Rapid Change and Long Memory

1
Rapid Change and Long Memory

• Carl Wunsch
• Leverhulme Symposium
• Cambridge UK
• March 2008

2
Abrupt climate change has three major
requirements (1) A component subject to rapid
shifts, and (2) a memory capable of sustaining
those shifts Stiff system. (3) A least large
regional scale, and of magnitude sufficient to
have human consequences (in the present
context). A considerable challenge to a
model---must straddle fast and slow time scales.
3
Dansgaard et al. Nature, 1993
4
Volatile elements of the system include, in rough
order Vector Wind Cloud Cover Tropical
Oceans/atmosphere Sea ice cover Ocean surface
layer . Memory elements Extratropical Ocean
(time scales from decades to 10,000
years) Glacial ice (time scales of years to
10,000 years) Biological albedos (decades to
thousands of years) Physical albedos (decades to
thousands of years) Orography (hundreds of
thousands to 10s of millions years)
5
First singular vector (EOF) of the global
meridional overturning circulation over 15 years.
Tropical! Annual cycle dominated
6

• How to get long memories
• Large heat or general reservoir capacities (e.g.,
  ocean for heat, CO2,)
• Slow propagation times (ocean baroclinic signal
  transmissions,)
• Potential wells with stochastic disturbances

Times between transitions depend only upon the
RMS driving fluctuations and the height of the
well. Shift is very rapid when it occurs.
7
Clearly exists a major element of random behavior
in the climate system
8
Some rapid changes appear reversible (D-O
events). Some appear basically irreversible
(breakup of a large ice sheet). The modeling
requirements are rather different. Is there any
example where the ocean is known, from paleo or
modern data, to plainly have induced by internal
variation, a shift in the wider climate system?
The poster-child abrupt events (D-O, Younger
Dryas, Bolling-Allerod, etc.) all appear to have
occurred in climate states so different from
those of today that their relevance is
obscure. Or, the -8200y event is not clearly
large-scale. The largest, fastest, climate
change is the annual cycle hemispheric, a few
months time scale, very large amplitude, and
serious human adaptation has taken place. Lots of
modern data. Would anyone claim it is fully
understood?
9
Why is there such an obsession with the North
Atlantic Ocean as the controller of abrupt
climate change? It actually seems very
implausible
Not so easy to change the circulation below about
100m. Only about 10 of the oceanic surface area
and so comparatively little contact with the
atmosphere. How does this tail wag such a big
dog? Is it the geography of N. American and
European oceanographic institutions? A high
sedimentation rate? (The drunk looking for his
keys under the lamppost?)
10
(No Transcript)
11
The ocean, taken as a whole, has many of the
characteristics of an immense flywheel. The
stored potential (or even available) energies
are very large compared to any plausible
estimates of inter reservoir transfers.
There are about 1026J of energy in the ocean.
Modern transfer rates between reservoirs are no
larger than 1-2x1012W. An 0.1 shift in energy
levels of the general circulation would take of
order 1023/1012s or about 3,000y unless energy
transfer mechanisms undergo a qualitative change.
How might that work?
12
Inference Understanding of paleoclimate remains
much too primitive to say much more than that
rapid shifts have taken place in the past. But
cause, and effect, remain obscure. The field must
mature greatly before anyone should base public
policy on any of its findings other than that
very different climate states have occurred in
the past, and that sometimes, some elements have
changed comparatively rapidly.
13
Things to care about Sea level (? A curve?)
Drought Ocean acidification
To what extent does past behavior carry
implications for the future? We barely
understand the modern state and that makes it
doubly hard to understand what poorly observed
phenomena in a very different mean state are
telling us, or whether past states apparently
like the present one really are similar.