Zhu, J., Poulsen, C.J.
and Otto-Bliesner, B.L.
2020.
High climate
sensitivity
in CMIP6 model
not supported
by paleoclimate.
Nature Climate Change 10: 378-379
Climate modeling:
After billions of dollars spent. and almost 50 years of work, climate modelers (computer gamers) stuck with a HUGE ECS range for a doubling of CO2 (+1.5 to +4.5 degrees C.) that was wild guessed back in the 1970s.
The computer gamers have now expanded the ECS range even more -- to levels unsupported by proxy data of Earth's past climate.
The long-term warming predicted in response to a given rise in atmospheric CO2 defines a climate model's equilibrium climate sensitivity (ECS).
ECS is actually a incorrect theor, since Earth's climate is NEVER in thermodynamic equilibrium.
Over the past five decades, the range of ECS predicted by various versions of climate models has been a doubling of the air's CO2 content above its preindustrial value of 285 ppm is expected to cause a global mean surface air temperature increase between +1.5 and +4.5 °C.
Actual warming since the 1940's is in the +1.0 +1.5 range is you assume only CO2 causes warming -- a worst case estimate.
Recent model revisions have arbitrarily produced even larger ECS values,
as if scaring the general public was much more important than making accurate climate predictions.
Zhu et al. (2020) note that 10 of the 27 climate models participating in the most recent Coupled Model Inter-comparison Project phase 6 (CMIP6) "have an ECS higher than the upper end of the [long-predicted 1.5 to 4.5 °C] range,"
( only 2 out of 28 models were above this upper range
in the previous CMIP5 assessment, CMIP5).
Zhu et al. say their study "illustrates that the development and tuning of models to reproduce the instrumental record does not ensure that they will perform realistically at high CO2."
The three scientists evaluated predictions of global mean surface temperature (GMST), and meridional sea surface temperature gradient (MTG), from a key CMIP6 model against a historical benchmark period -- the early Eocene climatic optimum (~50 million years ago).
That period had atmospheric CO2 concentrations between 3 and 9 times the pre-industrial value (roughly 1750).
The CMIP6 model utilized was the Community Earth System Model version 2 (CESM2).
CESM2, according to Zhu et al.
"is among the best-performing CMIP6 models based on mean pattern correlations of a variety of climate fields."
Zhu et al. report its GMST and MTG projections for the early Eocene climatic optimum (EECO) are "not supported by geological evidence."
In the chart below are EECO simulations at one, two and three times preindustrial CO2 values.
Zhu et al. report that "modeled tropical land temperature exceeds 55 °C, which is much higher than the temperature tolerance of plant photosynthesis and is inconsistent with fossil evidence of an Eocene Neotropical rainforest."
The authors conclude that "the high ECS in CESM2 is incompatible with known Eocene greenhouse climate," adding "we expect that other models with similarly high ECS may also be biased too warm when driven by high levels of atmospheric CO2."
The authors say the hind-casting failure "has been attributed to cloud feedback -- that is, the amplification of surface warming through changes in clouds."
Figure 1.
Model-data comparison of GMST (panel a) and MTG (panel b) during the EECO.
Panel a plots GMST as a function of atmospheric CO2 based on proxy estimates (grey box) and model simulations using CESM2 (red line and dots) and CESM1 (orange line and dots) estimates.
Panel b is the same as Panel a, but for MTG in percent of the corresponding pre-industrial values. For reference, pre-industrial GMST and MTG are marked as open circles in both panels. Note the x-axis is logarithmic in scale and pre-industrial CO2 = 285 ppm.
