Halstead Harrison : This one’s a bit more specialized than most, and will interest
somewhat fewer.
————
“Uncertainty in Climate Change Projections”
A talk by Dr. Clara Deser, June 3, 2011
review by hh
Dr. Deser was a student of Prof. Mike Wallace, here at UW ‘many’
years ago. I knew her slightly then, but we did not strongly
interact. She is now a Senior Scientist at the National Center
for Climate Research [NCAR], at Boulder Colorado, with a list
of honors longer than mine.
Dr. Deser has conducted computer-modeled tests in an effort
to determine the effects of natural variability in the earth’s
climates system, over periods between days and years, on our our
ability to detect trends in sea-level pressure, temperature, and
precipitation over a projected future, 2000-2065.
With a middling emissions scenario [IPCC AR4 A1B for CO2, SO2,
Black Carbon] she exercised both a single model [NCAR CCSM3 with
~1 degree spatial resolution over land, ~3 degrees over oceans]
with 40 different initializations, to simulate winter weather,
and compared their output with single realizations from a set of
9 [I think?] different climate models chosen from the set of 22
discussed in the last IPCC review.
She found the variability of detected trends in both sets to be
ruffly comparable. That is, the variability in perceived trends
induced by natural weather fluctuations in one model, approxi-
mated the uncertainties introduced by structural differences
between models. She further found that trends in temperatures
were more robust, that is, more easily perceived above the
variability, than those in mean surface pressures and precipi-
tation.
In her talk here she .. unfortunately from my perspective ..
restricted display of her results to the Northern hemisphere
only, between .. ruffly .. eastern Asia and the mid Atlantic.
After her talk I asked of her what seems to me the central
question: “What is the probability that .. assuming her modeled
results well simulated the actual events .. a Climate Skeptic in
2060 might reasonably be able to say ‘Ah Ha! All those pointy-
headed climate guys were wrong!'” I understood her to reply,
“Very small, with respect to temperature trends and their
spatial patterns, and the spatial patterns of changes in mean
sea-level pressures in precipitation, but somewhat larger with
respect to trends in pressures and precip.” I wish she could
have been a bit more precise. Quantitative odds are useful.
Dr. Deser’s central contribution .. as it seems to me .. has
been to emphasize the contribution to model uncertainties from
natural internal variabilities, and the computationally
expensive need for ensembles of model runs to account for them.
The game gets harder.
Cheers,
Halstead