Tipping of the Atlantic Ocean Circulation

Project Information

TAOC

Grant agreement ID: 101055096

DOI

10.3030/101055096

EC signature date 4 May 2022

Start date 1 October 2022

End date 30 September 2027

Funded under

• European Research Council (ERC)

Total cost

€ 2 500 000,00

EU contribution

€ 2 500 000,00

Coordinated by

UNIVERSITEIT UTRECHT
Netherlands

Periodic Reporting for period 1 - TAOC (Tipping of the Atlantic Ocean Circulation)

Reporting period: 2022-10-01 to 2025-03-31

Summary of the context and overall objectives of the project

The Atlantic Ocean Circulation plays an important role in regulating Earth’s climate by redistributing heat through the global ocean. This large-scale ocean circulation consists of a northward transport
of relatively warm surface waters from the tropics to the North Atlantic Ocean. These relatively warm surface waters eventually reach the subpolar regions around Greenland and Iceland where they are
cooled by the atmosphere. As cold water is heavier
than warm water, the surface waters sink to depths of 2000 to 3000 meters before returning southward as a cold deep undercurrent. This Atlantic Meridional Overturning Circulation (AMOC) is crucial for
Europe’s climate, but it also plays an essential role in the ocean’s ability to absorb CO2 and supply oxygen as well as in regulating rainfall patterns in the tropics.
Since 2004, the AMOC strength is closely monitored along the 26N parallel. This
observational record is unfortunately too short to detect any long-term trend. However, climate reconstructions indicate that the circulation’s strength has fallen by 15% since 1950. A related sign of the
circulation weakening is the so-called “cold blob” over the North Atlantic Ocean. It is the sole region on Earth that experienced a cooling trend rather than a warming trend since the start of the last
century. A sign consistent with a declining transport by the AMOC.

The AMOC has been classified as a potential tipping element in the present-day climate. A tipping element is a system that can
(rapidly) shift from one state to another state as the result of a small change in an external forcing. In 1961, it was first realised that the AMOC may have
two stable states and that transitions between these states are possible i.e. the circulation can tip. Using a highly idealised model that represented the Atlantic Ocean circulation,
the feedback loop that causes the circulation to collapse was identified. A freshwater anomaly near the sinking regions will inhibit the sinking as freshwater is lighter than saline
water. This in turn reduces the northward transport of salinity to these regions, which freshens the regions even more. This in turn reduces the circulation strength again and so forth.
When this feedback is strong enough, the circulation will make a transition to a different stable state: the collapsed state.

The overall aim of the TAOC project is to determine reliable estimates of the probability that the Atlantic Meridional Overturning Circulation will undergo a collapse before the year 2100. To reach
this aim, we work along four objectives O1-O4, connected to the Work Packages WP1-WP4 in the project: (O1) to (further) develop novel computational methodology to determine transition
probabilities in high-dimensional multi-stable systems, (O2) to apply this new methodology to a hierarchy of ocean-climate models to determine AMOC transition probabilities and transition
paths, (O3) to simulate an AMOC collapse in one of the state-of-the-art climate models under at least one scenario of climate change, and (O4) to determine precursors of AMOC
transitions and develop a skillful prediction scheme for future AMOC behavior.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

Contributing to O1, we have developed in WP1 a new version of the Transient Adaptive Multilevel Splitting method (TAMS) using machine learning methods to estimate its score function.
We also developed a new method of computing transition probabilities using Dynamical Orthogonal Field methods to reduce the dimension of the problem. Methods to compute optimal transition
paths (instantons) were also further developed by solving the instanton equations numerically.

To realize O2, the methods developed in WP1 have been applied successfully in WP2 to AMOC box models and 2D ocean circulation models. For example, we have studied the optimal transition
paths of the AMOC in these models, leading to insights to which perturbations and noise the AMOC is most sensitive.

We already partly accomplished O3 by successfully simulating an AMOC tipping
event in the pre-industrial version of the a low-resolution version of the Community Earth System Model (LR-CESM). Moreover, we have demonstrated that a previously suggested physics-based
indicator of AMOC stability indeed shows the expected behaviour in the LR-CESM. The onset of the AMOC collapse is close
to a minimum of this indicator. We also performed a full hysteresis simulation with the same LR-CESM version, showing the strong asymmetry of AMOC collapse and recovery. This was a substantial
computation which took us about 6 months to complete on the Snellius supercomputer system in Amsterdam (NL). We were very happy to find such results already so early in the TAOC project
and it has stimulated an enormous amount of further work within WP3. The results so far indicated that Arctic sea-ice plays a significant role in AMOC stability and effectively creates a modified
statistical equilibrium state, which we have called the AMOC weak state. From additional simulations with the LR-CESM under historical and climate change forcing (SSP2-4.5 or SSP5-8.5) we
have provided an estimate of the critical global mean surface temperature for an AMOC collapse to happen. We have analysed major biases in CMIP6 climate models relevant for AMOC stability
and showed that one of the major ones is too much precipitation over the Indian Ocean. First results on the effects of these biases on the multi-stability regime of the AMOC
indicate that the width and position of this regime shift, such that CMIP6 models are likely have a too stable AMOC. Finally, within WP3 we have analysed the different feedbacks
involved in the AMOC collapse and showed that the salt-advection feedback is the dominant one.

Contributing to O4, we build on the LR-CESM results in WP3 to determine optimal observation locations
for an AMOC collapse onset. These turn out to be in the salinity field along the southern boundary of the Atlantic such as observed with the SAMBA array. Using the classical notion of critical
slowdown as an early warning indicator, we have provided an estimate of the onset tipping time of an AMOC collapse.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

So far, the five most significant achievements of the TAOC project are

1. The simulation of the pre-industrial AMOC collapse in the LR-CESM. The importance of this paper is two fold in that (i) it shows that AMOC collapses also can occur in state-of-the-art climate
models and (ii) it shows that a physics-based indicator (the AMOC freshwater transport at 350S in the Atlantic) can be used as an indicator of the onset of the AMOC collapse (paper published). .

2. The development of a TAMS scheme that uses machine learning methods to estimate the score function (papers published).

3. The computation of transition probabilities and transition paths in relatively idealized AMOC models, in particular the computation of the instantons in these models (papers published).

4. The determination of the optimal observation regions for early warning signals of an AMOC collapse and the estimate of a tipping time from reanalysis data (paper under review).

5. The simulation of the collapse of the AMOC in the LR-CESM under the climate change scenario’s SSP2-4.5 and SSP5-8.5. This result indicates that an onset of the AMOC collapse may already occur
under a 2.2C global warming level (paper under review).

amoc-schematic.png