land surface (including soil moisture, vegetation, etc.) in order to produce the best climate projections as well as state-of-the-art decadal and interannual predictions.

WCRP Strategic Framework WCRP Strategic Framework

(COPES)

(8)

Seamless Prediction Problem Seamless Prediction Problem

2. The shorter time-scales and weather are known to be important in influencing the longer-time-scale behaviour. In addition, the

regional impacts of longer-time-scale changes will be felt by society

mainly through the resulting changes in the character of the shorter time- scales, including extreme events. In recognition of this, climate models are being run with the highest possible resolutions, resolutions that were employed in the best weather forecast models only a few years ago.

3. Even though the prediction problem itself is seamless, the best practical approach to it may be described as unified: models aimed at different time-scales and phenomena may have large commonality but place emphasis on different aspects of the system.

WCRP Strategic Framework WCRP Strategic Framework

(COPES)

(9)

Seamless Prediction Seamless Prediction

Since climate in a region is an ensemble of

weather events, understanding and prediction of regional climate variability and climate change, including changes in extreme events, will require a unified initial value approach that encompasses weather, blocking, intraseasonal oscillations,

MJO, PNA, NAO, ENSO, PDO, THC, etc. and climate change, in a seamless framework.

(10)

A Generalized Seamless Prediction A Generalized Seamless Prediction

Concept Concept

Seamless across:

• Space scales (clouds to global climate system)

• Time scales (minutes to centuries; multi-scale interactions)

• Phenomena (Convection-MJO-ENSO-PDO-AMO-Climate Change)

• Scientific disciplines (weather, climate, Earth system, biodiversity, socio-economics)

• Institutions (academic, government, corporations, intra-institutional labs/divisions)

• Political boundaries (local, state, national and international governments)

(11)

Global changeGlobal change Global changeGlobal change

Some Examples of Seamless Processes Some Examples of Seamless Processes

Tropical ConvectionTropical Convection (SST)

(SST)

Tropical ConvectionTropical Convection (SST)

(SST)

Rossby WavesRossby Waves (Atmosphere) (Atmosphere) Rossby WavesRossby Waves

(Atmosphere) (Atmosphere)

North America Forest FiresNorth America Forest Fires (Land)

(Land)

North America Forest FiresNorth America Forest Fires (Land)

(Land)

Surface WindSurface Wind (Atmosphere) (Atmosphere) Surface WindSurface Wind (Atmosphere) (Atmosphere)

Eurasian SnowEurasian Snow (Cryosphere) (Cryosphere) Eurasian SnowEurasian Snow

(Cryosphere) (Cryosphere)

Pacific/IO SSTPacific/IO SST (Ocean) (Ocean) Pacific/IO SSTPacific/IO SST

(Ocean) (Ocean)

Walker cellWalker cell (Atmosphere) (Atmosphere) Walker cellWalker cell (Atmosphere) (Atmosphere)

Asian MonsoonAsian Monsoon (Land) (Land) Asian MonsoonAsian Monsoon

(Land) (Land) Propagation downPropagation down

Propagation downPropagation down Extra Trop. Surface Winds Extra Trop. Surface WindsExtra Trop. Surface WindsExtra Trop. Surface Winds Upp. Stratosphere Circ.Upp. Stratosphere Circ.

Upp. Stratosphere Circ.Upp. Stratosphere Circ.

ENSOENSO ENSOENSO ISO/MJOISO/MJO

ISO/MJOISO/MJO Global Warming Global WarmingGlobal WarmingGlobal Warming

Regional SSTARegional SSTA

Regional SSTARegional SSTA Hurricanes HurricanesHurricanesHurricanes Persistent DroughtPersistent Drought

(Land) (Land)

Persistent DroughtPersistent Drought (Land)

(Land)

Influence ENSOInfluence ENSO (Ocean) (Ocean) Influence ENSOInfluence ENSO

(Ocean) (Ocean)

Monson DroughtsMonson Droughts (Atmosphere) (Atmosphere) Monson DroughtsMonson Droughts

(Atmosphere) (Atmosphere) Wet/Dry soil, TsWet/Dry soil, Ts

(Land) (Land) Wet/Dry soil, TsWet/Dry soil, Ts

(Land) (Land)

1.

2.

3.

4.

5.

7.

6.

(12)

From Cyclone Resolving Global Models From Cyclone Resolving Global Models

to to

Cloud System Resolving Global Models Cloud System Resolving Global Models

1. Planetary Scale Resolving Models (1970~): Δx~500Km 2. Cyclone Resolving Models (1980~): Δx~100-300Km 3. Mesoscale Resolving Models (1990~): Δx~10-30Km 4. Cloud System Resolving Models (2000 ~): Δx~3-5Km

Organized Convection

Cloud System

Mesoscale System

Synoptic Scale

Planetary Scale

Convective Heating Convective

Heating MJOMJO ENSOENSO Climate

Change Climate Change

Seamless Prediction of Weather and Climate

(13)

Important Issues and Discussions Important Issues and Discussions

1. Lack of comprehensive

1. Lack of comprehensive model developmentmodel development efforts globally efforts globally 2. Low resolution IPCC models can not simulate blocking

2. Low resolution IPCC models can not simulate blocking 3. 3. Regional downscalingRegional downscaling of climate change: of climate change: questionablequestionable

4. 4. Seamless predictionSeamless prediction: IPCC projections as “: IPCC projections as “Initial Value ProblemInitial Value Problem”” 5. 5. Insufficient computingInsufficient computing for next generation models for next generation models

6. Realism versus complexity: chemistry, biology; physical climate 6. Realism versus complexity: chemistry, biology; physical climate 7. Data assimilation for

7. Data assimilation for next generation modelsnext generation models 8. Lack of progress in

8. Lack of progress in ENSO predictionENSO prediction (model error, IC) (model error, IC) 9. Common

9. Common data managementdata management strategy for all WCRP activities strategy for all WCRP activities 10. 10. Joint WCRP-IGBP-THORPEX effortJoint WCRP-IGBP-THORPEX effort for models and data for models and data

assimilation assimilation

WMP reports to JSC (Zanzibar, 2007) WMP reports to JSC (Zanzibar, 2007)

that climate models have serious that climate models have serious

problems problems

(14)

Systematic Error: MSLP (NDJ)

(15)

Infamous Double Infamous Double

ITCZ Problem

(16)

Annual Cycle of SST Climatology Annual Cycle of SST Climatology

4-6 month forecast, APCC/CliPAS & DEMETER CGCMs 4-6 month forecast, APCC/CliPAS & DEMETER CGCMs

Calendar Month

Center of Ocean-Land-

Atmosphere studies Jin et al. 2008

(17)

NINO 3.4 Index (Observed and CFS) NINO 3.4 Index (Observed and CFS)

HadSSTv1.1

CFS long run

Calendar year

Jin and Kinter 2009 Climate Dynamics

(18)

50 60 70 80 90 100

1 2 3 4 5 6

Forecast Lead [Month]

Anomaly Correlation [%]

CFS CMP CCA CA MRK

15-member CFS reforecasts

Skill in SST Anomaly Prediction for Skill in SST Anomaly Prediction for

Nino3.4 Nino3.4

DJF 1981/82 to AMJ 2004

(19)

NINO3: Warm minus Cold composite

SST anomalies

Influence of Systematic Error on CFS Influence of Systematic Error on CFS

Forecast Skill Forecast Skill

 Warm composite (82/83, 86/87, 91/92, 97/98) - Cold composite (84/85, 88/89, 98/99, 99/00)

 Dashed lines denote composite for Hindcasts at different lead times

Observation CFS long run

(Hindcast composite) Forecast lead month

Correlation

CORR. with respect to lead month based on 1^st SEOF mode of SST

Correlation between 1^st PCs based on

observation and hindcasts at different lead times

Correlation between 1^st PCs based on long run and hindcasts at different lead times

  Model Flaw Model Flaw: Slow coupled dynamics of : Slow coupled dynamics of CGCM CGCM

19

Jin and Kinter 2009, Climate Dynamics

(20)

Fundamental barriers to advancing weather and

climate diagnosis and prediction on timescales from days to years are (partly) (almost entirely?)

attributable to gaps in knowledge and the limited

capability of contemporary operational and research numerical prediction systems to represent

precipitating convection and its multi-scale organization, particularly in the tropics.

(Moncrieff, Shapiro, Slingo, Molteni, 2007)

(21)

Shukla and Kinter 2006

Effect of SST Anomaly

(22)

1982-83

1988-89 Rainfall

Zonal Wind

1988-89

1982-83

The atmosphere is so strongly forced by the underlying

ocean that integrations with fairly large differences in the atmospheric initial conditions converge, when forced by the same SST (Shukla, 1982).

(23)

(24)

Evolution of Climate Models 1980-2000

Model-simulated and observed, 1983 minus 1989

Rainfall (mm day-1) 500 hPa GPH anomaly (m)

KUO R15

RAS R40

Observed

(25)

Percent Variance of PNA Percent Variance of PNA region explained by Tropical region explained by Tropical

SSTSST

Probability Distribution

Boreal Winter (DJF) Rainfall Boreal Winter (DJF) Rainfall

Variance in AGCMs Variance in AGCMs

(26)

Evolution of Climate Models

1980-2000

Model-simulated and observed rainfall anomaly (mm day^-1)

1983 minus 1989

(27)

Evolution of Climate Models

1980-2000

Model-simulated and observed 500 hPa height anomaly (m)

1983 minus 1989

(28)

MRF8: high, middle, low clouds allowed to exist

MRF9: Only high cloud allowed to exist over regions of tropical deep convection

MRF8 MRF8 MRF9 MRF9

(29)

Kumar et al. 1996 Journal of Climate MRF8: high, middle, low clouds allowed to exist

MRF9: Only high cloud allowed to exist over regions of tropical deep convection

(30)

Vintage 1980 GFDL AGCM (Lau, 1997, BAMS)

Note: amplitude of model response quite weak; structure is PNA rather than ENSO

forced

Note: estimate of

predictability depends on model fidelity

(31)

Vintage 2000

AGCM

(32)

Bjorn Stevens, UCLA World Modelling Summit, ECMWF, May 2008

Annually & Zonally Averaged Reflected SW Annually & Zonally Averaged Reflected SW

Radiation

(33)

Annually & Zonally Averaged SW Radiation (AR4)

(34)

Clouds as Ultimate, rather than Proximate, Sources Clouds as Ultimate, rather than Proximate, Sources

of Bias of Bias

(35)

1. Predictions of climate change depends on the climate model’s fidelity in simulating the current climate.

2. Models with low fidelity in simulating

climate statistics (mean and variability) have low skill in predicting seasonal climate

anomalies.

Conjectures Conjectures

(36)

Model sensitivity versus model relative entropy for 13 IPCC AR4 models. Sensitivity is defined as the surface air temperature change over land at the time of doubling of CO₂. Relative entropy is proportional to the model error in simulating current climate.

Estimates of the uncertainty in the sensitivity (based on the average standard deviation among ensemble members for those

J. Shukla, T. DelSole, M. Fennessy, J. Kinter and D. Paolino

Geophys. Research Letters, 33, doi10.1029/2005GL025579, 2006

Climate Model Fidelity and Projections of Climate Climate Model Fidelity and Projections of Climate

Change

(37)

Climate Model Fidelity and Projections of Climate Climate Model Fidelity and Projections of Climate

Change Change

Relative Entropy: The relative entropy between two distributions, p₁(x) and p₂(x), is defined as

(1) where the integral is a multiple integral over the range of the M-

dimensional vector x.

(2) where _j^k is the mean of p_j(x) in the kth season, representing the

annual cycle, j is the covariance matrix of p_j(x), assumed

independent of season and based on seasonal anomalies. The

distribution of observed temperature is appropriately identified with p₁, and the distribution of model simulated temperature with p₂.

(38)

Climate Model Fidelity and Projections of Climate Climate Model Fidelity and Projections of Climate

Change Change

Model vs. Model Relative Entropy with respect to MIROC high-resolution Model vs. Model Relative Entropy with respect to MIROC high-resolution

(39)

Climate Model Fidelity and Climate Prediction Climate Model Fidelity and Climate Prediction

Interim Conclusions:

• If we conjecture that models that better simulate the present climate should be considered more credible in projecting the future climate change, then this relationship suggests that the actual changes in global warming will be closer to the highest projected estimates among the current generation of models used in IPCC AR4.

• Lack of understanding of causes of model differences – is source of uncertainty in predicting climate change.

Question: Will AR5 be any different?

(40)

1. Predictions of climate change depends on the climate model’s fidelity in simulating the current climate.

2. Models with low fidelity in simulating

climate statistics (mean and variability) have low skill in predicting seasonal climate

anomalies.

Conjectures

(41)

DelSole (research in progress)

(42)

Examples of Success Examples of Success

Understanding of Dynamics and Physics Understanding of Dynamics and Physics

of A, O, L Climate System of A, O, L Climate System

• Numerical Weather Prediction (NWP) – Steady improvement in skill

• Dynamical Seasonal Prediction (DSP) – Prediction of large Amp. ENSO

• Climate Change Prediction (“IPCC”)

– Human activities are changing climate

(43)

(44)

Time series VW 850 tropics Time series VW 850 tropics

850 hPa WIND RMSEV (m/s) VERIFICATION AGAINST ANALYSIS

TROPICS

VERIFICATION TO W.M.O. STANDARDS

1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5

DWD 00UTC T+120 CANADA 00UTC T+120

UK 12UTC T+120 NCEP 00UTC T+120 ECMWF 12UTC T+120

1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5

FRANCE 00UTC T+24 DWD 00UTC T+24 CANADA 00UTC T+24 UK 12UTC T+24 NCEP 00UTC T+24 ECMWF 12UTC T+24

(45)

ERA Forecast Verification ERA Forecast Verification

Anomaly Correlation of 500 hPa GPH, 20-90N Anomaly Correlation of 500 hPa GPH, 20-90N

(46)

ERA Forecast Verification ERA Forecast Verification

Anomaly Correlation of 500 hPa GPH, 20-90N Anomaly Correlation of 500 hPa GPH, 20-90N

(47)

Number of Northern Hemisphere Number of Northern Hemisphere

Cyclones Cyclones

T255 ERA T159 T95

Jung 2006

47

(48)

Nastrom

&

Gage,198 5

(49)

dx= 40km dx= 25km 10km

Spectra of Total KE

log

₁₀

k k

^-3

k

^-5/3

(50)

Masaki Satoh, Hirofumi Tomita, Hiroaki Miura, Shinichi Iga and Tomoe Nasuno, 2005: J. Earth Simulator, 3, 1-9.

far …

Closest attempt to global cloud resolving model so far …

54 layers, top at 40 km 15-second time step

~ 1 TF-day per simulated day Ocean-covered Earth

Geodesic grid

3.5 km cell size, ~10⁷ columns

Running on Earth Simulator

(51)

Obs.Obs. (Takayabu et al. 1999)(Takayabu et al. 1999) NICAM (7-km)

NICAM (7-km)

Matsuno (AMS, 2007)

(52)

Masaki Satoh, JAMSTEC

(53)

MJO in High Resolution Model MJO in High Resolution Model

A Madden-Julian Oscillation Event Realistically Simulated by a Global Cloud-Resolving Model.

H. Miura, M. Satoh, T. Nasuno, A. T. Noda, and K. Oouchi Science, 1763 (2007); 318, DOI: 10.1126/science.1148443

(A) Infrared image from the Multi Functional Transport Satellite (MTSAT-1R) at 00:30 UaTC on 31 Dec 2006.

(B) outgoing longwave

radiation from the 3.5km-run

averaged from 00:00 UTC to

01:30 UTC on 31 Dec 2006.

(54)

Towards a Hypothetical “Perfect”

Model Model

• Replicate the statistical properties of the past observed climate – Means, variances, covariances, and patterns of covariability

• Utilize this model to estimate the limits of predicting the sequential evolution of climate variability

• Better model  Better prediction (??)

Societal Needs Societal Needs

• Regional climate prediction from days to decades

– Global cloud system resolving models are required

• Science based adaptation and mitigation strategies

– Billion to trillion dollar decisions to be made by policymakers

• Optimum utilization of space and in-situ observation

(55)

Revolution in Climate Prediction Revolution in Climate Prediction

is Possible and Necessary is Possible and Necessary

Coupled Ocean-Land-Atmosphere Model ~2015

~1 km x ~1 km (cloud- resolving)

100 levels

(Unstructured, adaptive grids)

~100 m 10 levels

Landscape-resolving

~10 km x ~10 km (eddy-resolving) 100 levels

(Unstructured, adaptive grids)

Assumption:

Computing power enhancement by a factor of 10⁶

• Improved understanding of the coupled O-A-B-C-S interactions Improved understanding of the coupled O-A-B-C-S interactions

• Data assimilation & initialization of coupled O-A-B-C-S system Data assimilation & initialization of coupled O-A-B-C-S system

(56)

Interim Conclusion Interim Conclusion

• The largest obstacles in realizing the potential predictability of weather and climate are inaccurate models and insufficient observations, rather than an intrinsic limit of predictability.

– In the last 30 years, most improvements in weather forecast skill have arisen due to improvements in models and assimilation techniques

• The next big challenge is to build a hypothetical “perfect”

model which can replicate the statistical properties of past observed climate (means, variances, covariances and

patterns of covariability), and use this model to estimate the limits of weather and climate predictability

– The model must represent ALL relevant phenomena, including ocean, atmosphere, and land surface processes and the interactions among them

Lecture in the Lorenz Symposium, AMS, 2005

(57)

Events Leading to the Modelling Summit Events Leading to the Modelling Summit

1. WCRP established new strategic framework in 2004 (COPES).

2. COPES established WCRP Modelling Panel (WMP).

3. WMP reports to JSC (Zanzibar, 2007) that climate models have serious problems.

4. JSC asks WMP (Chair: J. Shukla) to organize World Modelling Summit (WMS).

5. WCRP forms WMS organizing committee.

6. WMS takes place at ECMWF (6-9 May 2008). Nearly 150 participants from all modelling centers of the world.

(58)

Complexity Duration an

d/or Ensem

ble size

Resolution Computing

Resources

(59)

1.E+00 1.E+02 1.E+04 1.E+06 1.E+08 1.E+10 1.E+12

1.E+14

Weather and Weather and

Climate Model Climate Model

Evolution Evolution

 Computer power has increased 10⁶X since 1970s, but numerical models used for NWP and climate simulation have remained roughly the same

 Same equations (non-hydrostatic)

 Spectral or finite-diff. methods (fv cores; geodesic grids)

 Simple parameterizations (some improvement)

 Resolution: 4X in horizontal, 3X in vertical Accounts for ~ 10³ X

 Remaining 10³X:

 Longer runs

 Ensembles of model integrations

 Computer power has increased 10⁶X since 1970s, but numerical models used for NWP and climate simulation have remained roughly the same

 Same equations (non-hydrostatic)

 Spectral or finite-diff. methods (fv cores; geodesic grids)

 Simple parameterizations (some improvement)

 Resolution: 4X in horizontal, 3X in vertical Accounts for ~ 10³ X

 Remaining 10³X:

 Longer runs

 Ensembles of model integrations (courtesy of David Randall)

59

(60)

Yelick, U.C. Berkeley

Petaflop with ~1M Cores by 2008

(61)

Computing Capability & Model Grid Size Computing Capability & Model Grid Size

( ( ~ ~ km) km)

Peak Rate: 10 TFLOPS 100 TFLOPS 1 PFLOPS 10 PFLOPS 100 PFLOPS Cores ₍₂₀₀₆₎^1,400 ^12,000₍₂₀₀₈₎ ^80-100,000₍₂₀₀₉₎ 300-800,000

(2011)

6,000,000?

(20xx?)

Global NWP⁰:

5-10 days/hr ^{18 - 29} ^{9 - 14} ^{4 - 6} ^{2 - 3} ^{1 - 2}

Seasonal¹:

50-100 days/day ^{17 - 28} ^{8 - 13} ^{4 - 6} ^{2 - 3} ^{1 - 2}

Decadal¹:

5-10 yrs/day ^{57 - 91} ^{27 - 42} ^{12 - 20} ^{6 - 9} ^{3 - 4}

Climate Change²:

20-50 yrs/day ^{120 - 200} ^{57 - 91} ^{27 - 42} ^{12 - 20} ^{6 - 9}

Range: Assumed efficiency of 10-40%

0 - Atmospheric General Circulation Model (AGCM; 100 levels) 1 - Coupled Ocean-Atmosphere-Land Model (CGCM; ~ 2X AGCM computation with 100-level OGCM)

2 - Earth System Model (with biogeochemical cycles) (ESM; ~ 2X CGCM computation)

* Core counts above O(10⁴) are unprecedented for weather or climate codes, so the last 3 columns require getting 3 orders of magnitude in scalable parallelization (scalar processors assumed;

vector processors would have lower processor counts) Thanks to Jim Abeles (IBM)

(62)

Important Issues and Discussions Important Issues and Discussions

1. Lack of comprehensive

1. Lack of comprehensive model developmentmodel development efforts globally efforts globally 2. Low resolution IPCC models can not simulate blocking

2. Low resolution IPCC models can not simulate blocking 3. 3. Regional downscalingRegional downscaling of climate change: of climate change: questionablequestionable

4. 4. Seamless predictionSeamless prediction: IPCC projections as “: IPCC projections as “Initial Value ProblemInitial Value Problem”” 5. 5. Insufficient computingInsufficient computing for next generation models for next generation models

6. Realism versus complexity: chemistry, biology; physical climate 6. Realism versus complexity: chemistry, biology; physical climate 7. Data assimilation for

7. Data assimilation for next generation modelsnext generation models 8. Lack of progress in

8. Lack of progress in ENSO predictionENSO prediction (model error, IC) (model error, IC) 9. Common

9. Common data managementdata management strategy for all WCRP activities strategy for all WCRP activities 10. 10. Joint WCRP-IGBP-THORPEX effortJoint WCRP-IGBP-THORPEX effort for models and data for models and data

assimilation assimilation

that climate models have serious that climate models have serious

problems

(63)

Dr. Michel Beland Dr. Cecilia Bitz Dr. Gilbert Brunet Dr. Veronika Eyring Dr. Renate Hagedorn Dr. Brian Hoskins Dr. Christian Jakob Dr. Jim Kinter

Dr. Herve LeTreut Dr. Jochem Marotzke Dr. Taroh Matsuno Dr. Gerald Meehl Dr. Martin Miller Dr. John Mitchell Dr. Antonio Navarra

World Modelling Summit for Climate World Modelling Summit for Climate

Prediction Prediction

(International Organizing Committee) (International Organizing Committee)

Dr. Carlos Nobre

Dr. Tim Palmer

Dr. Venkatchalam Ramaswamy Dr. David Randall

Dr. Jagadish Shukla (Chair) Dr. Julia Slingo

Dr. Kevin Trenberth WMO / WCRP

Dr. Len Barrie Dr. John Church Dr. Ghassem Asrar

JSC asks WMP (Chair: J. Shukla) JSC asks WMP (Chair: J. Shukla)

to organize WMS to organize WMS

63

(64)

1. Overview: societal drivers; current status of weather and climate modeling; strategies for seamless prediction; crucial hypotheses (Hoskins)

2. Strategies for next-generation modelling systems: balance between resolution and complexity; balance between multi-model and unified modeling framework; issues of parameterizing unresolved scales and regional models (Miller)

3. Prospects for current high-end computer systems and implications for model code design (Kinter)

4. Strategies for model evaluation, modelling experiments, and initialization for prediction of the coupled ocean-land-atmosphere climate system (Marotzke)

5. Strategies for revolutionizing climate prediction: enhancing human and computing resources; requirements and possible organizational frameworks (Slingo)

World Modelling Summit for Climate World Modelling Summit for Climate

Prediction

Prediction (Themes and Theme leaders) (Themes and Theme leaders)

(65)

WMS takes place at ECMWF (6-9 May WMS takes place at ECMWF (6-9 May 2008). Nearly 150 participants from all 2008). Nearly 150 participants from all

modelling centers of the world.

Article in Nature, May 2008

(66)

Summary of WMS Summary of WMS

(summit declaration)

(67)

Challenge Challenge

The world recognizes that the consequences of global climate change constitute one of the most important threats facing

humanity. The peoples, governments, and economies of the world must develop mitigation and adaptation strategies, which will

require investments of trillions of dollars, to avoid the dire

consequences of climate change. The development of reliable, science-based adaptation and mitigation strategies will only be possible through a revolution in regional climate predictions,

supported by appropriate climate observations and assessment, and the delivery of this information to society.

(68)

1.Most important requirement: Prediction of changes in the statistics of regional weather variations.

2. Models have serious problems and cannot provide information with accuracy required by society

3. “A revolution in climate prediction is necessary and possible.”

(one of the most important declarations of the summit) 4. Proposal to establish a Climate Prediction Project

5. Enhance national centers

6. Establish a small number of climate research facilities for decadal prediction.

Summary of Summit Declaration

(69)

7. Dedicated high-end computing facilities are required (at least a thousand times more powerful than the currently available computers)

8. More computing power will help to enhance resolution and include complexity (e.g. biogeochemical cycles).

9. Global observations and assimilations are needed for prediction project.

10. Better estimates of uncertainties in climate prediction.

11. Collaboration between weather and climate prediction research communities (Seamless prediction).

12. Encourage the participation of young generation of climate modelers

Summary of Summit Declaration Summary of Summit Declaration

(70)

• There is a scientific basis for revolutionizing climate prediction

• The problem is beyond a person, a center, a nation …

• International collaboration is required

International Research and International Research and

Computational Facility to Computational Facility to

Revolutionize Climate Prediction

(71)

International Research and International Research and

Computational Facility to Computational Facility to

Revolutionize Climate Prediction Revolutionize Climate Prediction

1. Computational Requirement:

- Sustained Capability of 2 Petaflops by 2011 - Sustained Capability of 10 Petaflops by 2015

Earth Simulator (sustained 7.5 Teraflops) takes 6 hours for 1 day forecast using 3.5 km global atmosphere model; ECMWF (sustained 2 Teraflops) takes 20 minutes for 10 day forecast using 24 km global model

2. Scientific Staff Requirement:

- Team of 200 scientists to develop next generation climate model - Distributed team of 500 scientists (diagnostics, experiments)

A computing capability of sustained 2 Petaflops will enable 100 years of integration of coupled ocean-atmosphere model of 5 km resolution in 1 month of real time

(72)

International Research and International Research and

Computational Facility to Computational Facility to

Revolutionize Climate Prediction Revolutionize Climate Prediction

Examples of International Collaboration

• CERN: European Organization for Nuclear Research (Geneva, Switzerland)

• ITER: International Thermonuclear Experimental Reactor (Gadarache, France)

• ISS: International Space Station (somewhere in sky..)

(73)

IPCC AR4 Climate Modeling Centers IPCC AR4 Climate Modeling Centers

UKMO

FRCGC

NCAR GISS LASG MRI

MPI

BCC BCCR

CCCma

CNRM

MIUB IPSL

CSIRO INMCM

KMA GFDL

(74)

Facilities Facilities

American node

American node European/African nodeEuropean/African node Asian/Australian nodeAsian/Australian node

(75)

Scientific/Political Domains of Climate Modeling Scientific/Political Domains of Climate Modeling

Facilities Facilities

American node

American node European/African nodeEuropean/African node Asian/Australian nodeAsian/Australian node

UKMO

FRCGC

NCAR GISS LASG MRI

MPI

BCC BCCR

CCCma

CNRM

MIUB IPSL

CSIRO INMCM

KMA GFDL

75

(76)

Summary Summary

1. Models that better simulate the present climate produce the highest values of global warming for the 21^st century.

2. Models with low fidelity in simulating climate statistics have low skill in predicting climate anomalies.

3.3. Revolution in climate prediction is necessary and possible.Revolution in climate prediction is necessary and possible.

4.4. Seamless PredictionSeamless Prediction: : From cyclone resolving global models From cyclone resolving global models to cloud system resolving global models

to cloud system resolving global models 5.5. International collaboration is essential for: International collaboration is essential for:

capacity building; model development; computational power capacity building; model development; computational power

(77)

How to Implement a Seamless How to Implement a Seamless

Prediction System in the midst of Prediction System in the midst of

Several

Several Pre-existing Separate, Pre-existing Separate, Independent National Centers

Independent National Centers for for Weather, Climate, and Earth

Weather, Climate, and Earth System Science?

System Science?

(78)

US Climate Modeling US Climate Modeling

Infrastructure and World Infrastructure and World Climate Computing Facility Climate Computing Facility

• The US has 4 major climate modeling groups

– GFDL, GISS, GSFC, NCAR (Three contribute to IPCC)

• The US has 2 major data assimilations/reanalysis groups – GSFC, NCEP (both in Washington, DC area)

• The US has multiple small groups that utilize climate models – COLA, CSU, FSU, IRI, MIT, UCLA, UH, UMCP….

• The US has a large number of individual researchers (with students, post-doc’s, etc.) utilizing climate models and/or model outputs for research- too many to list.

 How will the US participate in a World Climate Facility?

(79)

Suggestions for India (may apply to Suggestions for India (may apply to

Brazil) Brazil)

1. The total national capacity for model development is limited: must build additional capacity

• Must use the existing capacity wisely

• Foster collaboration among groups (challenging!!)

2. Ensure that the (planned) climate center uses the operational NWP model as the dynamical core of the climate-earth system model (may require large effort; will foster national collaboration)

3. Develop mechanisms for perpetual interaction among groups for mutual benefit to keep the integrity of at least one truly unified national model development effort.

(80)

How to Implement a Seamless How to Implement a Seamless

Prediction System in a “Free” Country Prediction System in a “Free” Country

(viz USA) (viz USA)

1.A Climate center (e.g. NCAR, GFDL) takes an operational NWP/DSP model as the dynamical core, and builds a complex climate-earth system model (energy balance; biogeochemistry etc.)

(UNLIKELY)

2. An Operational NWP/DSP center (e.g. NCEP) takes the dynamical core of a climate model.

(Note: The approximate transition periods for 1 and 2 is about 2 years)

(UNLIKELY)

3. An Operational NWP/DSP center gradually expands its effort/

excellence towards NWP-Climate-Earth System Prediction System (IS IT POSSIBLE IN USA?)

(81)

How to Implement a Seamless How to Implement a Seamless

Prediction System in a “Free” Country Prediction System in a “Free” Country

(viz USA) (viz USA)

4. Create one new national (govt.; academia) “model development group” consisting of current and new staff that builds a new

generation of weather-climate-earth system prediction system with complexity/ resolution/ensemble size unaffordable at

individual centers.

The current centers and research groups, universities use the model(s) for different applications (NWP, DSP, IPCC, mechanistic experiments etc.).

81

(82)