Transcript Hurrell_DecadalForec..

Decadal-to-Centennial Prediction:
Opportunities and Challenges
Jim Hurrell
Climate and Global Dynamics Division, NCAR
[email protected]
OceanObs09: Ocean Information for Society
Day III: Delivering Services to Society
Session B: Forecasting
OceanObs09
21-25 September 2009
Jim Hurrell
Climate Prediction
Seasonal to Interannual
Prediction
Initial value problem
Boundary value
problem
Boundary value
+ Initial value
Decadal
prediction
Centennial
prediction
OceanObs09
21-25 September 2009
Jim Hurrell
CWP on Decadal Climate Variability,
Predictability and Prediction
Contributing Authors
G. Danabasoglu
T. Delworth
D. Dommenget
H. Drange
K. Drinkwater
S. Griffies
W. Hazeleger
N. Holbrook
B. Kirtman
N. Keenlyside
M. Latif
J. Marotzke
J. Murphy
G. Meehl
T. Palmer
H. Pohlman
T. Rosati
R. Seager
D. Smith
R. Sutton
A. Timmerman
K. Trenberth
J. Tribbia
M. Visbeck
OceanObs09: Ocean Information for Society
Day III: Delivering Services to Society
Session B: Forecasting
OceanObs09
21-25 September 2009
Jim Hurrell
CWP on Decadal Climate Variability,
Predictability and Prediction
Contributing Authors
G. Danabasoglu
T. Delworth
D. Dommenget
H. Drange
K. Drinkwater
S. Griffies
W. Hazeleger
N. Holbrook
B. Kirtman
N. Keenlyside
M. Latif
J. Marotzke
J. Murphy
G. Meehl
T. Palmer
H. Pohlman
T. Rosati
R. Seager
D. Smith
R. Sutton
A. Timmerman
K. Trenberth
J. Tribbia
M. Visbeck
OceanObs09: Ocean Information for Society
Day III: Delivering Services to Society
Session B: Forecasting
OceanObs09
21-25 September 2009
Jim Hurrell
Linear trend of surface temperatures
1901 – 2005 ( C Century-1)
°
IPCC (2007)
Mixture of internal variability and forced climate change
OceanObs09
21-25 September 2009
Jim Hurrell
Decadal Climate Variability
Change in Winter Sea Level Pressure (1980-2008)
Dec-Mar
(hPa)
1900-2009
Pressure Falls
1900-2009
Pressure Rises
OceanObs09
21-25 September 2009
Jim Hurrell
Decadal Climate Variability
Sahel Rainfall
Interhemispheric SST Contrast
NH Warm
NH Cold
Strong relationship to SST changes
OceanObs09
21-25 September 2009
Jim Hurrell
North Atlantic SST
Warm North Atlantic
Drought
linked
to
…
Drought
Hurricanes
More rain
Forced component
Less rain
Ting et al. (2008)
Courtesy T. Delworth
OceanObs09
Internal variability
21-25 September 2009
Jim Hurrell
Open Questions and Challenges
• To what extent is decadal variability predictable?
Higher Extratropical SST Predictability
Potential Predictability in Surface Temperature
IPCC AR4 Models
(8900 yrs Control)
10 yr means
potential predictability variance fraction (σv2/σ2)
Boer and Lambert (2008)
OceanObs09
21-25 September 2009
Jim Hurrell
Open Questions and Challenges
• To what extent is decadal variability predictable?
Phenomena: Idealized Predictability Experiments
Perturbed
ensemble
members
evolve
coherently
for two
decades
Courtesy Tom Delworth
in Hurrell et al. (2009)
OceanObs09
21-25 September 2009
Jim Hurrell
Open Questions and Challenges
• To what extent is decadal variability predictable?
• What are the mechanisms for decadal variability?
Latif et al. (2009)
Projected Atlantic SST Change
 Forced by External Processes
o “Committed”
o Future Δradiative forcing
 Internal Processes
Internal Processes
o Low-pass filtering of noise
o Slow internal variations
o Coupled
External Forcing
OceanObs09
21-25 September 2009
 Interactions of Forced and
Natural Variability
Jim Hurrell
Open Questions and Challenges
• Does oceanic variability have atmospheric relevance?
+ AMOC
- AMOC
Pohlmann et al. 2006
OceanObs09
21-25 September 2009
Jim Hurrell
Open Questions and Challenges
• Do we have the proper tools to realize the
predictability?
 Adequate climate observing system?
Global Number of Temperature Observations (1980-2006)
A major challenge for climate analysis and
prediction: uneven observational coverage
in both space and time; deep ocean and
ice covered regions are poorly observed.
Argo
Courtesy Tony Rosati
in Hurrell et al. (2009)
OceanObs09
21-25 September 2009
Jim Hurrell
Open Questions and Challenges
• Do we have the proper tools to realize the
predictability?
 Adequate climate observing system?
et al.
 Reliable assimilation systems to initialize models? Rienecker
Balmaseda et al.
 Are models “good enough” to make skillful predictions?
Global
Decadal Mean Surface T
Regional
Internal Variability
Internal
Scenario
Scenario
Sources of
uncertainty
Model
Model
Hawkins and Sutton 2009
OceanObs09
21-25 September 2009
Jim Hurrell
First Attempts at Decadal Prediction
Correlation skill in predicting 10-yr mean SAT anomalies
(Hashing indicates skill over radiative forcing only)
10-yr, 3-member hindcasts
Every 5 years from 1955-2005
ECHAM5/MPI-OM
OceanObs09
Initial Conditions: coupled model
SSTs restored to observed anomalies
Keenlyside et al. 2008
21-25 September 2009
Jim Hurrell
First Attempts at Decadal Prediction
• Initialization strategies and other aspects differ considerably
• All include estimates of changes in external forcing
• Large differences in future predictions
Global Surface Temperature
Observations
Observations
OceanObs09
21-25 September 2009
Jim Hurrell
Decadal Prediction Protocol
(CMIP5)
• 10 year integrations with initial dates from 1960-2005
• Ensemble predictions (minimum 3 members)
• Ocean initial conditions should be in some way representative of the
observed anomalies or full fields for the start date
• Land, sea-ice and atmosphere initial conditions left to the discretion of
individual modeling groups
• 1960, 1980, and 2005 integrations extended 30 years
WCRP/IGBP Aspen Workshop
(July 2006)
OceanObs09
21-25 September 2009
Jim Hurrell
Concluding Thoughts
• Decadal climate variability:
– Mixture of internal variability and forced climate change
– Significant societal relevance: will be part of IPCC AR5
– Time to explore is now, but:
•
Decadal prediction will require:
– Better characterization and mechanistic understanding
(determines level of predictability)
– Advanced assimilation and initialization systems
– Advanced models (resolution, physics)
– Estimates of future changes in radiative forcing
– Sustained global observations
• Increased dialog with sector communities to
– make best use of predictions, better define requirements
and drive improvement in predictive systems
OceanObs09
21-25 September 2009
Jim Hurrell
Ocean Observations
• Ocean observations are critical for decadal prediction:
• For model improvement and verification
• For initialization and characterization of decadal variability
• Sustain existing observing system components: e.g.,
• Argo, altimetry, SST, sea ice
• Moorings (tropical, boundary current, OceanSITES)
• Address under-sampling
• Full water column observations (initialize MOC)
• Utility of Argo-type measurements below 2000 m
• Need for denser observing system in “critical regions” to
assess observational requirements for decadal information
• Must continually assess utility of existing and planned
ocean observations for decadal prediction systems
OceanObs09
21-25 September 2009
Jim Hurrell
Thank You
Happy Birthday Lisa!
OceanObs09
21-25 September 2009
Jim Hurrell