Thread by @cdkoven

New preprint posted. What happens when we run Earth system models beyond 2100 under both high emissions and net-negative emissions scenarios? Just in time for #Earthday!

Lots of coauthors, some of whom are on twitter: @PatriciaCadule @rosiealice2 @DaveLaw82031625 @ktlindsay88 @SabineMathesius @meinshausen @NichollsZeb @benmsanderson @Neil_C_Swart @KirstenZickfeld

As one might expect, the amount of warming in these scenarios is very very different. When the CO2 concentrations are high, so is the amount of global warming.

For the terrestrial carbon cycle, in both scenarios, the biosphere switches from a sink of CO2 to either a source or neutral conditions, but for very different reasons.

For the ocean carbon cycle, the biosphere remains a sink under high emissions, but also switches to a source under net negative emissions.

Basically, net negative CO2 emissions drives a negative CO2 fertilization effect on land and a reversed air-sea gas flux in the ocean, which act to partially counteract the negative emissions.

For the overshoot to net negative emissions scenario, there is a bit of a lag to the switch, so the reversal of the land and ocean sinks only happens after 2100, even though emissions in that scenario become negative a couple decades before then.

The other big difference between land and ocean is the degree of agreement (or not) between models. For the ocean, while ensemble spread increases after 2100, when we resolve by latitude bands, the models are all telling the same basic story.

On land, the wide spread in model trajectories looks even worse when we focus in by region. The models all do different things in different places, and for different reasons.

In both scenarios, warming remains roughly proportional to cumulative CO2 emissions, in most cases, even if the cumulative emissions start to decrease. But there are some lagged effects that show up on these long timescales.

In some of the models under the high emission scenario, there is a long tail of warming that continues after CO2 stabilize.

On the overshoot scenario, one model (CESM2) has a substantial warming in the 23rd century, even though CO2 concentrations have stabilized by then. The pattern of warming and relative timing point to a recovery of AMOC as the driver of that lagged warming.

Overall, these scenarios help us explore both the long-term consequences of our climate policies, as well as the limits of our models to projecting these dynamics. They are really useful scenarios, and we should be pushing ESMs beyond 2100 more often.