Royalston, MA
Change is upon us here.
The mist is rising from the trout stream this morning under the red,
orange, and pine-green trees, the swamp maples have dropped their leaves. We have had
our first frost and more are coming. We
will be away to Cerca Trova in two weeks.
But change is also brewing in the outside world. We have personally seen the change, the dying
reefs, the species shifts, the increased storms, the changing rain patterns,
and we are here to tell you it is absolutely happening. But a cousin sent around an article by a
vehement Denier about how, finally, a heroic climate scientist was calling bullshit
on the world-wide clamor to act. So I dug
into that article and the climate scientist in question and here’s what I
found. It runs a bit long because it
takes some work to unpack the truth.
The article is here https://quadrant.org.au/opinion/tony-thomas/2014/03/finally-real-climate-science/.
It definitely challenges the standing belief that climate change is both real
and man-made, so it definitely deserves some real attention. It makes the challenge with name-calling and
shaming, and only one minor scientific argument that is at least worth noting.
The way to credibly assess a scientific position is to
examine the people taking the position, the position itself, and the manner in
which the position is presented. Using this approach, several things
stand out with Tony Thomas’ article on Quadrant.au about Mototaka Nakamura.
The most obvious one is that Quadrant has an obvious
audience and the article sells well with that audience. Other material on
Quadrant is almost entirely name-calling that plays to the people who need to
be outraged. Tony Thomas is himself not a scientist and has never done
any modelling work. He is a “prolific contributor” to Quadrant. He
and Quadrant are clearly making a profit from promoting the controversy. I.e. he clearly has a conflict of interest.
The other key person is Nakamura. See https://www.researchgate.net/scientific-contributions/2013331662_Mototaka_Nakamura
ResearchGate is essentially the IMDB of
peer-reviewed science, ie the only credible science because without peer-review
you can say anything that will get printed.
Nakamura’s body of work stops in 2010. His last topic
of work largely dealt with baroclinic waves (like the meanders in the jet stream) as
factors in various elements of dynamic climate such as the polar vortex
phenomenon. He has published nothing that has a global integrated climate
output. His work has largely dealt with small-scale models of
particularly complex non-linear elements.
He has never published work that shows the global observed changes can
be ascribed to any mechanism other than green house gas emissions. I.e. he is not a published global
climate scientist.
Thomas paraphrases Nakamura, since what Nakamura
has actually written is not readily available. Nakamura’s book does exist
but there is no way to actually read it since it is indeed in Japanese https://www.amazon.com/kikoukagakushanokokuhaku-chikyuuonndannkahamikennshounokasetsu-Japanese-Nakamura-Mototaka-ebook/dp/B07FKHF7T2/ref=sr_1_1?keywords=mototaka+nakamura&qid=1570052979&s=digital-text&sr=1-1
. The excerpts in English that Thomas offers are not translation-cited.
But in those excerpts I see a lot of language of Nakamura “throwing sour
grapes”. Nakamura's specific complaint that Thomas reports is that climate
modelling is so complicated that drawing conclusions is useless. From the get-go this is
a bit strange since Nakamra's entire body of science work is in fact built on
computer modelling.
Just how accurate is Thomas’ interpretation of a book in
Japanese, since I am willing to be that Thomas doesn’t read Japanese? Note that
I am operating only on Thomas’ presentation. And he has already
been shown to have a conflict of interest. But it is all we have to go on for now.
There is a fundamental truth in science, a sine
dicens since it is so utterly obvious, which a Canadian economist George
Box put into eloquent words about two decades ago: “All models are wrong, but
some are useful”. In this specific case, the question is - can these
models be used to predict our near future?
Every single scientist who is an actual scientist knows that
his/her model (a.k.a theory-expressed-as-equations) is incomplete. The way computer models are built is to
assemble the equations which describe the phenomenon, fit the equations’
parameters (the A, B, C etc, typically multiplier factors) to a set of known
data (called the ‘learning set’), and then test that fit with a new set of data
(called ‘the test set’). This is indeed how nearly all climate modelling
has been done.
You might ask how is both a ‘learning set’ and a ‘test set’
obtained. Good question. In the climate situation where the only
climate data are historical, i.e. you can’t just run the experiment again and
collect a new set of data, one way is to draw the learning data from one subset
of the measurements, say every other reading, and the test set is the remaining
data. Another way is to fit the model to one output parameter, say global
temperature, and test against a different parameter, say sea level. This
latter approach is much more powerful and is the basis for most of the
predictions being made.
Predictions are the central point of contention in Thomas’
write-up, since Thomas says Nakamura says predictions can’t be done by
modelling even though he has done so himself. In an absolute sense,
this is correct. Another sine discens is that “extrapolations are
hazardous.” Predictions can be legitimately made out as far as the model
error diverges to where it’s judged the predictions are not confident,
typically the 95% confidence level. However, the way climate predictions are
being done, using about 100,000 years of historical data, is to only look at
model outputs for the next 50-100 years, which is only a few sample points
ahead of Now and are well within the 95% confidence interval of the
models. This is perfectly legitimate and is almost always done with
models. For models being built with recorded data from the
post-industrial age, the predictions are still being only pushed out to where
the model error is excessive.
In fact the echo cancellation technology in your long
distance phone connections does this every single time you make a phone call,
predicting the sounds that will bounce back on the line and subtracting them
before putting the result on to your ears. Do you remember hearing echos
on your long distance calls? Not for a long while now. I.e. the
models are now good enough to predict the signal to several seconds, which is
in fact several thousand sample intervals ahead of Now.
As for the manner in which Thomas presents the results, note
the language and tone of it all as disparaging and only one single scientific
argument. That argument is interesting. Thomas says that Nakamura
says that existing climate models can only be made to fit by adjusting some
model parameters to outrageous values. I have no way to assess the validity of
this statement itself since I am not active in the model-making work and have
not dug in to the models themselves. But if this is the case, then it is
a legitimate complaint.
In my own modelling work for my doctorate, I
arrived at bizarre parameter values, for example negative volumes for the
heart, in order to get the models to fit the data. My doctoral committee
pointed this out and I was able to show many more instances where the model fit
without resorting to weird parameter values. And in the end, my model was
able to successfully predict and remove the signal in question, much like the
echo cancellation example above, so we all said “there must be more to this
beastie” and drove on.
Which is the case with climate science. Everyone
knows there’s more to the beastie. Saying that the beastie is too
complicated so we are just going to ignore it is specious at best and fatal at
worst. The entire September issue of the
Economist is saying that the non-linear tipping points are beyond current
science, for now, but the results are unanimously foreboding. But science
progresses when the next kid up takes what’s been done and adds in another
factor to get to an even better model. Which is what has been going on
with climate science since the first warning report was presented to President
Johnson in 1965 (which he ignored BTW).
Ultimately, there’s my own observations in the real
world. I have traversed the Caribbean for 40 years of which the last 10 years have seen 15,000 miles under
sail. Climate has absolutely changed since 1980’s. No question, I
am here to testify to it. Will it continue to change? All the models say yes, it is changing 100
fold faster than has ever been seen in geobiologic history, and that the
factors which we can measure also can account for that change in ways that
predict no slowing of the change within our kids’ lifetimes (which is a blink
of an eye in geobiologic history)
But bottom line, are the predictions accurately, absolutely correct?
No. All models are wrong but some are useful. The translations of Nakamura reads like a
spoiled perfectionist, not a thoughtful scientist working to address the
situation. There is way too much hysteria and not enough
discussion. Nakamura does not help the situation.
But are the predictions useful? Absolutely. They
are clearly fitting what is being observed well enough that each decade the
situation becomes more clearly problematic. Does it matter if we see
another 1oC increase in 30 years or in 50? Not in the sense that it’s all
bad news and we need to be doing something about it. And are we thinking about the consequences of
that news to us and our cruising?
Absolutely.
The real question is whether we, the world village, are willing
to take the risk that the predictions are fundamentally wrong. The canaries in the coal
mine are screaming, the managers are saying there’s a problem, and we can’t get
out of the coal mine.
