Geophysical Research Letters
Volume 47, Issue 12
Climate Model Projections
of 21st Century Global Warming
Constrained Using the
Observed Warming Trend
by Yongxiao Liang
E-mail address: yongxiao@uvic.ca
https://orcid.org/0000-0003-3162-1494
School of Earth and Ocean Sciences,
University of Victoria, Victoria,
British Columbia, Canada
SUMMARY:
This study is not
a wild guess about
the future ...
( the usual
nonsense. )
But it states
the obvious.
The obvious
does not require
any study.
Who pays for this ?
The Obvious Conclusion:
If you examine a large
batch of climate models,
and then toss aside
those that had been
making the worst
predictions
(predicting far too
much global warming ),
then the remaining
models will seem
more "accurate".
I didn't need to read
a study to know that.
How about eliminating
every climate model,
except the Russian
model, that predicts
mild global warming
and "seems" to be
quite accurate ?
I call it the Russian
"no thinking model"
-- it seems to assume
past mild global warming
will continue indefinitely.
It's already been happening
for about 325 years, so that
seems like a safe bet !
DETAILS:
Observed (measured)
warming trends
for the post‐1970s period
may be a good metric
to better projected
future warming.
The Coupled Model
Inter-comparison Project
Phase 6 (CMIP6) archive
includes larger ensembles,
longer historical simulations,
and models with a broader
range of climate sensitivity
than CMIP5.
Different climate models
dict different amounts
of future warming
over the 21st century.
Most previous studies
have weighted models
equally to derive
a range of projected
future warming.
That sums
of the "wild guesses"
of climate change
when no one knows
which mode is best.
That makes sense,
because every model
is a different guess
about the effect of
(mainly) carbon dioxide
on the future global
average temperature.
This study gives more weight
to models that were better able
to match the observed
1970–2014 warming trend.
This approach substantially
reduces the upper bound
of projected warming
over the 21st century.
The Fifth Assessment Report
(AR5)
of the Intergovernmental
Panel on Climate Change
(IPCC)
included a range of model
projections of long‐term warming
without any performance‐based
weighting (Collins et al., 2014).
Projections in the IPCC's
Sixth Assessment Report
(AR6)
will be based largely on
CMIP6 (Eyring et al., 2016).
Compared to CMIP5,
the number of different models,
model variants, and ensemble
sizes of individual models
have all increased in CMIP6.
Future scenario simulations
in CMIP6 were coordinated
by the ScenarioMIP project
(O'Neill et al., 2016)
and are driven by a
new set of emissions
and land use scenarios,
known as Shared
Socioeconomic Pathways
(SSPs) (Riahi et al., 2017)
Some new CMIP6 models
show higher transient
climate response
(TCR)
and equilibrium
climate sensitivity
(ECS)
compared with previous
versions of these models
in CMIP5,
(Gettelman et al., 2019;
Sellar et al., 2019; S
wart et al., 2019;
Voldoire et al., 2019;
Zelinka et al., 2020)
An unweighted ensemble
of climate models
to make projections,
as not all models
are equally skillful
in reproducing
observations
(Brunner et al., 2019;
Gillett, 2015;
Knutti et al., 2017;
Lorenz et al., 2018).
Historical model simulations
(1850–2014) and projections
(2015–2100) of climate change
under each of the Tier 1
SSP scenarios are used
in this study
(O'Neill et al., 2016;
Riahi et al., 2017).
Thirty models with up to
50 ensemble members each
are included in the analysis.
The study focuses on
changes in monthly‐mean
global‐mean near‐surface
air temperature (GSAT)
in historical and
future periods.
The HadCRUT4 data set
consists of monthly
historical instrumental
temperature records,
combining sea surface
temperature data
from the UK Met Office
Hadley Centre with
land surface air
temperature records
from the University of
East Anglia Climatic
Research Unit (CRU).