Weekly-Seasonal

Decadal Predictability and Predictions Thomas Delworth

GFDL/NOAA

Collaborators: Keith Dixon, Shaoqing Zhang, Tony Rosati, Matt Harrison, Rong Zhang, Fanrong Zeng, Hyun-Chul Lee

Weekly-Seasonal

Decadal Climate VariabilityAnd Change

Multidecadal to CentennialClimate Change

Initial Value Problem

Boundary Value Problem

1. Predictability arising from knowledge of future changes in radiative forcing agents, and climate system response to those changes.

2. Predictability arising from initial state of the system

- “committed warming”

- natural variability of the system

Decadal predictability and predictions

1. On multi-year to decadal scales, is there any predictability associated with the initial state of the system?

- What phenomena might give rise to such predictability?

2. Can we realize that predictability?

- Dependence on models, observations, initialization

3. Are the benefits of realizing that predictability “worth the cost”?

Decadal predictability and predictions

1. Describe one phenomenon that might give rise to decadal scale predictability – observational and modeling results

2. Provide preliminary results from predictability experiments

3. Briefly discuss observational and assimilation system requirements

4. Summary and outlook

Outline

Atlantic Ocean Temperature(70

oW-0

oW,0

oN-60

oN)

Reconstruction of Atlantic Multidecadal Oscillation (AMO)

Gray et al., 2004

Atlantic meridional overturning circulation

SST Change: 1940-1960 minus 1971-1990

• Evidence (instrumental, paleo, modeling) that something like the Atlantic Multidecadal Oscillation exists

• Lack adequate theoretical understanding

• AMO remains a viable hypothesis for some of the observed Atlantic changes over the last century

KEY QUESTIONS:

Does the AMO impact large-scale atmospheric climate?

Can we predict AMO fluctuations?

Hybrid coupled model - based on GFDL CM2.1

AtlanticSlab Ocean

Global Atmosphere/Land System

Pacific Dynamic Ocean

Heat Water Momentum Heat

Indian Dynamic Ocean

Heat Water Mom.

Constant Flux Adjustment

Time varying heating to induce AMO-like SST variations

GFDL CM2.12o atm1/3 to 1o ocnnomads.gfdl.noaa.gov/CM2.X

Regression of modeled LF JJAS Rainfall Anomaly on modeled AMO Index

Modeled AMO Index

Regression of observed LF JJAS Rainfall Anomaly (CRU data) on observed AMO Index Observed AMO Index

ECMWF 40-yr Reanalysis

Regression of LF ASO vertical shear of zonal wind (m/s) on AMO index (1958-2000)

MODEL (10-member ensemble mean)

Simulated multidecadal JJAS surface air temperature difference (K) (1931-1960) –(1961-1990)

Summary so far …

• AMO fluctuations – Generate multidecadal fluctuations in Sahel and

Indian summer rainfall– Modulate the vertical shear of the zonal wind over the

main development region for Atlantic hurricanes– Influence summer temperature over North America

and Europe

• Crucial issues:– How much of AMO-like behavior is internal variability

versus forced climate change?– To what extent are AMO fluctuations generated

internally in the Atlantic versus forced from other parts of the globe?

SST anomalies associated with interdecadal MOC fluctuations

SmallTropicalAmplitude

Anomalous poleward heat transport in Atlantic/Arctic associated with MOC maximum

Atla

ntic

Hea

t T

rans

port

(1

014 W

atts

)

MOC increasing MOC weakeningMOC maximum

JJA Precipitation Anomalies Associated with Maximum MOC

Units: cm/day

JJA: Change in Vertical Shear of zonal wind(850mb-300mb) Associated with Maximum MOC

Key Issues

• What sets the timescale? (spectral peak around 20 yrs)

• How robust are these fluctuations?

• Are these related to the observed AMO?

• “Should” there be a larger tropical signal associated with these?

• NEXT: Predictability of these fluctuations.• LATER: Issues of initialization for prediction.

The

N.

Atl.

MO

C in

the

186

0 C

ontr

ol

Ensemble starting at year 1101

Ensemble starting at year 1201

Air temperature Histogram, 30N-90N

Looking at 21st Century SimulationsProjected Atlantic SST Change (relative to 1991-2004 mean)

Areal average70

oW-0

oW

0oN-60

oN

Results fromGFDL CM2.1Global ClimateModel(SRES A1B)

ObservedTrend from 1950-2004

Key Issues:

(a) How much impact is there for continental climate? Results to date are mixed, even in perfect predictability experiments.

(b) Does this translate into predictability of atmospheric circulation of climatic relevance (ie, tropical conditions relevant to hurricanes; Pacific SST patterns of relevance for North American drought).

(c) Are our current models a fair evaluation of the actual predictability in the system?

- Are our models good enough? - Do model atmospheres interact with the ocean realistically? - Are we missing inherent types of oceanic variability?

(d) Are observing and assimilation systems up to the challenge?

Impact of observational network on “observation” of MOC

CONCLUSION: ARGO network plus atmospheric assimilation allows accurate“observation” of MOC in perfect model context. (S. Zhang, personal communication)

Summary and Discussion

• Decadal prediction is a mixture of boundary forced and initial value problem

• Changing radiative forcing will be a key ingredient, particularly aerosols that can change more rapidly

• On multi-year timescales there is some basis for predictability, probably originating in ocean

• Substantial challenge for models, observations, and assimilation systems

• Unclear what the cost/benefit is – does this add much to the radiatively forced component?

• Some of predictability will arise from unrealized climate change already in the system

Directions and needed activities • AMO is one potentially predictable phenomenon … others?

• Predictability experiments of various sorts to quantify what can be predictable (given current capabilities)

• Improved models

• Sustained observation systems – ARGO looks quite promising

• Theoretical work on dynamical underpinning of phenomena that may give rise to decadal predictability (AMO and others)

• Challenge: If conditions were ripe for another “Dust Bowl” or “mega-drought”, would we know it?