predictability of weather and climate (seamless prediction of weather and climate) department of...

Predictability of Weather and Climate(Seamless Prediction of Weather and Climate)

Department of Atmospheric, Oceanic and Earth SciencesDepartment of Atmospheric, Oceanic and Earth Sciences

George Mason UniversityGeorge Mason University

Jagadish ShuklaJagadish Shukla

CLIM 751CLIM 751Fall 2012Fall 2012

Lecture on Aug 29, Lecture on Aug 29, 20122012

OutlineOutline• Weather and Climate for Poets

• Mechanisms of Variability of Weather and Climate

• Predictability and Prediction of Weather and Climate– Weather– Climate (Seasonal, ENSO, Decadal)– Climate Change

• Model Fidelity, Predictability and Sensitivity

• Factors Limiting Predictability: Future Challenges– Observational and Theoretical (Physics & Dynamics of the

Coupled Climate System)– Computational and Numerical

• Summary and Conclusions

ECMWF: Steady improvement of weather forecast skill, 1980-2010

NHSH

ECMWF: Skill of deterministic forecasting systems, 1980-2010

Nino3.4 SST anomaly predictions from March 2011

Global-mean Surface TemperatureGlobal-mean Surface Temperature

On the Time-Varying Trend in Global-Mean Surface Temperature by Huang, Wu, Wallace, Smoliak, Chen, Tucker

EEMD: Ensemble Empirical Mode Decomposition; MDV: Multi Decadal Variability

Courtesy of UCAR

1.0º C

Physics of Weather and Climate for Poets

Climate is what you expect,weather is what you get.

(quoted by E. N. Lorenz)

CLIMATE DYNAMICS CLIMATE DYNAMICS OF THE PLANET EARTHOF THE PLANET EARTH

S

Ω

a

g

T4

WEATHER

CLIMATE .

hydrodynamic instabilities of shear flows; stratification & rotation; moist thermodynamics

day-to-day weather fluctuations; wavelike motions: wavelength, period, amplitude

T_

y,U

_

yT

_

z,U

_

z

S, , a, g, ΩO3

H2OCO2

stationary waves (Q, h*), monsoons

h*: mountains, oceans (SST)w*: forest, desert (soil wetness)

(albedo)

Examples of Weather and Climate Examples of Weather and Climate VariabilityVariability

• Annual CycleAnnual Cycle

• Daily WeatherDaily Weather

• Seasonal ClimateSeasonal Climate

• Interannual (ENSO)Interannual (ENSO)

• Decadal Decadal

• Centennial (Climate Change)Centennial (Climate Change)

Daily, Intraseasonal, Seasonal, Daily, Intraseasonal, Seasonal, Interannual, and Decadal Interannual, and Decadal

VariationsVariations “Short range” weather variation

• Hours; thunderstorms, tornadoes, squall lines, fronts, ….• Diurnal cycle; Organized convection• “Cyclones”, Eeasterly waves, Depressions, ….

“Medium range” weather variations

• Blocking; Growth, decay of tropical, tropospherical disturbances

Intraseaonal variations

• Madden Julian “Oscillation” (MJO), Monsoon Intraseasonal variations, Pacific North American (PNA) variations, Annular modes

Seasonal mean variations

• Persistent droughts; Floods; Persistent “hot” and “cold” days; “Anomalous” number and tracks of cyclones

Interannual variations

• ENSO, QBO, TBO, NAO, NAM, SAM

Decadal variations

• PDO, Thermohaline circulation, Sahel drought, Decadal ENSO

Climate change • Solar, Volcanoes, Greenhouse gases, Land use change

Mechanisms of VariabilityMechanisms of VariabilityInternal External

Weather: 1. Internal Dynamics of Atmosphere

• Boundary Condition of SST, Soil wetness, Snow, Sea ice, etc.

Climate:(seasonal-decadal)

2. Internal Dynamics of Coupled Ocean-Land-Atmospshere

• Solar, Volcanoes

Climate Change:

3. Internal Dynamics of Sun-Earth System

• Human effects: (Greenhouse gases, land use changes)

14

2008

Bivariate Correlation

between ECMWF Ensemble Mean

Forecast and Observations

Based on 80 Dates (1st Feb, May, Aug,

Nov, 1989-2008)

MJO

Model 1: (Tropics) a = 1.98

Model 2: (Mid-latitude) b = 1.60

An ensemble of 10000 initial random errors was allowed to evolve for each model.

22222

2

21111

1

EsEdt

dE

EsEdt

dE

aXX nn 2

1

bYY nn 101.0 21

37.0

63.0

2

1

21

Growth of Random Errors in the simple Growth of Random Errors in the simple modelmodel

of Tropics and midlatitudes of Tropics and midlatitudes

Empirical fit for Error growth

ERA Forecast VerificationERA Forecast Verification Anomaly Correlation of 500 hPa GPH, 20-90NAnomaly Correlation of 500 hPa GPH, 20-90N

1980-2006


Schematic Error Growth Schematic Error Growth for the Winter (Red) & Summer for the Winter (Red) & Summer

(Blue)(Blue)

RMS Error and Differences between Successive RMS Error and Differences between Successive ForecastsForecasts

Northern Hemisphere 500 hPa Height in WinterNorthern Hemisphere 500 hPa Height in Winter

Current Limits of Predictability, A. Hollingsworth, Savannah, Feb 2003

Evolution of 1-Day Forecast Evolution of 1-Day Forecast Error, Lorenz Error Growth, Error, Lorenz Error Growth,

and Forecast Skill for ECMWF and Forecast Skill for ECMWF Model Model

((500 hPa NH Winter)500 hPa NH Winter)

1982 1987 1992 1997 2002

“Initial error”

(1-day forecast error) [m]20 15 14 14 8

Doubling time [days] 1.9 1.6 1.5 1.5 1.2

Forecast skill [day 5 ACC ] 0.65 0.72 0.75 0.78 0.84

-5/3 spectrum

-3 spectrum

synoptic scales mesoscales

The “Knife’s Edge” – The Observed Spectrum The “Knife’s Edge” – The Observed Spectrum Nastrom & Gage 1985

• In spite of the k -5/3 spectrum,

• NWP history (~40 years) suggests: Higher resolution models, improved physical parameterizations, and data assimilation techniques reduced initial errors; Increased the range of predictability (even though initial error growth increased).

• Despite 40 years of research, we still cannot definitively state whether the range of predictability cannot be increased by reducing the initial error.

Interim Summary (NWP)Interim Summary (NWP)

From Numerical Weather Prediction (NWP) From Numerical Weather Prediction (NWP) To Dynamical Seasonal Prediction (DSP) (1975-To Dynamical Seasonal Prediction (DSP) (1975-

2004)2004)

•Operational Short-Range NWP: was already in place

•15-day & 30-day Mean Forecasts: demonstrated by Miyakoda (basis for creating ECMWF-10 days)

•Dynamical Predictability of Monthly Means: demonstrated by analysis of variance

•Boundary Forcing: predictability of monthly & seasonal means (Charney & Shukla)

•AGCM Experiments: prescribed SST, soil wetness, & snow to explain observed atmospheric circulation anomalies

•OGCM Experiments: prescribed observed surface wind to simulate tropical Pacific sea level & SST (Busalacchi & O’Brien; Philander & Seigel)

•Prediction of ENSO: simple coupled ocean-atmosphere model (Cane, Zebiak)

•Coupled Ocean-Land-Atmosphere Models: predict short-term climate fluctuations

Effects of SST AnomalyEffects of SST Anomaly

Observed 5-month running mean SOIObserved 5-month running mean SOI

IC: Dec. 1982

IC: Dec. 1988

JFM Mean Rainfall JFM Mean Rainfall AnomaliesAnomalies

IC: Dec. 1988

IC: Dec. 1982

“Predictability in the Midst Of Chaos”

Model

Model

IC: Dec. 1982

1982-83 SST The atmosphere is so strongly forced by the underlying ocean that integrations with very large differences in the atmospheric initial conditions converge, when forced by the same SST.

Zonal Wind (m/s) at 200 Mb (10°S to 10°N, 120°W to 160°W)

IC: Dec. 1988

JFM Mean Rainfall JFM Mean Rainfall AnomaliesAnomalies

IC: Dec. 1988

IC: Dec. 1982

“Predictability in the Midst Of Chaos”

Observations

Model

Model

B.C.(SST): 1982 -83

When tropical forcing is very strong, it can enhance even the predictability of extratropical seasonal mean circulation, which, in the absence of anomalous SST, has no predictability beyond weather. Observed SST JFM83

Observed SST JFM83

IC: Dec. 1988

IC: Dec. 1982


Observed SST JFM83

IC: Dec. 1988

IC: Dec. 1982

Observed ϕ’ (meters)


Observed SST JFM83 Observed SST JFM89

Observed SST JFM89


IC: Dec. 1982

IC: Dec. 1988


Observed SST JFM83 Observed SST JFM89

Observed SST JFM89

IC: Dec. 1988

IC: Dec. 1982

IC: Dec. 1988

IC: Dec. 1982


1982-83

1988-89

Rainfall

Zonal Wind

1988-89

1982-83The 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).

(Trenberth, et al. 1998)

Northward Propagating Rossby-Wave TrainNorthward Propagating Rossby-Wave Train

MJO

Convection

JET

Tropical ConvectionTropical Convection

Vintage 2000AGCM

Model Simulation of ENSO EffectsModel Simulation of ENSO Effects500 hPa Height Anomalies (ACC = 0.98)500 hPa Height Anomalies (ACC = 0.98)

Predictability Limited Due to Initial Condition Uncertainty:Predictability Limited Due to Initial Condition Uncertainty:Two Time Scales in the Error Growth?Two Time Scales in the Error Growth?

dE1

dt1E1 1E1

2

E1

dE2

dt2E2 2E2

2

E2

E(t) E1(t) E2(t)

Goswami and Shukla (1991, J. Clim.)

50

60

70

80

90

100

1 2 3 4 5 6

Forecast Lead [Month]

An

om

aly

Co

rre

lati

on

[%

]

CFSCMPCCACAMRK

15-member CFS reforecasts

Skill in SST Anomaly Prediction for Skill in SST Anomaly Prediction for Nino3.4Nino3.4

DJF 1981/82 to AMJ 2004

-3

-2

-1

0

1

2

3

Selected Dynamical Models, 5-month lead

2002 2003 2004 2005 2006 2007 2008 2009

OBSGMAO

CFS LDEO

CFS

GMAO

LDEO

.

.

..

.

.

Tony Barnston and Mike Tippett

20 Years: 1980-19994 Times per Year: Jan., Apr., Jul., Oct.6 Member Ensembles

Kirtman, 2003

Current Limit of Predictability of ENSO Current Limit of Predictability of ENSO (Nino3.4)(Nino3.4)

Potential Limit of Predictability of ENSOPotential Limit of Predictability of ENSO

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)

44

Factors Limiting Predictability

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

Pro

bab

ility

Dis

trib

uti

on

Boreal Winter (DJF) Rainfall Variance in AGCMsBoreal Winter (DJF) Rainfall Variance in AGCMs

(mm2)

Observed and Simulated Surface Observed and Simulated Surface Temperature (°C)Temperature (°C)Observed Simulated

Scientific Basis for Decadal Predictability

• Slowly varying climate components • Atmosphere-ocean interactions (Pohlmann et al., 2006;

Stouffer et al., 2006, 2007; Latif and Barnett, 1996; Held et al., 2005; Knight et al., 2006; Zhang and Delworth, 2006).

• Decadal predictability in oceans (Griffes and Bryan, 1997; Collins and Sinha, 2003; Collins et al., 2006, Msadek et al., 2010, DelSole et al., 2010).

• Potential predictability of temperature, precipitation, sea level pressure (Collins, 2002; Boer, 2004; Boer and Lambert2008; Pohlmann et al., 2004, 2006, Smith et al., 2007; Keenlyside et al., 2008).

• Predictable external forcing (Hegerl et al., 2007).

Boer &Lambert, 2008, Geophys.Res. Lett.

Percent of potential predictable variance of 5-yr mean

Example of Unforced Predictability Study

Little to no predictability over land !

Recent Results on Multi-year Recent Results on Multi-year Predictability over LandPredictability over Land

Diagnosis of Multi-year Predictability on Continental ScalesLiwei Jia and Timothy DelSole

(J. Climate, 2011)

Robust Multi-Year Predictability on Continental ScalesLiwei Jia

(Ph.D. Thesis, George Mason University, 2011)

Leading Predictable Component (APT):Leading Predictable Component (APT):Internal Multi-decadal Pattern (IMP)Internal Multi-decadal Pattern (IMP)

(°C)

Scientific Basis for Decadal PredictabilityScientific Basis for Decadal Predictability

CMIP5: Global Mean 2m Temp (°C) for 20th Century Historical Runs

NOAA

Dynamical Prediction Experience Dynamical Prediction Experience

Model predictability depends on model fidelity

ERA Forecast VerificationERA Forecast Verification


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

Models that simulate climatology “better” make better predictions.

Definition: Fidelity refers to the degree to which the climatology of the forecasts (including the mean and variance) matches the observed climatology

HypothesisHypothesis

DEMETER Data• 7 global coupled atmosphere-ocean

models• 9 ensemble members• 1980-2001 (22 years)• Initial conditions: 1 February, 1 May, 1

August, 1 November• Integration length: 6 months

Testing the Hypothesis: Testing the Hypothesis: DataData

Climate Model Fidelity and PredictabilityClimate Model Fidelity and Predictability

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 jk is the mean of pj(x) in the kth season, representing the

annual cycle, j is the covariance matrix of pj(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.

Fidelity vs. SkillFidelity vs. Skill

Courtesy of Tim DelSole

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 models for which multiple realizations are available) are shown as vertical error bars. The line is a least-squares fit to the values.

J. Shukla, T. DelSole, M. Fennessy, J. Kinter and D. PaolinoGeophys. Research Letters, 33, doi10.1029/2005GL025579, 2006

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

Black: Reanalysis (ERA); Red: T 159; Blue: T 1279 (ECMWF)(Higher Resolution Model Improves Simulation of Blocking Frequency)

Blocking FrequencyBlocking Frequency

ERA-40

T159

T1279

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

Matsuno (AMS, 2007)

(b) Coupled model (2 degree)- Climatology -

Monsoon Rainfall in Low Resolution Monsoon Rainfall in Low Resolution ModelModel

Oouchi et al. 2009: (a) Observed and (b) simulated precipitation rate over the Indo-Chinamonsoon region as June-July-August average (in units of mm day -1). The observed precipitation is from TRMM_3B42, and the simulation is for 7km-mesh run.

Monsoon Rainfall in High Resolution Monsoon Rainfall in High Resolution ModelModel

Dynamical Prediction Experience Dynamical Prediction Experience (~30 years)(~30 years)

• Weather 500,000 (30 years X 365 days X 50 centers)

• Seasonal 5,000 (30 years X 12 months X 15 centers)

• Decadal 5

Seamless PredictionSeamless 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.

Conceptual/TheoreticalENSO: unstable oscillator?ENSO: stochastically forced, damped linear system?(The past 50 years of observations support both theories)

– Role of weather noise?

Modeling• Systematic errors of coupled models - too large• Uncoupled models not appropriate to simulate Nature in some

regions/seasons: CLIMATE IS A COUPLED PROCESS• Atmospheric response to warm and cold ENSO events is nonlinear

(SST, rainfall and circulation)• Distinction between ENSO-forced and internal dynamics variability

ChallengesChallenges

ChallengesChallenges

Observational• Observations of ocean variability • Initialization of coupled models

Computational • Very high resolution models of climate system need million fold

increases in computing• Storage, retrieval and analysis of huge model outputs• Power (cooling) and space requirements-too large

Summary (1)Summary (1)• In spite of the k -5/3 spectrum, NWP history (~40

years) suggests: Higher resolution models, improved physical parameterizations, and data assimilation techniques reduced initial errors; and Increased the range of predictability.

• 35 years ago, Dynamical Seasonal Prediction (DSP) was not conceivable; DSP has achieved a level of skill that is considered useful for some societal applications. However, such successes are limited to periods of large, persistent SST anomalies.

Summary (2)Summary (2)• The most dominant obstacle in realizing the potential

predictability of seasonal variations is inaccurate models, and unbalanced initial conditions rather than an intrinsic limit of predictability.

• Because of a lack of suitable computing and modeling infrastructure (dedicated powerful computers and critical mass of scientists) we are unable to derive the benefits of expensive space and in-situ observing systems, and apply the scientific and technological advances for accurate and reliable regional climate prediction.

• 25 years ago, a dynamical seasonal climate prediction was not conceivable.

• In the past 20 years, dynamical seasonal climate prediction has achieved a level of skill that is considered useful for some societal applications. However, such successes are limited to periods of large, persistent anomalies at the Earth’s surface. Dynamical seasonal predictions for one month lead are not yet superior to statistical forecasts.

• There is significant unrealized seasonal predictability. Progress in dynamical seasonal prediction in the future depends critically on improvement of coupled ocean-atmosphere-land models, improved observations, and the ability to assimilate those observations.

Conclusions, Conjectures and SuggestionsConclusions, Conjectures and Suggestions

• Improvements in dynamical weather prediction over the past 30 years did not occur because of any major scientific breakthroughs in our understanding of the physics or dynamics of the atmosphere

• Dynamical weather prediction is challenging: progress takes place slowly and through a great deal of hard work that is not necessarily scientifically stimulating, performed in an environment that is characterized by frequent setbacks and constant criticism by a wide range of consumers and clients

• Nevertheless, scientists worldwide have made tremendous progress in improving the skill of weather forecasts by advances in data assimilation, improved parameterizations, improvements in numerical techniques and increases in model resolution and computing power

Conclusions, Conjectures and Conclusions, Conjectures and SuggestionsSuggestions

THANK YOU!

ANY QUESTIONS?

Schematic diagram illustrating the error growth in summer (red) and winter (blue). The thick lines in both panels depict the rates at which initially different states reach the boundary-forced state. The thin lines show typical spread of forecasts initialized with slightly perturbed initial conditions on day 0.

Summer

Winter

Schematic Error Growth Schematic Error Growth for the Winter (Red) & Summer (Blue)for the Winter (Red) & Summer (Blue)

predictability of weather and climate (seamless prediction of weather and climate) department of...

Documents