+------------------------------------------------------------------------------+
Dot Earth: A Sharp Ocean Chill and 20th Century Climate
2010-09-22 17:10:55.172 GMT
By ANDREW C. REVKIN
Sept. 22 (New York Times) -- A new paper closely examining
ocean temperatures throws a twist into understanding of the
pattern of global warming seen in the 20th century, but does it
throw established concepts and climate models into question?
A number of scientists focused on the interplay of oceans
and climate say that's highly unlikely, although they acknowledge
that the paper, and other similar work, reveals how the Atlantic
Ocean, particularly, can be a powerful, and unpredictable,
influence on the atmosphere above it. You'll hear from some of
the paper's authors and other researchers below.
The paper, "An abrupt drop in Northern Hemisphere sea
surfacetemperature around 1970," is in the Sept. 23 issue of
Nature. It is based largely on a close-grained analysis of masses
of sea surface and air temperature data collected over the
century. The analysis focuses on a period beginning after World
War 2 when a decades-long climb in global temperatures stopped,
with the northern hemisphere, particularly, cooling somewhat.
Warming resumed and intensified later in the 20th century,
and it's that trend that's been linked primarily to the building
influence of human-generated greenhouse gases.
The mid-century difference in temperature trajectories in
the northern and southern hemispheres has largely been linked,
until now, to a burst of hazy aerosol pollution (which cools the
surface by reflecting sunlight) from the United States and Europe
as the great post-war industrial boom got into high gear.
The paper identifies an abrupt short-term cooling of the
surface of the North Atlantic, largely hidden in the data until
now, and concludes that this was the prime influence shaping that
mid-century climate pattern. This line says it all: Here we show
that the hemispheric differences in temperature trends in the
middle of the twentieth century stem largely from a rapid drop in
Northern Hemisphere sea surface temperatures of about 0.3 6C
between about 1968 and 1972.
Does this conclusion raise questions about the level of
confidence in methods used to determine the mix of ocean
conditions and other influences, like aerosols and greenhouse
gases, that shape climate? Read various climate scientists'
responses to that question below.
In the meantime, some of those eager to cast doubt on the
human element in recent global warming may try to trumpet these
findings. But they might have trouble doing so when they realize
that the work is largely based on temperature data collated by
the same Britain-based team that came under vicious attack last
year in the episode skeptics labeled as Climategate. One of the
authors, in fact, is Phil D. Jones of the University of East
Anglia, a climatologist beset by death threats and attacks after
that episode.
They'll also have trouble because, as is always the case in
science, picking out one study as a keystone development
threatens to cause a kind of intellectual whiplash. A conclusion
is provocative and important. An influential journal focuses on
it. A press release condenses the "front-page thought" and those
whose interests are served by the finding run with it.
Any honest effort to transform energy policies based on such
work has to accommodate the reality that climatology is still a
young, evolving science and could get things wrong -- in either
direction.
It's also important to remember that clarifying the causes
and consequences of the last century's climate changes is just a
small part of the overall picture illuminating rising risks from
the unrelenting buildup of greenhouse gases. This is what leads
me back, always, to tracking the overall trajectory of
understanding of global climate change and its implications.
What you're seeing is the herky-jerky process of knowledge
building, as scientists -- many of them colleagues and co-authors
on previous papers -- challenge each others' methods and
conclusions until a fuzzy image of reality emerges, along with a
path for further inquiry.
Below you can read what that back-and-forthing looks like.
I've collected some thoughts from the paper's authors, and also
from a batch of other researchers separately studying the
interplay of oceans and atmospheres and attempts to sift out
different drivers of climate variability and change.
This all also illustrates why I trust the process of climate
science (as I wrote recently in an otherwise pretty lousy post)
and why I have faith in the established, though still imprecise,
picture of a building human influence on an inherently turbulent
climate.
Below you can read questions I sent to the authors -- David
W. J. Thompson of Colorado State University, John M.Wallace of
the University of Washington, John J. Kennedy of Britain's Met
Office and Phil D. Jones at the University of East Anglia's
Climatic Research Unit -- along with their answers.
Further down you can read some reactions to the paper and
the same questions from others studying oceans and climate.
The questions:The findings on ocean cooling mid 20th century
are fascinating on several levels, perhaps most in relation to
modeling of 20th century climate change.
- I'd love it if you could weigh in on whether the findings
challenge current model designs?
- Or is this simply a parallel set of shifts to what
happened with aerosols through that period, in a way that could
point the finger to one (or both) as driving forces?
- Does this ocean cool spell match known cycles or is this
something novel?
- What are next steps in terms of clarifying what drove mid
20th century climate fluctuations?
The replies from the authors and other climate scientists
are pretty technical, quite long and include some abbreviations
(SH is Southern Hemisphere, for instance; GSA for the Great
Salinity Anomaly) so dive in with that understanding. I think
they help reveal how the process of science works.
David Thompson (author):The key point in our paper is the
suddenness of the drop in Northern Hemisphere sea surface
temperatures around 1970, and thus the suddenness of the
difference between Northern Hemisphere and Southern Hemisphere
sea-surface temperatures around 1970.
The community already knows that there is multidecadal-scale
variability in North Atlantic temperatures. But most of that work
is based on "smoothed" or "filtered" versions of the data (ie,
filters that remove all variability on timescales less than ~10
years). Thus the suddenness of the difference around 1970 was
largely overlooked.
The suddenness has implications for the interpretation of
20th century temperature variability. For example, we already
knew that the Northern Hemisphere did not warm as much as the
Southern Hemisphere during the middle 20th century. One
prevailing theory for this difference is that the NH experienced
the cooling effects of tropospheric pollution more so than the SH
during the middle 20th century. But the current paper shows that
the differences in midcentury warming rates occurred much more
rapidly than previously thought (over a few years, not a few
decades). It is hard to reconcile the sudden difference in NH and
SH sea-surface temperatures with the effects of tropospheric
pollution in the NH, since the pollution presumably ramped up
over a longer timeframe.
Kevin Trenberth, National Center for Atmospheric Research:
I wonder if there is not too much hype here? It seems to me
that what is shown here is mostly related to the Atlantic
Multidecadal Oscillations: e.g. see:
Trenberth, K. E., and D. J. Shea (2006), Atlantic hurricanes
and natural variability in 2005, Geophys. Res. Lttrs., 33,
L12704, doi:10.1029/2006GL026894. [Paper (.pdf)]
The paper brings out the fairly sudden drop in NH vs SH
temperatures around 1970 that is certainly seen in Fig 1 of our
paper, for instance. In that regard it seems to be real, and it
is not due to aerosols or spurious effects, as they state. By
differencing the NH and SH they magnify the effect. Their
mapping of it strongly resembles our Fig 4, although they used an
idealized time series. So this is supposed to relate to ocean
dynamics and the Atlantic Meridional Overturning Circulation. It
is nice to bring this out as natural change of the sort that
occurs and we should be prepared for. But it does seem to me to
be likely due to ocean dynamics.
Michael Wallace (author, reacting to Trenberth):
I agree that we're seeing the same phenomenon through
different lenses. The cooling of the North Atlantic is implicit
in Fig. 1 of Trenberth and Shea, but it doesn't stand out clearly
relative to other features in that particular time series and it
isn't emphasized in the text. The "take home message" in our
paper is that the cooling of the high latitude North Atlantic in
1968-72, in combination with the weaker and more diffuse cooling
of the North Pacific, was strong enough to cause a major shift in
the NH minus SH SST difference.
I find it interesting that this cooling event coincided with
the emergence of the so-called Great Salinity Anomaly. To the
extent that this abrupt cooling event can be identified with
ocean dynamics, regardless of whether it involves the GSA or an
abrupt change in the intensity of the AMOC, it provides a
plausible explanation of why the NH warmed less rapidly from
around the time of the end of WW II to 1980 than the SH. In this
sense, our findings (and I suppose we could say the findings of
Trenberth and Shea as well) call into question the widely held
belief that the different rate of warming was due to the more
rapid buildup of aerosols in the NH during this period. I could
also cite at least two other recent papers that could be viewed
as making the same point, based on other lines of evidence, but
don't have time to do that here.
Had the signal in our Fig. 3a been entirely confined to the
high latitude North Atlantic, I think the attribution of the
interhemispheric SST signal to ocean dynamics would be
compelling. The fact that there are features in the North Pacific
of the same polarity as those in the high latitude North Atlantic
raises the possibility that the abrupt NH cooling from 1968 to
1972 might be merely the result of a fortuitous superposition of
largely independent cooling events in the two oceans. To my eye,
the high latitude North Atlantic stands out quite clearly in Fig.
3a, but as they say, beauty is in the eye of the beholder.
David Thompson:
Just to chime in: Another "take home message" is the very
sudden nature of the major shift in the NH minus SH SST
difference (~ 4 years). That abruptness is masked in previous
discussions of multi-decadal variability in the Atlantic (i.e.,
previous characterizations of the AMO). But the abruptness is
important. It calls into question the importance of aerosols in
the driving the differences in NH and SH temperatures (as Mike
and Kevin noted). And it opens up the possibility of oceanic
processes that are not strictly "oscillatory" ....
At first, I had a hard time believing that the abrupt drop
in the NH minus SH SST time series is not due to data bias (ie,
as is the case for the drop in 1945, discussed in our 2008
paper). But John Kennedy examined the data from every possible
angle (his exhaustive work in this regard is laid out in our
Supplementary Material). The abrupt nature of the drop is
unbreakable.
Gabriele Hegerl, University of Edinburgh:
I like this paper, and I find it great to draw attention to
the evolution of the North Atlantic, which has shown very
interesting variability in the 20th century (early 20th century
warming, the great salinity anomaly with cooling that Dave's
paper highlights, recent relatively strong warming). Dave's paper
raises fascinating science questions, as Kevin's paper does, and
it will be important to check this out in the models. However,
the paper does not, in my mind, question the overall conclusions
of the causes of 20th century climate change.
After all, those conclusions are based on the idea that the
20th century will be a combination of external forcing and
internal variability. As long as the variability patterns are
broadly consistent with the models (and thus it will be really
useful to check out the one shown in the new paper).
In terms of the aerosols: If you want to argue really
simplistic, you could still explain what is seen in Dave's NH-SH
time series: due to the larger thermal inertia of the SH, you
would expect slower warming there with greenhouse gas forcing, so
an increase in NH-SH early on, which would then be reduced as
aerosol forcing becomes stronger in the NH. This simplistic way
of arguing shows you why models are so useful - you can actually
look at more sophisticated climate physics than what I can do
with hand waving (which may or may not be right, Peter would have
more insight on the space time pattern of aerosol forcing that we
detect)!
The early 20th century warming is a good parallel to the
questions asked here: that warming is also seen in Dave's time
series (fig. 3), and its pattern also suggests a role of the
North Atlantic (see Delworth's cited paper). However, this does
not mean that the North Atlantic is the only contributor to the
early 20th century warming, since analyses (for example, 20th
century attribution studies, Stott et al., or my paper
http://www.geos.ed.ac.uk/homes/ghegerl/hegerletal_jclimate.pdf )
show that hiatus in volcanism, an early greenhouse gas signal,
and maybe solar forcing contributed to the early 20th century
warming as well. This is a really good parallel: what we are
looking at in the period in the 60s to 80s may well again be a
complicated interplay of forcing and climate variability. Having
found that climate dynamics contributed doesn't mean that forcing
didn't as well!
Peter Stott, Met Office:
This paper was co-authored by my Met Office Hadley Centre
colleague John Kennedy who has devoted a lot of time and effort
to developing these datasets and taking account of all the
different ways of measuring sea surface temperatures (including
buckets over the sides of ships, engine room intakes, moored and
drifting buoys). Part of the story here is that it is this very
sort of very careful work done by John Kennedy and Phil Jones and
other colleagues working on these datasets that has allowed us to
start challenging the models and our understanding in such a
detailed way - in some ways it is quite remarkable that the
observational data is now good enough to identify this level of
detail in how the climate varies and changes.
So the analysis done in this paper by David Thompson et al
is a valuable piece of additional evidence in enabling us to get
down into the real detail as Gabi has says and pick out the
interplay between the different factors that can affect climate.
As to the models, this interplay between climate change and
climate variability is fascinating and this richness of behaviour
is indeed seen in the climate models. Recently I have been
looking at the climate models collected in the CMIP3 archive
which have been analysed and assessed in IPCC and it is very
interesting to see how the forced changes - i.e. the changes
driven the external factors such as greenhouse gases,
tropospheric aerosols, solar forcing and stratospheric volcanic
aerosols drive the forced response in the models (which you can
see by averaging out several simulations of the same model with
the same forcing) - differ from the internal variability, such as
associated with variations of the North Atlantic and the ENSO
etc, which you can see by looking at individual realisations of a
particular model and how it differs from the ensemble mean. Like
I say, you see a richness of behaviour in the models including in
some occasions behaviour that at first sight looks not dissimilar
to that highlighted in the observations by the Thompson paper and
this on top of the "external control" as we called it in our 2000
paper in Science of the external forcings in a particular model
which drives much of the multi-decadal hemispheric response in
these models and which, in terms of the overall global warming
response, is dominated by greenhouse gases. Like Gabi say this
new study adds to the evidence that we can use to break down in
more detail the interplay of different factors. There remains
important science to do to quantify with much greater accuracy
how these different factors have contributed to past variability.
And given the inherent unpredictability of the internal modes of
climate variability - as distinct from the external control
imposed by the external drivers of climate, which themselves are
also uncertain - such attribution statements will always be
subject to uncertainty and therefore probabilistic. But this
fact, given our understanding of how the large scale climate
system is expected to work and observations from many different
aspects of the climate system showing it responding in a
consistent way, doesn't challenge the large scale view that
warming over the last decades is dominated by greenhouse gases.
In addition there is still clear evidence in my view for aerosols
having played a significant role in holding back that warming,
which acts on top of the effects of internal variability which
play an important role in fluctuations about the forced changes.
Climate models capture much of the complexity and richness
of this behaviour - behaviour that is not programed into the
climate models but which - actually a remarkable success for the
science of climate modelling - but which emerges from these
models when we makesimulations of climate over the last century.
Reference: Stott, P. A., Tett, S. F. B., Jones, G. S.,
Allen, M. R., Mitchell, J. F.B. Jenkins, G. J., 2000, External
control of twentieth centurytemperature by natural and
anthropogenic causes. Science, 290,2133-2137.
Ray Schmitt, Woods Hole Oceanographic Institution:
The association of the abrupt NH cooling in 1970 with the
"Great Salinity Anomaly" (GSA) is further evidence of the
importance of salinity to the ocean circulation and climate. It
directly affects the density, the stratification, the vertical
mixing rates and the dynamics of currents. Freshwater capping
inhibits vertical mixing and diminishes the transfer of heat from
the ocean to the atmosphere. Of course, salinity is also the
variable that tells us how the global water cycle is changing.
The GSA seems to have originated from a large discharge of
ice from the Arctic to the key deep water formation regions of
the North Atlantic. As the Arctic sea ice continues to decay,
such large discharges may become unlikely. However, there is a
large modeling community of "hosers" who like to suggest the
rapid melting of Greenland could do as much. My evaluation of the
quantities involved suggests that it cannot, the accelerated
melting of Greenland has only a very modest impact on North
Atlantic salinities. Salinity is thoroughly dominated by
evaporation and precipitation over the basin, river runoff and
glacial discharges are just too small.
This issue points out the importance of establishing a
salinity monitoring network that would provide insight into the
global water cycle that we do not presently have. Society is much
more sensitive to changes in the water cycle than to temperature
changes alone. Since most of the water cycle is over the oceans,
we cannot begin to understand the water cycle without better
observing capabilities for salinity. Next year's launch of the
AQUARIUS salinity satellite will be a step towards a
salinity-observing system, but we need to strengthen the in situ
observing as well.
I really like this last word from Carl Wunsch, a climate and
oceans scientist at the Massachusetts Institute of Technology:
This seems like a good, mainstream, very technical,
contribution to the difficult problems of (a) determining the
statistical significance of apparent trends in climate-related
measurements and, (b) figuring out what they mean. In a more
conventional field, in which highly technical papers were
published in professional journals rather than Nature or Science,
the paper would be read by the few experts, who over the next few
years would try to understand what it all means, whether it is
really new, what the weaknesses might be, do their own analyses
to see how robust the results are, and ask if there are
conflicting data sets. Historically, some of the most important
advances in science have taken years and even decades to be
understood and appreciated.
In the present situation, we are going to get an
instantaneous, public, response (solicited by you), with no time
for thought, for trying to understand where it fits in the
existing literature, or attempts to try out their analysis
methods oneself, or to ask whether there are other data sets in
conflict or consistency. The Science/Nature publishing methods
relegate crucial information to tiny type footnotes or numerical
references (4, 17, 23-29) which take hours to track down and
figure out what the relevance really is.
Apart from the fact that it is appearing in Nature (and
chosen by often inexperienced editors charged with attracting
public attention), is it obvious that of all of the
climate-related papers appearing in the last couple of months,
that this one warrants having particular attention called to it?
Perhaps it does. But I'd need some time to think about it....
Here's my reply to Wunsch and the group:
Bravo. I couldn't agree more. In fact, I'm soliciting all of
this feedback precisely because the media, more generally, are
likely to give everyone whiplash on this paper in the next 24
hours.
This cycle you describe is an issue I've written about
before. and the journals are complicit.
Whiplash in the GreenhouseClimate Research + Media Focus =
Whiplash
Cimate Experts Tussle Over Details. Public Gets Whiplash
Copyright 2010 The New York Times Company
-0- Sep/22/2010 17:12 GMT
