Rapid climate change and climate surprises Richard Wood Hadley Centre for Climate Prediction and Research, Met Office Thanks to Michael Vellinga, Robert Thorpe and Peili Wu

1
Rapid climate change and climate
surprisesRichard WoodHadley Centre for Climate
Prediction and Research,Met OfficeThanks to
Michael Vellinga, Robert Thorpe and Peili Wu
2
Overview
0. Possible causes of rapid climate change 1. Is
the THC important for climate? 2. Different types
of THC change 3. Smooth changes 4.Non-smooth
changes 5.Conclusions and prospects
3
Possible sources of rapid climate change

• Atlantic THC collapse
• see rest of talk
• Respiration vs growth
• land biosphere changes from carbon sink to source
  (Cox et al. 2000)
• Methane clathrate instability
• stable at high pressure, low temperature
• potential destabilisation under global warming
• Ice sheet instability
• unlikely (IPCC) but possible instability due to
  ice streams
• Common features
• Low probability, high impact
• poorly developed understanding

4
Overview
0. Possible causes of rapid climate change 1. Is
the THC important for climate? 2. Different types
of THC change 3. Smooth changes 4.Non-smooth
changes 5.Conclusions and prospects
5
Impact of ocean heat transport on surface
temperature
January mean
(Seager et al. 2002)
6
Surface temperature anomalies due to THC shutdown
Surface Temperature
Cooling over UK 3-5 C (1st decade), 2-3 C
(3rd decade)
7
Central England Temperature Anomalies 1772-2002
Based on Parker et al. 1992
Coldest year since 1659 -2.5?C (1740) Mean
cooling from THC collapse -3?C
8
Change in primary production of carbon due to THC
collapse
Surface Temperature

Assumes current vegetation types
continue Significant regional changes Europe
-16 Asia -10 Indian subcontinent -36 Central
America -109 (current vegetation types
unsustainable)
9
Combined surface temperature effect of THC
collapse and global warming
Surface Temperature
Anomalies (vs. preindustrial climate) in 1st
decade after a hypothetical THC collapse in 2049
(under IS92a greenhouse forcing) Cooling over UK
1-3C N.B. This is a what-if? scenario, not a
prediction
10
Overview
0. Possible causes of rapid climate change 1. Is
the THC important for climate? 2. Different types
of THC change 3. Smooth changes 4.Non-smooth
changes 5.Conclusions and prospects
11
Multiple Equilibria and state transitions of the
THC
(Based on Stommel 1962)
12
Overview
0. Possible causes of rapid climate change 1. Is
the THC important for climate? 2. Different types
of THC change 3. Smooth changes 4.Non-smooth
changes 5.Conclusions and prospects
13
Response of THC to the IS92a greenhouse gas
forcing scenario, in a group of GCMs
Source IPCC (2001)
14
GHG? response in HadCM3 anomalous fresh water
flux (P-ER) into various basins
(Thorpe et al. 2001)
15
Overview
0. Possible causes of rapid climate change 1. Is
the THC important for climate? 2. Different types
of THC change 3. Smooth changes 4.Non-smooth
changes 5.Conclusions and prospects
16
Hysteresis behaviour of THC
(Rahmstorf Ganopolski 1999)
17
Thresholds in GCMs?

• Two GCMs show THC off equilibrium with
  preindustrial greenhouse gases
• (Manabe Stouffer 1999, Rind et al. 2001)
  possibly not very stable?
• THC off state has not been found in other
  models
• (Schiller et al. 1997, Vellinga et al. 2002)
• but it may exist nevertheless!
• No spontaneous state transitions have been seen
• but rapid changes seen when forced by fresh water
  (Manabe Stouffer 1995)
• No threshold effects seen in response to IPCC
  forcing scenarios
• but long-term shutdown seen at 4xCO2 in one model
  (Manabe Stouffer 1999)

18
Rapid THC changes in GFDL GCM
(Manabe Stouffer 1995)
19
Response of MOC to CO2 stabilisation at 4 x
preindustrial in GFDL GCM
(Manabe Stouffer 1999)
20
Overview
0. Possible causes of rapid climate change 1. Is
the THC important for climate? 2. Different types
of THC change 3. Smooth changes 4.Non-smooth
changes 5.Conclusions and prospects
21
Conclusions

• Most likely response over next century is gradual
  weakening of THC
• offsets warming somewhat in NH
• uncertainty over magnitude
• effect is included in current GCM projections
• Threshold events rapid or irreversible THC
  changes
• low probability but possible?
• high impact
• Large modelling uncertainty in THC projections

22
What next?

• Reduce uncertainty
• Live with uncertainty

23
What next?Reducing uncertainty

• Understand key processes in a range of models
  (e.g. CMIP)
• Test processes against reality (difficult!)
• Improve models (difficult!!)
• Some uncertainties will remain...

24
Labrador Sea Freshening and the THC
400
Obs (Dickson et al. 2002)
3500
1950
2000
HadCM3 All Forcings ensemble
25
Is the observed freshening a sign of a weakening
THC?
Deep salinity
1930
2000
2070
THC strength
(Wu et al. 2003)
26
What next?Living with uncertainty

• Model ensembles probabilistic statements
• GCM ensembles for smooth changes (see James
  Murphys talk)
• Need large ensembles for low-probability events
• Use spectrum of models cheap models to explore
  parameter space, GCMs to make link with reality
• What is the question?
• Engage with policymakers
• What observational filters to use?
• What are necessary/sufficient conditions to get a
  good prediction of future THC?

27
The End
28
Millennial Northern Hemisphere proxy temperature
reconstruction (blue) instrumental data (red)
(Mann et al.)
Coolest year in record -0.6?C. Mean cooling
from THC shutdown -1.7?C
29
Fate of THC depends on CO2 pathway as well as
final concentration
Surface Temperature
(Stocker Schmittner 1997)
30
Response of MOC to CO2 stabilisation at 4 x
preindustrial in HadCM3
(R. Thorpe, pers. comm.)
31
A Possible Future Trend in the Labrador Sea

• Under historical external forcings, HadCM3 has
  reproduced an observed freshening trend in the
  late 20th century.
• But this freshening trend may reverse in the 21th
  century under a climate change projection.
• The water mass structure in the Labrador Sea may
  be completely altered.

Model simulated and projected salinity changes in
the Labrador Sea.
32
Simulated MOC variability during the (recent)
Little Ice AgeSimon Tett and many others
Non-temperature influence gt 1960
33
Simulated Conveyor Belt1500 to 1975
conveyer belt is stable in forced simulations.
34
GHG? response in HadCM3 components of anomalous
Atlantic ocean freshwater transport
(Thorpe et al. 2001)
35
GHG? response in HadCM3 relationship of MOC to
steric height gradient in a range of experiments
(Thorpe et al. 2001)
36
No_THC contributions to density change
(Vellinga et al. 2002)
37
No_THC timeseries of anomalous freshwater
transport (25N)
Thick No_THC Thin control G gyre M
overturning
(Vellinga et al. J. Clim. 2002)