Read on Twitter
New paper out today on how prior assumptions in simple climate models impact Zero Emissions Commitments. Bottom line: a carbon-budget framework to meet mitigation targets only works if we're sure the real world is not high ECS/low TCR. THREAD: https://esd.copernicus.org/articles/11/563/2020/
A two-box model of the Earth (representing shallow/deep ocean), is a reasonable proxy for century scale evolution of global mean temperatures in GCMs (see https://journals.ametsoc.org/jcli/article/26/6/1859/33259 or https://advances.sciencemag.org/content/3/7/e1602821)
Coupling to a simple carbon cycle, we can see how different types of transient observation can constrain the model's parameters. Using historical temperatures and concentrations alone leaves huge uncertainties in (long term) carbon budgets for 1.5 or 2C stabilization
Adding prior paleo knowledge about ECS helps a little, but the only prior information which really constrains the model to linear TCRE-like behavior is prior knowledge of the ratio of TCR and ECS.
Most simple models used in assessment assume either implicitly or explicitly that this TCR/ECS ratio (the Realized Warming Fraction, RWF) is known - justified by TCR/ECS ratios from CMIP models which appear to show a strong relationship (see e.g. https://advances.sciencemag.org/content/6/26/eaba1981)
But this observed CMIP relationship doesn't use equilibrium climate sensitivity (which is the relevant parameter in a simple model) - it uses effective climate sensitivity - and in some models this is a big understimate. https://agupubs.onlinelibrary.wiley.com/doi/epdf/10.1029/2019GL083898
We can estimate possible CMIP ECS by finding simple model parameters which reproduce the first 140 years of the 4xCO2 simulation. For some models, with large slow components of response (e.g. CESM2, UKESM1) - 140 yrs 4xCO2 doesn't constrain the upper bound of ECS.
So - we might be overestimating our confidence in RWF, and therefore we might be too confident that ZEC is small and observed TCRE will hold under net negative emissions. But how does this gel with ZECmip results, which find that ESMs have near-zero ZEC? https://www.biogeosciences.net/17/2987/2020/bg-17-2987-2020.pdf
There are a few possibilities: longer 4xCO2 runs might eliminate the risk of very high ECS, and there may be a relationship between slow-mode ocean carbon uptake and slow radiative feedbacks which constrains ZEC to be small even in low RWF cases.
But the 100 year ZECmip ESM runs aren't long enough to constrain these multi-century carbon climate dynamics - and so our confidence comes primarily from intermediate complexity models. In short - it's difficult to rule out the necessity for multi-century negative emissions.
So - what's the way forward? One option is to frame policy in terms of near-term emissions (rather than indefinite carbon budgets) which allows us to risk-frame uncertainties in long term response as being conditional on near-term mitigation.
Even if we don't know ECS or RWF, there's a good relationship between total emissions pre-2040 and required emissions in 2070-2100. And *by* 2070, under strong mitigation - the observations will tell us more about RWF.
Does any of this change optimal climate policy? A little -in absence of certainty of near-zero ZEC, we need to factor the risk of long term commitments to negative emissions - a risk which may decrease as we learn. As such, near-term mitigation is disproportionately valuable.
